Meta Layoffs May 2026: 8,000 Jobs Cut Starting May 20

TL;DR

Meta is starting its biggest layoff round since 2023 on May 20. Three Reuters sources put the first wave at roughly 8,000 people, about 10% of the global workforce, with another wave coming in the second half of 2026. Leadership is pitching it as an AI-productivity restructuring rather than a cost cut: AI-assisted engineers plus flatter org charts make a chunk of the headcount redundant. If Reuters’ earlier reporting holds, total cuts could reach 20%+ before year-end.

What Reuters reported

Three sources told Reuters yesterday that Meta has set May 20 as the start date for the first wave. The number quoted is 8,000 employees, give or take, against a global headcount of roughly 79,000 (Meta’s own end-2025 number). That puts the cut at about 10%.

A second wave is planned for the second half of the year. Size and timing of that one are not locked in. Meta declined to comment.

The same Reuters team reported a month ago that Meta was modeling cuts of 20% or more across the year. The May 20 number lines up with the first half of that range. Whether the back half of 2026 brings the rest depends on what executives see from the AI productivity tools they’re betting on.

For context: Meta’s last big restructuring, the 2022-23 “year of efficiency”, eliminated about 21,000 jobs over two rounds. If the 20% target lands, this year’s total ends up roughly the same.

Why now: the AI productivity bet

Meta is not in financial trouble. Q4 2025 revenue grew 24% year-over-year and operating income grew in absolute dollars (margins did compress from 48% to 41% as AI infra spending ramped). That makes the layoff narrative different from 2023, when the cuts were a reaction to overhiring and a market panic.

Leadership’s stated logic runs in three steps:

1. AI agents can now do meaningful chunks of the work an engineer or analyst used to do: writing code, reviewing PRs, running data pulls, writing first drafts of legal and policy memos.
2. With AI doing more, fewer people are needed at every layer.
3. Fewer people means fewer managers per layer too. The “flatter org” pitch is back, and this time it has a productivity story that wasn’t available in 2023.

Whether that story is real or a comfortable narrative for cuts that would have happened anyway is the question every engineer at every big tech company is asking. The honest answer: probably some of both. Coding agents have gotten dramatically better in the last 18 months, and that’s not in dispute. But the leap from “the agent helps a senior engineer ship faster” to “we need 10% fewer engineers” is exactly the kind of leap leadership teams have an incentive to make whether the data supports it or not.

Who gets cut

Reuters didn’t publish a department-by-department breakdown for May 20, but the trail of restructurings over the last four months telegraphs the answer.

Reading the pattern: Reality Labs has eaten the heaviest cuts in every round so far, recruiting and middle management roles in Facebook’s app side have been thinned, and the surviving engineers are being funneled into AI-product orgs. The May 20 wave is likely to follow the same shape, just at much larger scale.

The teams least at risk are the ones building the AI infrastructure itself: the Applied AI org, the inference and training infra teams, the LLM research org Alexandr Wang has been staffing since his $14.3 billion Scale AI hire in June 2025. Wang’s organization has been a hiring magnet inside Meta while the rest of the company contracts.

The “Applied AI” pivot, decoded

The Applied AI org is the unit Meta has built specifically to ship AI agents that write code and execute multi-step tasks autonomously. It’s the corporate analog to what Anthropic, OpenAI, and Google have been doing for the last year, and by external benchmarks Meta trails the leaders on agent quality.

Internally, the move means two things at once:

• Senior engineers from product orgs are being reassigned into Applied AI to accelerate it.
• The product orgs they came from are then declared overstaffed, because the people doing the load-bearing work just left.

That reading follows from how Reuters sequenced the moves. The teams getting drained for Applied AI are the same teams whose headcount math no longer adds up by mid-May.

What this means for engineers, at Meta and elsewhere

A few things stand out, and most of them apply industry-wide, not just to Meta.

flowchart LR
    A[Senior engineer] --> B{AI productivity story}
    B -->|Believed by leadership| C[Headcount reduction]
    B -->|Engineer adopts coding agents| D[Higher per-person output]
    C --> E[Fewer roles, more competition]
    D --> F[Smaller teams ship more]
    E --> G[Hiring bar climbs]
    F --> G

The “AI is doing my job” defense doesn’t work. It used to be reassuring to say the agent can’t actually replace a senior engineer. That’s still true. The thing is, layoffs hinge on team output, not one-to-one replacement. If a team of 10 ships what a team of 14 used to ship, the four roles disappear, and “the agent isn’t really a senior engineer” doesn’t change the math.

Middle management is the most exposed layer. Meta’s pitch (fewer management layers, more individual contributor output per person) is the same pitch Amazon made when it cut 14,000 corporate roles in late 2025, and the same pitch Google has been making since the 2023-24 cuts. There is now broad consensus among big-tech CFOs that the engineering manager / TPM / program management layer was overgrown during the ZIRP era.

Reality Labs is the cautionary tale of the round. Headcount there has shrunk every quarter for two years. Working on a long-horizon bet inside a public company that’s also under pressure to ship AI products this quarter puts the calendar against you.

The AI infra jobs are the safe ones in 2026. Inference, training infrastructure, GPU scheduling, model evaluation, agent harnesses: these are where headcount is growing inside the same companies that are cutting elsewhere. Backend engineers and SREs probably find the path of least pain through one of these teams.

For a wider view of the 2026 layoff picture beyond Meta, see Tech Layoffs Q1 2026: 80,000 Jobs Cut and the AI Restructuring Behind Them.

The macro picture

Meta is far from alone. Q1 2026 already saw roughly 80,000 tech layoffs across the industry. Oracle, Amazon, and now Meta are the headline names, but the same pattern is playing out at every company over 5,000 people: AI as the framing, productivity as the justification, middle management as the primary target.

Two numbers worth holding in mind:

The 80K figure is more impressive than it looks, because the broader US labor market is not in recession. Tech is contracting against a backdrop of an otherwise-healthy job market. That’s the part that should worry engineers more than the headline numbers. When the rest of the economy is fine and your industry is shrinking, the cuts are structural rather than cyclical.

What’s not in the Reuters reporting

A few things to flag honestly, since the breaking-news story is still moving:

• Severance terms remain undisclosed. Meta’s 2023 packages were generous (16 weeks base + 2 weeks per year). Whether the 2026 round matches is unknown.
• Geographic breakdown. Reuters’ sources didn’t split the 8,000 by region. Past Meta rounds have hit US headcount disproportionately, but 2026 may differ.
• H-1B and other visa-dependent staff at Meta have a 60-day grace period to find new sponsorship after termination. Past rounds have included specific provisions; this one’s unclear.
• WARN Act filings. California, Washington, and New York require advance notice for cuts of this size. Filings in those states over the next two weeks will give a much harder geographic and divisional breakdown than Reuters’ anonymous-source reporting.

FAQ

When are Meta’s 2026 layoffs?

The first wave is set to begin May 20, 2026, according to three Reuters sources. A second wave is planned for the second half of 2026, with size and date not yet finalized.

How many people are being laid off at Meta in 2026?

Around 8,000 in the May 20 wave, or roughly 10% of Meta’s global workforce. Reuters reported earlier that the full-year total could reach 20% or more, which would put the year’s cuts on par with the 21,000 jobs eliminated in the 2022-23 “year of efficiency.”

Why is Meta laying off employees if they’re making record profits?

The official framing is AI-driven productivity gains: fewer engineers and managers needed because AI agents are doing more of the work. Meta is not cutting due to financial pressure; it grew revenue and margins last quarter. Whether the AI productivity story is the real reason or the convenient narrative is debated.

Which Meta divisions will be affected?

Based on the pattern of January, March, and April 2026 restructurings, Reality Labs, recruiting, sales, global operations, and parts of Facebook’s app team have been hit hardest. The Applied AI organization and inference/training infrastructure teams have been growing rather than shrinking.

Is this related to Alexandr Wang and Scale AI?

Indirectly. Meta acquired Wang’s expertise (and a chunk of Scale AI’s team) for $14.3 billion in mid-2025 and put him in charge of AI strategy. The Applied AI org he’s been staffing is the destination for engineers being reassigned out of older product teams, which then makes those teams’ headcount math no longer work.

Bottom line

If you’re at Meta and not in Applied AI, AI infra, or a team Wang’s org pulls from, May 20 will be tense. Everywhere else in tech, the Q1 industry numbers (80K and climbing) suggest the same productivity-driven restructuring is happening at every company over 5,000 people, just on a slower clock.

The honest read: AI agents aren’t yet good enough to replace senior engineers, but they’re good enough to give CFOs cover for cuts that the financial story alone wouldn’t justify. Some companies will pull it off and look smart in 18 months. The ones that misjudged the productivity gains will be hiring back through 2027 and 2028, quietly, at higher salaries, into a much tighter market.

For everyone else: “AI-proofing” your role stopped being a cliche about a quarter ago. It’s now closer to a job description.