Meta is doing both. Cutting 8,000 people on May 20 and writing a $1.5 billion check to one engineer in the same quarter. If you're holding a Meta offer or interviewing there, those two facts tell you everything about what to ask for.
The headline is "Meta layoffs." The reality is a two-track sort: legacy roles out, elite AI talent in, at prices that break the salary band. If you don't know which track your offer is on, you're negotiating blind.
What actually happened in May 2026
On April 23, Meta told its workforce that roughly 8,000 employees — about 10% of headcount — would be cut starting May 20, 2026. The first wave hit on schedule. California WARN Act filings confirmed the headcount, and Meta's chief people officer Janelle Gale called it a restructuring tied to "efficiency."¹
The same announcement froze 6,000 open roles. Positions Meta had been actively recruiting for were cancelled the same week.²
Then, three weeks later, Meta hired five founders out of Thinking Machines Lab — Mira Murati's startup — including AI researcher Andrew Tulloch on a reported package worth $1.5 billion over six years. The hires went into Meta Superintelligence Labs, the org Zuckerberg announced in March 2026 to consolidate the company's frontier AI work.³
Both things happened. They're not contradictions. They're the strategy.
The capex math that explains the cuts
Meta's 2026 capital expenditure guidance is $115–$135 billion. In 2025 it was $72.2 billion. The company is nearly doubling infrastructure spend in one year — almost entirely for AI datacenters and compute.⁴
The cuts aren't because revenue collapsed. Q1 2026 revenue was up. The cuts are because Meta needs to redirect operating cash from headcount into infrastructure to fund the AI buildout. Zuckerberg said it directly at an April 30 company-wide town hall: AI tools are not driving the job cuts. The capex is.⁵
That distinction matters when you're negotiating. If a recruiter tells you "we're tightening across the board," they're describing the symptom. The cause is a capital reallocation decision. The bands on AI-adjacent roles are not tightening. They're widening.
The comp signal nobody is reading
Meta employee median total compensation dropped from $417,400 in 2024 to $388,200 in 2025. In February 2026, the company cut the stock portion of raises by an additional 5%, on top of a prior 10% cut.⁶
For legacy IC and management roles, that's the trajectory: down. The PSC review cycle is harder. Stock refreshers are smaller. Promotions are slower.
For the AI track — Superintelligence Labs, infra, research — the trajectory is the opposite. The Tulloch package is the public number. The unpublished numbers for senior ML engineers and research scientists in the same org are reportedly in the $3–8M annual range, with signing bonuses that match or exceed base salary.
So when someone asks "is Meta paying well in 2026," the answer is: which Meta? The median is falling. The ceiling is breaking records. The middle is being squeezed out.
What this means for your offer
Five questions to answer before you accept, counter, or walk:
1. Is your role on the cut list or the buy list?
Meta has not published org-by-org cut details, but the May 20 wave skewed toward Reality Labs (Quest, AR), middle-management in ads, and recruiting itself. The buy side is Superintelligence Labs, AI infra, and core ranking ML.
If your offer is in the cut zone, you have less leverage than you think — even a strong offer in a shrinking org gets re-leveled within 12 months. If your offer is in the buy zone, you have more leverage than the recruiter is telling you. The band published in your offer letter is not the ceiling. It's the floor for that level.
2. What's the actual hiring temperature on your team?
A recruiter pitching a role at Meta in May 2026 is doing one of two things: backfilling a frozen-then-unfrozen req, or actively building a buy-side team. Ask: "Was this role on the freeze list in April? When was the headcount approved?" A clean answer means active build. A vague answer means you're a backfill, and your offer reflects backfill economics.
Read Meta before the call. Funding stage. Hiring temperature on the org you'd join. Layoff signals in that surface. Recent news. The signals tell you which track you're on before you ever ask the question.
3. What does the open market pay for your exact role?
The AMMO Bench has 1M+ comp data points across 529 role families and 50 metros, refreshed monthly. For a Senior ML Engineer in Menlo Park in May 2026, the P75 total comp is $612K. For a Research Scientist on a frontier AI team, P75 is $1.1M. For a Senior Product Manager on a non-AI surface, P75 is $487K.
Meta will quote you a band. The band is real. The band is also negotiable, especially when the company is publicly paying $1.5B for one hire — that establishes that "the band" is a fiction at the top end. Grade your offer against the live market before you respond.
4. What's the stock vesting structure actually worth?
Meta's stock is up materially YTD. A four-year grant at today's price is not the same instrument as a four-year grant at 2023's price. The dollar value the recruiter quotes assumes a flat stock. If you believe the AI capex bet works, the grant is worth more than quoted. If you think the capex is overcommitted and 2027 earnings get pressured, the grant is worth less.
Either way, ask for the share count, not just the dollar value. And ask whether the offer includes a refresher cycle in year one or whether you wait until your first PSC.
5. What happens if there's a second wave?
NPR reporting confirmed Meta is planning additional layoff waves in the second half of 2026.¹ If you start in June and there's a September cut, your severance clock — typically 16 weeks base plus 2 weeks per year of service — has barely begun. Negotiate a signing bonus that fully vests on day one and a guaranteed minimum severance floor for the first 12 months. Both are gettable in 2026. Recruiters won't offer them. You ask.
The data Meta doesn't want in your hands
Per the Pew Research Center, 66% of workers who negotiate their starting salary succeed in raising it. Only 30% even ask.⁷
At Meta specifically — a company that is publicly demonstrating that comp is a strategic lever, not a fixed band — the gap between asking and not asking is wider than at any other Big Tech employer right now. The legacy median is falling because most employees aren't asking. The AI ceiling is breaking because a small group is asking, hard, with leverage.
You don't need a competing offer from OpenAI to negotiate at Meta in 2026. You need a defensible market read on your role and metro, a clear ask, and a counter to the "we have a band" objection. That objection is publicly false. Bring the receipts.
Compare your offer against the market range for the level and metro. Read the methodology so you can defend the number when the recruiter pushes back.
The honest part
Two analyst views worth taking seriously:
HR Executive's read is that "AI is eliminating jobs" is an oversimplification — many companies quietly rehire offshore or lower-wage talent after headline cuts, and the real driver is efficiency reorg, not automation. Meta fits that pattern in middle layers.
Zuckerberg's own framing — that AI tools are not driving the cuts — is consistent with the capex math. The cuts fund the buildout. They don't reflect AI replacing the cut workers' jobs.
Both views point to the same negotiating reality: don't accept "we're tightening" as a reason to take a low offer. The tightening is selective. If you're in the part of Meta that's growing, the band is wider than disclosed. If you're in the part that's shrinking, you should be asking what your offer looks like after the next wave.
Sector context
103,000+ tech workers have been laid off in 2026 year-to-date, per Layoffs.fyi, on pace to exceed 2025's full-year total of 124,000.⁸ Meta is not an outlier in the sector. It is one of the more visible cases because the cut-and-buy contrast is so stark.
For workers in the buy zone across the industry — frontier AI, infra, ML — 2026 is the strongest comp year on record. For everyone else, the median is flat or down, the bands are real, and the leverage comes from market data, not from your manager's pep talk.
Know which side you're on before you sign anything.
¹ NPR, "Meta will lay off 10% of its staff," April 2026. https://www.npr.org/2026/04/23/nx-s1-5797855/meta-layoffs-10-percent-staff
² NPR, ibid. — 6,000 open roles frozen alongside the cut announcement.
³ The Next Web, "Meta hires five Thinking Machines Lab founders including a reported $1.5 billion engineer," May 2026. https://thenextweb.com/news/meta-thinking-machines-lab-talent-raid
⁴ The Next Web, "Meta to cut 8,000 jobs on 20 May," May 2026. https://thenextweb.com/news/meta-layoffs-may-2026-ai-restructuring-thousands
⁵ TheStreet, "Mark Zuckerberg sends stunning message to Meta employees," May 2026. https://www.thestreet.com/employment/mark-zuckerberg-tells-meta-employees-ai-not-driving-layoffs
⁶ TheStreet, ibid. — median total comp $388,200 (2025) vs. $417,400 (2024); 5% additional cut to stock portion of raises in February 2026.
⁷ Pew Research Center, "How Today's Workers Feel About Their Job Prospects and the State of the U.S. Economy," April 2023, n=5,775. https://www.pewresearch.org/social-trends/2023/04/13/how-todays-workers-feel-about-their-job-prospects-and-the-state-of-the-u-s-economy/
⁸ InformationWeek, "2026 tech company layoffs tracker," May 2026. https://www.informationweek.com/it-staffing-careers/2026-tech-company-layoffs
Stop reading. Grade your offer free. Know which track you're on before the recruiter calls back.
Come to the table loaded.