By Lee Flanagan
Three years ago, Microsoft paid $75bn for Activision Blizzard, betting on gaming as a growth engine. This month it axed 4,800 roles, most of them in the Xbox unit built on that bet, resetting the business it had only just acquired. Executives across the sector will tell you artificial intelligence is finally doing the job of cutting. Nearly 140,000 US tech jobs have disappeared since the start of 2026, according to a Financial Times analysis of company filings and data from executive outplacement firm Challenger, Gray and Christmas, even as the largest hyperscalers pour a combined $725bn into AI infrastructure this year. The convenient explanation is productivity. The evidence points somewhere less flattering: years of overhiring and bets that did not pay off, with AI serving as the easiest cover story for admitting either.
The Bet That Got Reset
Microsoft’s Xbox cuts are not isolated. Amazon, Oracle, Meta and Microsoft together account for almost 50,000 of the roughly 140,000 tech jobs lost since the start of 2026, around 6 per cent of their combined corporate workforce, according to the FT’s analysis. RBC analyst Rishi Jaluria offered a blunter read than most executives are willing to give: “They are moving from one bet to the next.” We think Xbox was the last bet, AI infrastructure is the next one, and the workforce hired to chase the previous bet is paying for the pivot.
The wider US labour market has not followed tech down. Unemployment sits at 4.2 per cent, according to the Bureau of Labor Statistics, and AI-focused start-ups such as Anthropic and OpenAI are hiring fast enough to soften the industry’s overall job losses. UC Berkeley economist Enrico Moretti frames the contrast directly. “Employment in AI is growing at a rapid pace,” he said. “What tech companies are trimming is everything else.” In our view, if AI were quietly automating work across the sector, the cuts would not cluster so heavily in gaming and infrastructure units built for yesterday’s bet.
Oracle’s Balance Sheet Explains the Cuts: Cash for AI, Not Automation
Oracle ended fiscal 2026 with 21,000 fewer employees than a year earlier, following cuts made in March. The company is also committing $70bn to data centre infrastructure to serve clients including OpenAI. This month S&P lowered Oracle’s credit rating to one notch above junk grade, citing weaker cash flow and uncertainty over AI returns. Jaluria put the mechanism plainly: “The money has to come from somewhere.” That is a capital allocation problem: Oracle is freeing up cash for AI investment, not automating the work away. Oracle needed cash for a different bet, and its credit rating already shows the strain of making it.
Overhiring, Rebranded as Automation
Some companies do lean on AI directly. Block chief executive Jack Dorsey laid off nearly half of the company’s 10,000-strong workforce in May and told employees in a memo that AI was shifting headcount requirements. Challenger data cited by the FT links as many as 170,000 corporate job losses to AI since May 2023. Enrico Moretti does not think the attribution holds up under scrutiny. “The typical attitude of tech executives has been to say that AI allows us to gain efficiency rather than admit that they overhired,” he said. “It’s an easy way out.” Pinning the cause on technology lets executives skip a harder conversation about who they hired and why. In our view, the decisions being disowned are often the ones talent acquisition was asked to execute at speed.
Wall Street Is Not Buying It
The FT’s own analysis found that companies citing AI as a factor in layoffs underperformed the Nasdaq by almost 10 per cent in the 30 trading days following the announcement. Companies citing other reasons underperformed by roughly 4 per cent over the same window. Our read is that investors treat an AI-attributed layoff as a warning sign, not a productivity signal. Amazon and Microsoft have noticed: both have explicitly said AI’s rollout did not underpin their decisions to cut staff.
What This Means for the Next Hiring Cycle
For a VP of talent acquisition or a TA Ops leader, the real story here has nothing to do with AI. It has everything to do with what happens when headcount decisions made without rigor during a boom eventually come due. The pandemic hiring spree was not a talent acquisition failure in isolation. It was a company-wide bet, and in our view TA often carried it out under pressure, without the evidence base to justify who got hired and why. Years later, the correction lands as layoffs, and leadership reaches for whatever explanation is most convenient. Right now, that is AI.
In our work with hiring teams, the pattern is consistent: when hiring decisions are not built on evidence at the point of hire, the reckoning does not disappear. It gets deferred, and it comes back attached to a narrative TA did not choose. Ask yourself whether your own hiring data could survive this kind of scrutiny. AI is not coming for your workforce. The unexamined hiring decisions from the last boom are catching up with it instead. Unless TA can show its own hiring was evidence-based from the start, it becomes the easiest target when the bill arrives.
Original reporting: Financial Times.
Frequently asked questions
Does the FT’s data prove AI is not causing any tech layoffs?
No. Some companies, including Block, explicitly link cuts to AI, and Challenger data ties as many as 170,000 job losses to the technology since May 2023. What the analysis shows is that markets are skeptical of the AI explanation, and economists including Enrico Moretti dispute how often AI is the real driver.
Why did companies citing AI in layoff announcements underperform the Nasdaq?
The FT found a gap of almost 10 per cent versus roughly 4 per cent for companies citing other reasons, measured over the 30 trading days after the announcement. The analysis reports the gap without stating the exact mechanism behind the investor reaction.
Why are Amazon and Microsoft distancing themselves from the AI explanation?
Both companies have said explicitly that AI’s rollout did not drive their decision to cut staff, a distinction that lines up with the FT’s finding that AI-attributed layoffs underperform the Nasdaq by a wider margin than layoffs blamed on other factors.
What does Enrico Moretti say is actually growing inside AI?
Moretti points out that employment in AI itself is growing quickly at firms such as Anthropic and OpenAI, even as broader tech companies trim roles outside their core AI bets. That complicates any claim that AI overall is shedding jobs.
Does the pandemic hiring boom explain today’s tech layoffs?
Sweeping tech layoffs became routine after companies hired heavily during the pandemic on bets about sustained demand for digital services. Today’s cuts read as the deferred cost of those bets, not evidence that AI has started replacing the roles being eliminated.