Meta spends $7.8M in AI infrastructure per job cut. Amazon spends $12.5M. The Big 4 combined spend $12M per eliminated position. An NBER study of 6,000 executives found 90% of firms report zero AI impact on jobs.
Photo by Vitaly Gariev on Unsplash
Meta AI Spend Per Job Cut
Firms Report Zero AI Impact on Jobs (NBER)
Big 4 AI Spend in 2026
Meta is reportedly preparing to cut up to 20% of its workforce — roughly 16,000 positions — while simultaneously committing $115-135 billion to AI infrastructure in 2026 alone. Amazon already cut 16,000 in January and may cut another 14,000 by May. Block halved its staff in February. In each case, AI was cited as the justification.
We ran the numbers on what these companies are actually spending per eliminated position. The results don’t support the narrative that AI is replacing workers. They suggest something more uncomfortable: companies are cutting costs to fund a bet that hasn’t paid off yet, and calling it transformation.
The DropThe AI Replacement Cost Index 2026
A simple calculation: divide each company’s annual AI infrastructure budget by the number of jobs cut. The result is the AI cost per job eliminated — how much the company is spending on AI for every position it removes.
This is not what AI literally costs to replace one worker. It is the ratio that reveals whether the spending and the cuts are proportional, or whether something else is going on.
| Company | 2026 AI Capex | Jobs Cut (2026) | AI Cost Per Job | Years of Salary Equivalent |
|---|---|---|---|---|
| Amazon | $200B | 16,000 (phase 1) | $12.5M | 56 years |
| Meta | $125B (midpoint) | ~16,000 (planned) | $7.8M | 31 years |
| Alphabet | $180B (midpoint) | TBD | — | — |
| Microsoft | $120B+ | TBD | — | — |
| Big 4 Combined | $660B | ~55,000 (industry Q1) | $12M | 48 years |
Amazon is spending $12.5 million in AI infrastructure for every job it eliminates. At an average corporate compensation of $224,000, that is 56 years of salary. Meta‘s ratio is $7.8 million per position — 31 years at the company’s median total comp of $280,000.
No company recovers 31 years of salary by firing one person. The math only works if the AI spending and the layoffs serve entirely different purposes — which, according to the evidence, they do.
What the Layoffs Actually Look Like
Meta‘s cuts reportedly span product development, content moderation, and corporate functions. They are not touching AI and machine learning engineering — those teams are expanding. The median total compensation at Meta is $280,000, but the positions being eliminated skew toward mid-level corporate roles, not the $430,000+ engineering positions the company is actively hiring for.
Amazon‘s January cuts hit software development, engineering management, program management, and business intelligence. CEO Andy Jassy framed them as “anti-bureaucracy.” More than half came from core product and engineering organizations — the middle layer between leadership and execution.
Block’s cuts are the most explicit. Jack Dorsey told employees that “intelligence tools” made the reductions possible and predicted most companies would follow within a year. Bloomberg responded by flagging “suspicions of AI-washing.”
The 90% Problem
In February 2026, the National Bureau of Economic Research published a survey of 6,000 C-suite executives across the United States, United Kingdom, Germany, and Australia. The findings are hard to reconcile with the AI replacement narrative.
Roughly 90% of firms reported that AI had no impact on employment or productivity over the past three years. More than 80% reported zero productivity gains from AI despite billions invested. Executives use AI tools approximately 1.5 hours per week. A quarter don’t use them at all.
The executives themselves forecast a 1.4% productivity increase from AI over the next three years. Not 14%. Not 40%. One point four percent.
Economists are calling this the AI productivity paradox — an echo of Robert Solow’s observation in 1987 that “you can see the computer age everywhere but in the productivity statistics.” Nearly four decades later, the pattern repeats with a different technology and much larger capital commitments.
Even Sam Altman Says It’s Overblown
At the India AI Impact Summit in February 2026, OpenAI CEO Sam Altman acknowledged the gap between rhetoric and reality:
“I don’t know what the exact percentage is, but there’s some AI washing where people are blaming AI for layoffs that they would otherwise do, and then there’s some real displacement by AI of different kinds of jobs.”
A month later, he went further: “Almost every company that does layoffs is blaming AI, whether or not it really is about AI.”
This from the CEO of a company that just raised $110 billion at an $840 billion valuation — the largest venture deal in history, funded in part by $50 billion from Amazon and $30 billion from Nvidia. If anyone benefits from the narrative that AI replaces workers, it is Altman. That he is publicly qualifying it says something.
Where the Money Actually Goes
The $660 billion in combined Big 4 AI spending is not buying robot replacements for content moderators. Approximately 75% goes to data center construction, GPU procurement, and cloud infrastructure. The goal is not workforce replacement. It is platform revenue.
Meta‘s $125 billion funds custom chips (MTIA 300), data center expansion, and the AI models that power ad targeting and recommendation systems across Facebook and Instagram. These systems generate advertising revenue. The fired employees — many in corporate functions and content review — were a cost center. Cutting costs and building revenue infrastructure happen to coincide. They are not the same decision.
Amazon‘s $200 billion funds AWS AI services sold to enterprise customers. Amazon makes money when other companies use its AI infrastructure. The 16,000 people cut from corporate roles were not standing between Amazon and that revenue.
The Dallas Federal Reserve’s February 2026 research confirms the nuance: AI is simultaneously raising wages for experienced workers with tacit knowledge in AI-exposed occupations while reducing entry-level hiring in those same fields. The effect is bifurcation, not replacement.
The Real Number That Matters
Of the 45,000+ tech layoffs in Q1 2026, only 9,238 — about 20% — were explicitly attributed to AI and automation by the companies themselves. Challenger, Gray & Christmas tracked this. In all of 2025, AI accounted for fewer than 55,000 layoffs total, less than 1% of total job losses across the economy.
The gap between “AI is replacing everyone” and “AI accounts for 1% of layoffs” is where the $660 billion question lives. The spending is real. The job losses are real. The connection between them is, for most companies, a press release.
Sources: TechCrunch. “Meta reportedly considering layoffs.” March 14, 2026. (link) | CNBC. “Meta AI costs, mass layoffs.” March 16, 2026. (link) | CNBC. “Amazon layoffs, anti-bureaucracy, AI.” January 28, 2026. (link) | NBER. “AI Adoption and Firm Performance.” Working Paper 34836, February 2026. (link) | Fortune. “Sam Altman confirms AI washing.” February 19, 2026. (link) | Challenger, Gray & Christmas. AI-linked job cuts, Q1 2026. (link) | Dallas Fed. “AI, wages, and labor markets.” February 2026. (link)