Microsoft AI Chief Walks Back White-Collar Jobs Warning

Microsoft AI chief Mustafa Suleyman has walked back his warning that artificial intelligence would automate most white-collar work within 18 months, telling the Decoder podcast this week that he meant AI would speed up sub-tasks such as email and slide decks, not replace lawyers, accountants, and project managers. The clarification puts the head of Microsoft AI alongside OpenAI’s Sam Altman, who said in May that he was “delighted to be wrong” about AI’s near-term impact on entry-level jobs.

Both clarifications land months before the original 18-month window expires, and they arrive as enterprise AI costs climb and adoption rates inside white-collar firms trail the predictions.

What Suleyman Said on Decoder

Suleyman, who runs the Microsoft AI unit that builds the company’s frontier models, appeared on Decoder on Monday. Host Nilay Patel pressed him on the Financial Times interview he had given in February and the headlines it produced, including phrases about lawyers and accountants losing their jobs within a year to 18 months. This time the language was different. AI, Suleyman said, will help with sub-tasks that can be digitised, including drafting emails, talking through ideas with colleagues, and building PowerPoints. Faster and more efficient, he said, does not mean the role disappears. Microsoft frames the bet as building AI agents that handle the rote, labour-intensive pieces of white-collar work, leaving the role itself intact.

I said ‘tasks’ in the quote that you’ve just said. So that does not mean jobs… Jobs and roles are the broader category, and tasks are the components of that.

That was Mustafa Suleyman, chief executive of Microsoft AI, on Decoder on June 8, 2026, walking back the comment that set off the original round of coverage.

The February Warning in His Own Words

The original statement came in a Financial Times interview published in February 2026. Suleyman told the FT that white-collar work, including roles for lawyers, accountants, project managers, and marketers, would be fully automated by AI within a year to 18 months, framing it as a question of when, not if.

He also predicted “human-level performance on most, if not all professional tasks” from AI in the same window, and pointed to the exponential growth in computing power as the reason. Suleyman described compute advances as a “flashing red signal” that AI would replace large groups of professionals, the same kind of phrasing that put his name on a string of “AI is coming for your job” headlines. Anthropic chief executive Dario Amodei had said a year earlier that AI could wipe out half of all entry-level white-collar jobs, and Ford chief executive Jim Farley had warned that AI would halve the number of white-collar jobs in the United States. Suleyman’s framing was the new volume on a record that had been playing for a year.

His broader pitch was about Microsoft building its own frontier models toward “superintelligence,” a goal he tied to the company’s push to reduce its dependence on OpenAI. The same Suleyman-run unit that made the warning unveiled seven in-house models at Build 2026 in early June, including its first reasoning system, MAI-Thinking-1, the opening move in the same self-sufficiency story.

A Wider Retreat Among AI Chiefs

Suleyman is not the only one softening the line. Sam Altman, the chief executive of OpenAI, told a Commonwealth Bank of Australia conference in Sydney on May 26 that his earlier warnings had been overblown. He said he had expected to see more entry-level white-collar jobs gone by now, and that the “human part” of work had proved harder to replace than the model predicted.

I don’t think we’re going to have the kind of jobs apocalypse that some of the companies in our space advocate or talk about.

That was Sam Altman, chief executive of OpenAI, at a Commonwealth Bank of Australia conference in Sydney on May 26, 2026. Altman had said in a 2023 interview with The Atlantic that “jobs are definitely going to go away” and told a Federal Reserve conference in 2025 that “entire classes” of jobs would vanish. By May 2026 his position was that “we really do care about our interactions with people,” a phrase that pointed the other way. Anthropic’s Amodei, who told Axios in 2025 that up to half of all entry-level white-collar jobs could dissolve within five years, has also reportedly revised his tone, according to the February coverage of the original prediction.

What the Layoff Data Actually Shows

The retreat is not happening in a vacuum. AI is cutting jobs; it is just not cutting them on the timeline the chiefs named.

Research from outplacement firm Challenger, Gray & Christmas, reported in May, found that AI had been cited as the justification for nearly 50,000 job cuts through April 2026. That is a real number, and a real human cost. It is also far below the pace implied by Suleyman’s February framing, which had most white-collar tasks gone by mid-2027.

• Meta said in May it was eliminating roughly 8,000 roles, about 10% of its workforce, citing a shift to AI
• Intuit cut 17% of its staff, or 3,000 people, in May
• Amazon and Alphabet have run layoffs while pushing AI adoption and efficiency
• Block and Snap have also pointed to the technology in layoff announcements
• Microsoft let go of 15,000 workers in 2025, though it did not cite AI as the reason at the time

The wider tech layoff count is bigger but does not separate cleanly by cause. The live tech-layoffs tracker, captured on June 9, 2026, lists 150,612 people laid off across 377 tech events in 2026, an average of 941 per day, against 245,953 across 783 events in 2025, an average of 674 per day. The Yale Budget Lab, in a May study, found no meaningful change in unemployment through March 2026 for workers in jobs with high AI exposure, and judged AI unlikely to be the cause of recent labour-market weakness. Brookings reached a similar conclusion: “rapid advances in AI capability are not translating automatically into broad economic gains or meaningful adoption.”

Andy Challenger, the firm’s chief revenue officer, said tech companies are leading all industries in layoff announcements and “often citing AI spend and innovation,” but added that “regardless of whether individual jobs are being replaced by AI, the money for those roles is.” Jerome Powell, the then-Federal Reserve chair, said in October 2025 that a significant number of companies were freezing hiring or doing layoffs, and “much of the time, they’re talking about AI.” Both statements sit on the same evidence, and both are six months or more older than the AI chiefs’ recent walk-backs.

AI Costs Are Pushing Back Too

Even when AI can do the work, paying for it is another matter. Praveen Neppalli Naga, Uber’s chief technology officer, admitted in April that the company had burned through its 2026 Claude Code budget in four months. Uber’s president and chief operating officer, Andrew Macdonald, said on a podcast on May 22 that it is becoming harder and harder to justify AI costs inside the company. Nvidia’s Bryan Catanzaro, vice president of applied deep learning, told Axios that “for my team, the cost of compute is far beyond the costs of the employees,” a reversal of the usual pitch that AI replaces labour expense. Microsoft has reportedly begun cancelling engineer licences to Anthropic’s Claude because of the cost.

The same economics, in the other direction, showed up in a study from nonprofit Model Evaluation and Threat Research (METR) cited by Fortune, which found that AI tools made software developers’ tasks take 20% longer rather than shorter. The upshot for the original 18-month prediction is that adoption is gated as much by the bill as by the model’s capability, and the bill is the part the chiefs did not size in February.

Where the Two Stories Meet

The two halves of the story sit next to each other without cancelling out. AI is changing white-collar work, and tens of thousands of cuts are being attributed to it. It is also not the broad wipeout the February forecasts described, and the executives who made the forecasts are now publicly walking them back, sometimes within the same year they gave them.

Booking Holdings chief executive Glenn Fogel said at the TIME100 Summit that “the lowest rung on the ladder has been pulled away by AI,” a description of customer service roles now done by AI inside his own company. The doomsday version and the rung-by-rung version are both real, and the AI chiefs are choosing to talk about the second one.

Altman, asked for his scorecard at the CBA conference, said he and his peers had been “roughly right on technological predictions and pretty wrong on the social and economic implications.” He had said in February that some companies were blaming AI for layoffs that would have happened anyway. The 18-month clock Suleyman set in February runs to mid-2027, and the layoff data, the cost reports, and his own follow-up interview are the first three data points the public has on whether the original prediction will land, in the same way Altman’s CBA remarks reset OpenAI’s posture.