Meta Begins 8,000 Layoffs While Routing 7,000 Into AI Pods

Meta began emailing layoff notices to roughly 8,000 workers this morning, the first 10% workforce cut since 2023, even as chief executive Mark Zuckerberg directs as much as $145 billion into artificial intelligence (AI, the umbrella term for machine learning and large language models) infrastructure this year and reroutes another 7,000 staff into newly minted AI teams. Notifications hit inboxes in three waves at 4 a.m. local time, beginning in Singapore and rolling west across the United Kingdom and the United States, according to an internal memo from chief people officer Janelle Gale.

That swap is the structural story. Meta is rewriting which kind of worker the company wants on payroll, and the bill for that rewrite shows up in the capex line, not the severance line.

The Number That Frames This Morning’s Email

The arithmetic is plain. 8,000 cuts out of 78,000 employees lands at about 10.3%, the largest single round since the November 2022 reorganization that removed more than 11,000 roles. Singapore staffers were first to learn their status, followed by colleagues in Europe and finally the Americas. United States employees were told to stay home on Wednesday.

The shape of the exit package matters because Meta is using it as the model for further reductions later this year. Severance for United States workers covers 16 weeks of base pay plus two weeks for every year of continuous service, with health coverage extended to 18 months. That healthcare provision is triple what laid-off Meta employees received in earlier rounds.

• 8,000 roles eliminated in the first wave, with another 6,000 open requisitions also canceled
• 16 weeks of base salary as the floor of the United States severance offer
• 18 months of COBRA (Consolidated Omnibus Budget Reconciliation Act, the federal law that lets former employees pay for continued group health coverage) included in the package
• 4 a.m. local-time delivery in each region, so the morning ended before commuting began

Gale told employees at an internal meeting last month that morale had taken a hit during the four-week limbo since Meta first flagged the round on April 23. She conceded the run-up was a “shitty” situation the company was trying to manage with as much dignity as it could.

Why a Record Revenue Quarter Triggered the Cut

Meta did not cut because the business is weak. The opposite is true. The company posted $56.3 billion in first-quarter revenue, up 33% year over year, with operating income of $22.9 billion and a 41% operating margin. By any standard read of a quarterly print, that is a company hiring, not firing.

The pressure sits in the spending line below the revenue beat. Meta raised its 2026 capital expenditure range to $125 billion to $145 billion at the April 29 earnings call, lifting the midpoint by $10 billion from prior guidance. First-quarter capex alone came in at $19.8 billion. To fund part of the year, the company priced a six-tranche $25 billion investment-grade bond sale, one of the largest corporate offerings of the year.

How the Capex Curve Looks

Set against the prior two years, the trajectory reads less like a step up and more like a regime change in how Meta deploys cash.

Why Headcount Becomes the Adjustment Variable

When a company doubles its data-center bill two years running, the operating-expense line has to give. People are the only large lever left that Meta controls on a quarterly cycle. The headcount cut returns roughly $1.8 billion to $2.4 billion of annualized salary cost depending on average compensation, a rounding error against $145 billion but a number Wall Street can score immediately.

The April 29 reaction told that story crisply. Shares fell about 7% in after-hours trading, even as Alphabet and Amazon, which also raised capex, climbed on the back of cloud revenue. Meta has no cloud business to monetize its compute build. The investor question, said Melissa Otto, head of Visible Alpha research at S&P Global, is whether the spend will translate into top-line growth or efficiency gains anyone can see in a model.

Where the Cuts Landed First

Public posts and two sources cited by Business Insider point to specific functions hit on day one. Three patterns are visible.

• Integrity, the team responsible for policy enforcement, civic and election work, and platform abuse, took notifications in the early wave.
• Cybersecurity roles inside the broader infrastructure and security organization received notices.
• Content design, the discipline that writes the interface copy, error states, and product flows that govern how billions of users interact with Facebook, Instagram, and WhatsApp, was hit broadly.
• Managerial layers across the company are being trimmed, with Gale’s memo arguing that “many orgs can operate with a flatter structure with smaller teams of pods/cohorts that can move faster and with more ownership.”
• 6,000 open requisitions were also canceled, bringing the effective headcount reduction to about 14,000 positions Meta will not now fill.

The fingerprint is familiar from prior rounds. Trust, safety, and integrity functions, along with the writers and designers who build user-facing experience layers, absorb a disproportionate share of cuts when an engineering-heavy company refocuses on infrastructure. In the 2023 round, more than 100 trust, integrity, and responsibility roles were eliminated based on Department of Labor filings reviewed by reporters at the time. Today’s round appears to follow the same pattern at smaller scale.

Gale told staff that further job cuts beyond Wednesday’s round are not ruled out, with additional waves possible in the second half of the year.

Severance Math Compared to Block and Amazon

Whether 16 weeks is generous depends on the benchmark. Compared to recent technology-sector packages from companies of similar scale, Meta sits at the upper end on healthcare and middle on cash, though half of the 16-week base is already required under the federal Worker Adjustment and Retraining Notification (WARN) Act for layoffs of this size.

The doubled-and-then-some healthcare line is the most consequential change. Meta previously offered six months of COBRA support in 2023 rounds. Tripling that figure addresses a specific complaint from earlier waves, where former staff with chronic conditions or dependents faced a hard six-month deadline to find new employer coverage. Eighteen months stretches that runway across most full benefit-enrollment cycles.

Severance is also negotiable, particularly for senior staff. Separation dates, restricted stock unit (RSU) vesting acceleration, and outplacement-service duration are the three terms most often adjusted, according to specialist firms that advise departing technology workers. Meta is also offering immigration support and career-services placement to affected employees.

The 7,000-Person Reassignment Into AI Pods

The other half of Gale’s memo is where the second-order story sits. Roughly 7,000 employees will be moved into new organizations built around artificial-intelligence work. Three named teams have surfaced: Applied AI Engineering, Agent Transformation Accelerator XFN, and Central Analytics.

These pods report into the broader umbrella of Meta Superintelligence Labs, the consolidated AI research and product unit set up in mid-2025 after Zuckerberg paid roughly $14 billion to bring in Alexandr Wang, the founder of data-labeling firm Scale AI, as chief AI officer. The Wang appointment also triggered the November 2025 departure of Yann LeCun, Meta’s former chief AI scientist and a Turing Award laureate.

We’re now at the stage where many orgs can operate with a flatter structure with smaller teams of pods/cohorts that can move faster and with more ownership.

That line, from Gale’s Monday memo to staff, captures the organizational logic. Pods, originally piloted inside parts of Reality Labs, are small autonomous teams owning a single product or capability end to end, with thinner management layers above them. Multiplied across the company, the structure produces a flatter org chart and removes intermediate director and senior-manager positions, the layer where today’s cuts have been heaviest among managers.

What the AI Pods Actually Build

Applied AI Engineering takes research outputs and ships them inside Facebook, Instagram, WhatsApp, and Threads. Agent Transformation Accelerator XFN is positioned as a cross-functional unit deploying agentic AI inside existing workflows. Central Analytics consolidates measurement work that was previously spread across product groups.

The Workforce Composition Change

The arithmetic of the two moves together is the part most outlets are missing. 8,000 out plus 7,000 reassigned equals 15,000 staff whose role on payroll has changed in a single week. Layer in the 6,000 canceled requisitions and the move affects close to 27% of what Meta’s headcount would otherwise have looked like by year end. Even if the company’s total head count barely shifts in a quarter, the function mix underneath has been rewritten.

What Wall Street Is Pricing In

The April-29 selloff after the capex raise was not a one-day swing. It established a thesis that has trailed the stock through May: Meta is the only Magnificent Seven name spending at hyperscaler levels without a hyperscaler customer base to pay for the buildout. Alphabet’s cloud line grew 63.4% year over year in the same quarter, and that line is the easiest investor narrative for why a $90 billion-plus capex budget is justified. Meta does not have an equivalent.

During the earnings call, Zuckerberg told analysts the company does not yet have a “very precise plan” for how each AI product will scale to revenue. That admission, fair as a long-term framing, is the sort of comment that costs a stock several percent on the day.

The May 20 cuts are partially a response to that pressure. Reducing operating expense lets Meta tell the market it can sustain a doubled infrastructure bill without compressing margins, while preserving its ability to spend on the model-side wages, including the multi-year contracts inside Superintelligence Labs that are reportedly in the eight-figure range for senior researchers.

Leaders have already told staff that further reductions are possible later in 2026. If those waves match the scale of today’s round, the year ends with a Meta workforce roughly the size of its 2022 peak but composed of materially different people doing materially different work, paid out of a budget that has effectively traded a tier of human capital for a fleet of Meta’s most recent quarterly filings on EDGAR show as a $145 billion AI infrastructure commitment. Whether that trade clears the return-on-investment bar Wall Street keeps asking about is the question the next four quarters answer, not this one.