Stankovic.
← Back to News
AIJUL 07, 2026 · 9 MIN READ

The Layoff Everyone Blamed on AI, Where the Company Said AI Wasn't the Reason

LA
Lazar Stankovic

There's a particular tell in corporate communication: when a company volunteers a denial nobody asked for out loud, it's usually answering a question everyone was already thinking. Microsoft cut about 4,800 roles on Monday, and its Chief People Officer used the announcement to state, plainly, that the eliminated roles are not being replaced by AI. Which is a strange thing to lead with, unless you know that "AI did it" is precisely the story the market was going to write for you.

So let's write the more accurate one instead. This week handed us an unusually well-lit version of a debate that has been running all year: how much of the 2026 layoff wave is AI actually causing, versus how much is AI just a flattering label stapled onto ordinary cost-cutting? The honest answer, which almost nobody selling you a headline wants to give, is both, at once, on the same balance sheet.

What actually happened this week

Start with the facts, verified against primaries rather than the aggregator telephone-game.

Microsoft, July 6. Roughly 4,800 roles, about 2.1% of a global workforce that GeekWire pegs at around 220,000 before the cuts. The gaming division takes the deepest wound: Xbox CEO Asha Sharma announced a plan to reduce the team by approximately 3,200 throughout FY27, with roughly 1,600 immediately and four studios leaving Xbox. The rest lands mostly in the Commercial (sales and consulting) organization. Chief People Officer Amy Coleman drew the line explicitly: the roles eliminated are not being replaced by AI, even as, in her framing, AI changes how the work gets done.

Read that formulation twice, because it's doing two jobs. It denies automation-as-cause for these specific roles, and in the same breath keeps AI in the frame as the strategic weather system everyone's operating under. Both halves are true. Neither half is the whole thing.

The financial backdrop is the actual trigger, at least for Xbox. Sharma was blunt in her own memo: excluding Activision Blizzard King, Microsoft has spent over $20 billion over five years on gaming content, platform, and hardware subsidy while annual revenue actually declined by nearly half a billion. That's a division bleeding, not a division being automated. Meanwhile the parent company's shares fell nearly 23% in the first half of 2026, their worst first-half showing since 2022, even as it committed to $190 billion in infrastructure and data-center spending for 2026. A surge in memory prices, driven by data-center demand, is part of what forced Xbox to raise console prices into soft demand. So AI is in this story, just not where the headline wants it: it's inflating the cost base and consuming the capital, not doing the laid-off workers' jobs.

Two smaller in-window cuts round out the picture. British American Tobacco announced roughly 9,000 role reductions around June 29 under its "Fit2Win" program, framed as technology-led restructuring, though the CFO's specific AI framing reached the press secondhand rather than through BAT's own public statement, so treat that attribution as unconfirmed. And Oracle continued its Romania reductions, sitting atop the 21,000-job, roughly 13% workforce reduction it disclosed in its June 22 SEC filing, where the language is refreshingly direct about AI adoption resulting in workforce reductions. That 21,000 figure is SEC-primary; the Romania headcount rests on trackers and employee accounts, so hold it more loosely.

Layoffs announced this week, by company

In-window role reductions, June 25 – July 6, 2026.

British American Tobacco9,000 roles
Microsoft4,800 roles
Oracle (Romania)500 roles
Source: Company disclosures & SEC filings; Challenger, Gray & Christmas

The number that anchors the trend, and its asterisk

Challenger, Gray & Christmas released its June report on July 1, and it's the closest thing the US has to a real-time layoff tracker. The topline: US employers announced 45,849 job cuts in June, down 53% from May, with AI the leading reason for the fourth straight month. AI was tied to 14,029 June cuts, roughly 31%, about one in three, and has been cited in 101,743 job-cut announcements so far in 2026, approximately 23% of all cuts. Tech remains the epicenter: 139,156 cuts through June, up 83% from the same period in 2025, nearly a third of all job cuts this year.

Here's the asterisk you must carry every time you cite that number: Challenger records the reason the company gives, not a reason anyone verified. When Andy Challenger notes that regardless of whether individual jobs are being replaced by AI, the money for those roles is, he's telling you the quiet part. The tracker measures the label, and the label is chosen by the same executives who benefit from it. Which brings us to the actual argument.

The skeptics have names, and a good week

The counter-narrative to "AI is eating jobs" is that AI is mostly cover for jobs firms wanted to cut anyway, a phenomenon that already has a name: AI washing. OpenAI's own Sam Altman conceded earlier this year that some of it is real displacement and some is companies blaming AI for cuts they would make regardless. Apollo's Torsten Slok has been more caustic, arguing there's essentially no macro evidence of AI-caused job losses and that the spending boom is stoking employment, not shrinking it. Nvidia's Jensen Huang, who profits more than anyone from the AI story, called CEOs who blame AI for layoffs lazy. When the person selling the shovels tells you to stop blaming the shovels, listen.

And this week, two pieces of evidence landed squarely on the skeptics' side of the table.

First, Zuckerberg's admission. At a July 2 internal town hall, per a recording Reuters heard, Meta's CEO said the trajectory of agentic development over the prior four months hadn't accelerated the way the company expected, and that its bets on the new structure "haven't come to fruition yet". This matters because of what it followed: Meta cut about 10% of its workforce in May and reassigned roughly 7,000 employees to AI teams, a reorg premised on agents absorbing the work. He now says the agents aren't there yet, but expects benefits within three to six months. He was, by his own account, super optimistic about tools like Claude Code back in January and February when the plan was drawn up. Firing ahead of a capability that then doesn't arrive on schedule is the worst of both worlds: you lose the institutional knowledge and the promised efficiency.

Second, the rehiring reversals. CNBC reported July 1 on a growing set of companies quietly walking back AI-driven cuts. Commonwealth Bank of Australia replaced 40+ customer service staff with an AI voice bot, watched call volumes rise instead of fall, and reversed the layoffs, later admitting it had not adequately considered all the relevant business factors. IBM's AI handled about 94% of routine HR requests but not the 6% involving ethical dilemmas, and then moved to triple US entry-level hiring. Ford rehired hundreds of experienced engineers after automated quality control missed defects that veterans catch. The supporting survey data is damning in the mild way surveys are: Orgvue found 39% of leaders made AI-driven redundancies, a majority of whom now regret it, and Robert Half found nearly a third of US hiring managers eliminated a role for AI and later rehired for the same or a similar one.

So which is it

Neither camp gets to be entirely right, and the discipline here is refusing the clean answer both sides are offering.

The washing thesis is strongly supported this week. Microsoft explicitly denied AI-as-cause. BAT's cuts are cost-and-outsourcing wearing a technology-transformation coat. Oracle's mechanism is capital reallocation, cutting payroll to fund a data-center program, not one-for-one automation. And Zuckerberg just admitted the agents his layoffs were premised on underdelivered. If you were building a case that "AI" is often a communications-friendly wrapper around ordinary margin defense, this was a good week for your brief.

But the structural thesis survives too, because the money is the mechanism. The combined 2026 hyperscaler capex, roughly $650 to $725 billion across Amazon, Alphabet, Meta, and Microsoft, is real, and it creates real pressure to protect margins by trimming headcount. AI doesn't have to be doing the specific job to be the reason the job is gone; it just has to be the thing eating the budget the job was funded from. That's not automation. It's opportunity cost at civilizational scale. And separately, the hiring suppression at the entry level appears genuine and worth watching, independent of the theatrical announcements.

The stoic read: control what you can, which is your own reasoning about the evidence, and refuse the two stories that require you to control nothing. "AI is taking all the jobs" flatters the technology. "AI is taking none of the jobs, it's all a con" flatters your skepticism. The 30,000-foot view is less satisfying and more correct, both things are true simultaneously: AI is real cover and a real cause. It is the cover for cuts firms wanted anyway, and it is the capital sink that makes those cuts financially rational. You don't have to pick. You just have to notice which one a given company is doing, and this week most of them told you, if you read past the denial.

What would change my mind

Two clean signals, so this isn't just vibes.

If Q3 hyperscaler earnings show measurable AI productivity ROI, the washing thesis weakens, because then the cuts start looking like genuine efficiency rather than relabeled retrenchment. Right now the evidence runs the other way, with survey work finding executives reporting workforce reductions that, in their own assessment, aren't translating into return.

And if entry-level hiring keeps falling while capex stays elevated, the structural-displacement thesis strengthens, quietly, in the part of the labor market that doesn't get a press release.

Watch what companies do with compute, not what they say about their people. The announcements are marketing. The capital allocation is the truth.

Sources