Why Big Tech keeps laying off while making record profits

One scene kept repeating all year, and it reads like a typo. A company announces its best quarter ever, the stock climbs, and in the same week it sends thousands of people home. Meta notified roughly 8,000 employees in May 2026, in the same window it reported record quarterly revenue of $56.3 billion.⁸ Microsoft eliminated more than 15,000 positions across 2025, a year in which it cleared $27.2 billion in net income in the fiscal fourth quarter alone.⁴ ⁶ The math doesn't close. Unless you change the question. The money saved on people isn't vanishing; it's being spent somewhere else.

The number that won't balance on its own

Start with the size of the wave. In 2025, the tech sector announced 154,445 cuts, according to Challenger, Gray & Christmas, the outplacement firm that tracks U.S. layoff announcements. That's up about 15% from 2024 and the heaviest-cutting sector in the private economy.² The independent tracker Layoffs.fyi, which counts actual tech layoffs rather than announcements, logged somewhere between 122,000 and 124,000 over 2025.¹ The pace didn't ease in 2026: more than 100,000 jobs cut worldwide by early May.¹

The two trackers measure different things. One counts economy-wide announcements, the other counts completed tech cuts, so their totals don't match. The direction does, and it points up.

The other side of the scale: capex

Now the part that explains the paradox. Capex is capital expenditure, the money a company sinks into long-lived infrastructure. In this case data centers, chips, and networking to run AI. The big four (Alphabet, Amazon, Meta, Microsoft) spent roughly $381 billion on it in 2025. For 2026, the estimate runs as high as $725 billion, up about 77%.¹⁰ ¹⁴

By company, 2026 guidance lands near $200 billion at Amazon, $190 billion at Microsoft, $175 to $185 billion at Alphabet, and $115 to $145 billion at Meta.⁸ ¹⁰ Hold the two sides next to each other. Meta cuts 8,000 jobs, and Bank of America analysts estimate that saves on the order of $7 to $8 billion a year.⁸ In the same year it plans to spend up to $145 billion on infrastructure. The payroll savings are a small fraction of the GPU bill, but they help fund it, and the market reads the gesture as discipline.

The line going around among analysts says it better than any earnings deck: payroll is being converted into capital expenditure.⁹ Under that reading, the layoffs aren't cost-cutting in the old sense. They're the cash that bankrolls the AI build-out.

"Year of Efficiency," where the script came from

The playbook didn't start in 2026. On March 14, 2023, Mark Zuckerberg christened the "Year of Efficiency," announcing about 10,000 layoffs plus another 5,000 open roles cancelled, promising a flatter, leaner org.¹⁵ The market approved on the spot: Meta's stock jumped around 7% that day.¹⁶ Wall Street's takeaway was simple and blunt. Cutting people and flattening the hierarchy gets rewarded. In 2026 the script is the same. What changed is where the freed-up money goes, and now it goes to data centers.

A caution: not every executive confirms the thesis

Here's the caveat. The idea that "AI capex is financing the layoffs" is an analyst-and-press interpretation, supported by the arithmetic above, and not a corporate confession. And there's a direct counterexample. When Amazon announced 14,000 corporate cuts on October 28, 2025,¹¹ CEO Andy Jassy said on the earnings call that it "isn't even really financial, and it's not even really AI-driven, at least not right now," attributing the move to "culture," too many layers, and a need to move faster.¹³ The same Jassy, in a June 2025 memo, had written he expected the total corporate workforce to shrink over time as the company leaned harder on AI.¹¹ ¹³ Both statements coexist. Treat the stated motive with the skepticism it deserves, but don't mistake a strong economic hypothesis for a confessed fact.

What the community is saying

On tech-worker forums the mood is mostly skeptical and frustrated, with a pragmatic minority. (All of this is community opinion, not a receipt, and Acta Verum's crawler was blocked by Reddit for this reporting, so the read comes from Teamblind/Blind and from sentiment characterized in press coverage.)

The dominant read on Blind is that AI works less as a cause than as a narrative alibi: saying "we cut because of AI" lets a company shrink headcount and please investors in one move, since AI isn't actually doing the work of a huge slice of the staff. One thread captures the resignation: in this market, more layoffs read as more value, so it'll keep happening, "it worked for Meta, it becomes everyone's playbook." On the other side, some investors on the same forum treat the cuts as margin discipline and recession armor, the rules of the game rather than a scandal; and a few take AI substitution seriously where it's plausible, in repetitive coding roles. The complaint that surfaces most often isn't "AI is going to replace me." It's something colder: record profit doesn't protect me.

The verdict

The paradox is only on the surface. Laying off while profitable makes sense once a company has decided its capital is worth more buying GPUs than paying salaries, and once the market rewards exactly that move. The solid part of the story is the arithmetic: record profit, record capex, record cuts, all in the same quarter. The interpretive part, that one directly funds the other, is a strong hypothesis, not a confession, and Jassy's Amazon is the reminder that not every cut fits the same narrative. For people working in the sector, the practical takeaway is uncomfortable: in this phase, a company's performance and your job security have stopped moving together.