Your Job Funded the AI That Replaced You. You Are Welcome.

Mark Zuckerberg laid off 8,000 people in May and was remarkably honest about why. The company needed the money for GPUs. Not a metaphor, not corporate speak for “restructuring towards growth” — he meant the actual GPUs sitting in the actual data centers that Meta is currently spending $145 billion to build this year alone. He looked his employees in the eye, figuratively speaking, and said the budget line beat them. At least he did not call it a “difficult but necessary step forward for the business.”

Amazon, Microsoft, and Alphabet watched and apparently thought: yes, that is the move. Between the four of them, Meta, Amazon, Microsoft, and Alphabet will spend a combined $725 billion on AI infrastructure in 2026. That is a 77 percent jump over last year’s record of $410 billion, which was itself a jump over the year before. Meanwhile, the industry has cut somewhere between 95,000 and 150,000 jobs this year depending on which tracker you trust, and they are all tracking the same catastrophe from slightly different angles.

What $725 Billion Actually Looks Like

To put $725 billion in perspective, these four companies now spend more on AI infrastructure than the entire global oil and gas industry spends on finding new oil. Let that sit. The people who brought you targeted ads and a search engine are outspending the people who power the planet.

Amazon committed $200 billion. Alphabet set its number at $180 billion, which is double what it spent last year. Microsoft came in at $190 billion, a figure so far above analyst forecasts that even the analysts seemed briefly embarrassed for themselves. Meta raised its range to between $125 and $145 billion, which, if you break it down, is roughly $370 million a day on data center construction. Every single day. Including weekends.

All of it goes into data centers, GPUs, custom chips, networking equipment, and AI models. The payroll budget did not get a similar upgrade. You may have noticed.

The Layoff Numbers, Since Someone Should Say Them Out Loud

More than 81,000 tech workers lost their jobs in the first quarter of 2026, the highest quarterly figure in at least two years. Other trackers put the year-to-date number closer to 113,000. One tracker had already logged over 150,000 by mid-April, before Meta and Microsoft had even finished handing out the bad news. The industry is cutting around 864 people every day, which is not a statistic anyone put on a slide at a shareholder meeting.

Amazon cut 30,000 roles in five months while reporting AWS growth of 24 percent, its fastest in 13 quarters. So business was booming and people still got cut, which really tells you everything you need to know about what this moment actually is. Meta’s 8,000 May cuts hit recruiting and HR hardest, which has a certain poetry to it — the people whose job was to bring other people in are now the ones being shown the door. Microsoft quietly handed voluntary departure offers to around 125,000 employees, which is a polite way of describing one of the largest workforce reductions in corporate history. Alphabet trimmed 1,500.

Cloudflare cut 1,100 people, 20 percent of its entire company, after proudly announcing that internal AI usage had jumped 600 percent in three months. The staff who spent years building that culture apparently helped make themselves redundant. Progress.

The Math That Nobody Wants to Do in Public

Meta spends roughly $27 billion a year on total human compensation — every salary, every bonus, every stock grant for every person who works there. Its 2026 AI capex is up to $145 billion. That means Meta spends five times more on compute than on the humans operating the company. If Meta fired literally every single employee tomorrow, it would not generate enough savings to close the gap between what it spends on people and what it is spending on machines. The layoffs are not paying for the AI. The company is rich enough to do both. The cuts are a choice, not a necessity.

Zuckerberg was the only one honest enough to say it directly. Everyone else is still hiding behind language like “strategic realignment” and “optimising for the next phase of growth,” which translates, fairly reliably, to: we would rather buy servers.

There Are 275,000 AI Jobs Open and Nobody to Fill Them

Here is the part of the story that tends to get buried. There are currently 275,000 open AI-related roles across the tech industry. They are sitting empty. The people losing their jobs in customer support, HR, recruiting, and mid-level software development cannot walk into those roles, because the skills required are entirely different and the gap between the two is not something a three-week online course is going to close.

So the industry is not replacing humans with AI in the clean, linear way the headlines suggest. It is eliminating one category of work and creating demand for a completely different one, with a massive skills canyon sitting in the middle that nobody has a credible plan to bridge. Salesforce CEO Marc Benioff announced 4,000 customer support cuts and told analysts he simply needs fewer people in that function. He did not say he needs those people retrained and doing something else. He needs fewer of them. Full stop.

Why the Bills Keep Getting Bigger

Part of what is pushing costs higher is not just ambition — it is prices. Microsoft CFO Amy Hood told investors that rising memory chip costs account for $25 billion of the company’s capex budget on their own. Meta raised its forecast by $10 billion citing the same pressure. Land is expensive, power is expensive, and the construction workers needed to build data centers at this scale are very expensive. Everyone is competing for the same physical resources at the same time and acting surprised that it costs more than expected.

The revenue case for all of this spending is real, to be fair. Google Cloud grew 63 percent last quarter. AWS grew 24 percent, which Amazon will tell you is its fastest rate in 13 quarters. The money is coming in. The question is whether it comes in fast enough and at enough scale to justify what is going out. One investor described Meta as a once capital-light money machine that may be turning into a capital-intensive incinerator, which is the kind of sentence that gets you quoted everywhere but does not actually stop anyone from raising their capex guidance.

So What Is Actually Happening Here

The $725 billion and the 100,000-plus layoffs are not two stories running in parallel. They are the same story told from opposite ends of the org chart. These companies have made a calculated decision that AI infrastructure will generate more value per dollar than the roles they are cutting, and they are betting their own balance sheets on it. Markets largely agree with them, for now. Big Tech stocks have held up fine through all of this, which means the people with money in the game are not particularly worried.

The workers who lost jobs in Q1 2026 do not have the option of waiting to see how the bet resolves. For them it already resolved. Their company picked a GPU over them, and at least one CEO was decent enough to say so.