They cut people for AI before the AI was ready
On July 2, 2026, at an internal town hall, Mark Zuckerberg told Meta employees something CEOs almost never say out loud about their own strategy. In a recording heard by Reuters, he said the trajectory of AI agent development over the prior four months “hasn't really accelerated in the way that we expected,” and that the company's bets on its new organizational structure “haven't come to fruition yet.” He went further, saying the restructuring itself, which included mass layoffs, was not as “clean” as planned, and that executives had misjudged the timing of the changes.
That admission lands three months after Meta cut 8,000 jobs and rescinded 6,000 already-accepted offers to fund exactly the kind of AI agent development Zuckerberg just said is behind schedule. This article lays out the full 2026 timeline, the real dollar figures behind it, and what workers inside Meta, or watching from anywhere else in tech, should take from a CEO admitting the bet hasn't paid off yet.
The full 2026 layoff timeline, wave by wave
This was never a single event. Meta's 2026 job cuts arrived in a rolling sequence that accelerated sharply as the year went on. In January, the company eliminated more than 1,000 positions in Reality Labs, roughly 10 to 15% of that division, shutting down several VR game studios and cutting the unit's budget by 30%.
In March, Meta cut several hundred more roles, reported at around 700, spanning Facebook, global operations, recruiting, and sales, alongside continued reductions in Reality Labs. The company described these as routine restructuring rather than AI-driven cuts at the time.
Then came May 20: approximately 8,000 employees, about 10% of Meta's global workforce, cut in a single companywide reduction, described internally as structural rather than performance-based. In the same week, Meta rescinded roughly 6,000 already-accepted job offers, meaning thousands of people who believed they had a job at Meta lost it before their first day. Reports at the time indicated additional cuts were already being planned for the second half of 2026, with timing and scope still undetermined.
The number that makes this hard to explain as a cost problem
Meta's own 2025 results make it difficult to frame any of this as financial necessity. Full-year 2025 revenue reached $201 billion, up 22% year over year. Fourth-quarter net income was $22.8 billion, beating analyst expectations, and free cash flow for the year hit $43.6 billion. Meta's stock rose nearly 10% after that earnings report, and rose again when the market learned cuts of up to 20% of the workforce were being discussed, a signal that investors specifically wanted to see spending paired with lower headcount.
This is not a company in trouble reducing costs to survive. It is a highly profitable company redirecting labor spending toward a specific, enormous capital bet. For the broader version of this same pattern across tech and banking in 2026, see our full explainer on why layoffs are happening this year.
What the layoffs are actually funding: up to $145 billion
Meta's 2026 capital expenditure guidance started at $115 to $135 billion when CFO Susan Li disclosed the range on the Q4 2025 earnings call in January, then was raised again by as much as $10 billion, bringing the top end to $145 billion. That is nearly double the $72.2 billion the company spent in 2025, itself a record at the time. Total company expenses for 2026 are projected at $162 to $169 billion, and analysts at Barclays have forecast a near-90% drop in free cash flow as a direct result.
The spending has names attached to it. Prometheus, a one-gigawatt AI supercluster, is coming online in Ohio this year. Hyperion, a 2,250-acre, $10 billion facility in Louisiana, is designed for five gigawatts of capacity. Meta is also reportedly part of a $27 billion joint venture with Nebius for additional gigawatt-scale compute, and spent at least $2 billion acquiring Chinese AI startup Manus.
The part that will genuinely make workers angry: what the AI talent costs
In June 2025, Meta paid $14.3 billion for a stake in Scale AI specifically to bring on its CEO, 28-year-old Alexandr Wang, as Meta's first Chief AI Officer, now running the company's Superintelligence Labs division. Individual pay packages for elite AI researchers have reportedly reached as high as $1.5 billion for a single engineer. Executive stock option grants tied to the restructuring have been reported as high as $921 million each.
Put that next to the 8,000 people let go and the 6,000 rescinded offers in the same restructuring, and the contrast is not subtle: the people being hired at extraordinary cost are not the people being let go. That is, in a very literal sense, the entire point of the reorganization.
Where the cuts landed hardest
Reality Labs absorbed the earliest and among the deepest cuts, losing roughly 1,800 employees, about 12% of the division, on top of the January reductions and VR game studio closures. Recruiting and the broader People organization reportedly absorbed the steepest percentage cut of any function, in the 35 to 40% range, a direct consequence of a company hiring far fewer people overall. Family of Apps growth marketing and middle-management layers across Infrastructure also saw significant reductions. Meta separately cut about 600 researchers from FAIR, its long-running AI research division, as part of the broader restructuring around Wang's Superintelligence Labs.
The surveillance detail that added to the anxiety
Layered on top of the layoffs themselves, Meta faced internal backlash over a paused program that used keystroke and mouse-tracking software on employee devices to collect data described as being used for AI training purposes. At the same July 2 town hall where Zuckerberg discussed the AI agent delays, CTO Andrew Bosworth said a security review found no employee data had actually been used to train AI models under the program, and that if it resumes, it will be opt-in only. For employees already anxious about AI-driven restructuring, a company simultaneously monitoring their keystrokes to train the same AI systems is not a detail that reads as reassuring, regardless of the review's conclusion.
Why this specific admission matters more than the layoffs alone
Layoffs tied to AI investment are now common enough across tech and banking that they rarely surprise anyone by themselves. What makes this moment different is timing and candor. Zuckerberg told employees in May that he did not expect further companywide layoffs this year, though workers were reportedly skeptical even then. Two months later, he is on record saying the restructuring wasn't clean, that leadership misjudged the timing, and that the underlying AI agent capability the whole plan depended on is behind schedule.
That sequence, cut first, then admit the technology isn't ready yet, is precisely the pattern workers across the industry have quietly feared without a company ever confirming it this directly. Meta just confirmed it.
What this means if you work in tech, even outside Meta
This pattern is not unique to Meta, and treating it as a one-company story misses the broader signal. Companies across tech and banking are running the same basic sequence: reorganize around an AI capability that is still maturing, cut headcount to fund it, and only later discover the tools were not ready to absorb the work that left with those employees. For the fuller data on how widespread this specific fear has become among workers, see our breakdown of AI job loss fear statistics. If your own company is showing similar signals, hiring freezes paired with AI announcements, reassignment to AI-focused teams, leadership language about efficiency ahead of results, our guide on spotting layoffs before they're announced covers the internal warning signs worth tracking.
Quiet power moves if you're inside Meta or a company running the same playbook
If you were reassigned into an AI-focused team or pod, document what you are actually being asked to produce and on what timeline, since leadership's own admission that agent development is behind schedule means performance expectations tied to that timeline deserve scrutiny, not quiet acceptance.
If your offer was rescinded or you are watching colleagues affected by these waves, do not treat a single round as the end of the story. Reports already point to further cuts planned for the second half of 2026 with scope not yet finalized. Build your options now rather than waiting for a second wave to confirm what the first one already signaled.
If you feel stuck between an unstable AI-driven restructuring and the fear of leaving a well-paying role, that tension is common enough right now that it has its own name. For a deeper look at why workers are staying in jobs they would otherwise leave, and what that's actually costing them, see our piece on job hugging.
The Grind Hotline read
Strip away the corporate language and the sequence is plain. Meta committed up to $145 billion to an AI infrastructure bet. It funded part of that bet by cutting 25,000 jobs since 2022, including 8,000 in May 2026 alone, and rescinding 6,000 offers to people who had already accepted them. None of this reflects a company in distress; profit and revenue were both climbing sharply the entire time.
Three days ago, the person who made that bet told his own employees the core technology it depended on isn't moving as fast as promised. That is not a leaked memo or an anonymous source. That is the CEO, on a recording heard by Reuters, admitting it directly. The workers who protect themselves best in a moment like this are not the ones waiting for their own company's version of this admission. They are the ones who already understood that a restructuring built around unproven technology can outrun the technology itself, and planned accordingly before the CEO had to say so.
Bottom line
Meta has cut roughly 25,000 jobs since 2022, including a rolling 2026 sequence of about 1,000 to 1,700 in January, 700 in March, and 8,000 in May, alongside 6,000 rescinded job offers the same week. This is funded by a 2026 capital expenditure plan raised to as much as $145 billion, nearly double 2025's $72.2 billion, even as the company reported $201 billion in 2025 revenue and record profitability. On July 2, 2026, CEO Mark Zuckerberg told employees directly that AI agent development has not accelerated as expected and that the restructuring built around it hasn't come to fruition yet.
Further cuts are reportedly already planned for the second half of 2026. If you work inside Meta or any company running a similar AI-first restructuring, the moves that matter are documenting your actual workload against leadership's own stated timelines, building options before a second wave confirms what the first one signaled, and treating this admission as public confirmation of a pattern workers have suspected across the industry for a long time.
About The Grind Hotline
The Grind Hotline is a global media platform and business podcast reaching professionals in more than 150 countries, founded and hosted by an entrepreneur, author, sales coach, and sales trainer. He is a Fortune 100 and Fortune 500 global sales leader who has managed sales teams across dozens of industries and hundreds of companies, the founder of CallTeam, a global outbound B2B lead generation and cold-calling agency, and the creator of the Quiet Power methodology. He works directly with companies through the 90-Day Revenue Engine and the Sales Execution Lab, and runs Layoff Career Counselling for workers navigating job loss, PIPs, and severance.
If Meta's restructuring, or a similar AI-driven reorganization at your own company, has already become personal, Layoff Career Counselling offers confidential, practical support for reading your situation clearly and building your next move.