Silicon Valley has a new business model: fire humans, hire GPUs.

Meta is planning to cut roughly 20% of its workforce — about 16,000 people — to offset the staggering cost of its AI infrastructure ambitions. Reuters broke the story on March 14, citing sources who say top executives have already told senior leaders to start identifying where the axe falls.

Meta spokesperson Andy Stone called it “speculative reporting about theoretical approaches.” Translation: we’re not announcing it yet.

But the writing isn’t just on the wall. It’s in the earnings calls, the restructuring memos, and the CEO interviews. Meta isn’t alone in this, and that’s what makes it terrifying.

The Pattern Is Unmistakable

This would be Meta’s largest reduction since the 2022-2023 “Year of Efficiency,” when Zuckerberg cut 21,000 jobs. But back then, the story was pandemic over-hiring correction. This time it’s different. These cuts aren’t about trimming fat — they’re about replacing human capability with machine capability.

The dominos have already started falling:

  • Amazon cut 16,000 jobs in January 2026, nearly 10% of its workforce, restructuring around AI.
  • Block slashed half its staff in February. Jack Dorsey said the quiet part loud: AI tools mean smaller teams can do more.
  • Meta is now preparing the biggest single round yet.

These aren’t struggling companies. They’re among the most profitable businesses on Earth, making a calculated bet that AI does it cheaper, faster, and at scale.

Zuckerberg’s $600 Billion Bet

To understand why 16,000 people might lose their jobs, follow the money.

Meta has committed $600 billion to data center construction through 2028. The company is paying AI researchers compensation packages worth hundreds of millions over four years. It acquired Chinese AI startup Manus for at least $2 billion. It bought Moltbook, a social networking platform built for AI agents.

Zuckerberg tipped his hand in January: “Projects that used to require big teams can now be accomplished by a single very talented person.”

One person plus AI equals ten. So why keep the other nine?

The math is cold. The execution is colder.

Morgan Stanley Says the “Intelligence Explosion” Is Imminent

Days before the Meta news broke, Morgan Stanley released a report warning that a “transformative leap” in AI could arrive in the first half of 2026. Not from a breathless AI startup — from one of the world’s most conservative financial institutions.

Their evidence: OpenAI’s GPT-5.4 “Thinking” model scored 83% on the GDPVal benchmark, placing it at or above human expert level on economically valuable tasks. Morgan Stanley’s conclusion is blunt — AI is becoming a “powerful deflationary force,” and executives are already executing large-scale workforce reductions because of it.

The report even flagged a timeline for recursive self-improvement — AI systems autonomously upgrading their own capabilities — as early as the first half of 2027.

When Morgan Stanley starts talking about intelligence explosions, the corporate world listens.

The Physical Wall Nobody Mentions

Here’s the twist: all this AI ambition is crashing into physics.

Morgan Stanley projects a net U.S. power shortfall of 9 to 18 gigawatts through 2028 — a 12% to 25% deficit in the energy needed to run AI infrastructure. Companies are converting Bitcoin mines into compute centers, deploying gas turbines, and fighting over every available megawatt.

A new “15-15-15” dynamic is emerging: 15-year data center leases at 15% yields, generating $15 per watt in net value. The AI race isn’t just a software competition anymore. It’s a land grab for power, cooling, and physical infrastructure.

Meta’s $600 billion isn’t optional spending. It’s the price of admission. And that bill has to come from somewhere. For 16,000 employees, the answer is their paychecks.

Who Gets Crushed, Who Gets Ahead

The uncomfortable truth is that this restructuring creates winners and losers, and the dividing line is brutally clear.

The exposed: Mid-level knowledge workers. Project coordinators, data analysts, content teams, routine engineering. If AI handles 80% of your job, you’re a line item waiting to be optimized.

The beneficiaries: Small teams and startups. If five people plus AI can outperform fifty, the startup advantage just multiplied. Sam Altman’s vision of hyper-lean companies is becoming a blueprint, not a talking point.

The elite: Top-tier AI talent is commanding astronomical pay. Meta is offering researchers hundreds of millions. The premium on building AI versus being replaced by it has never been wider.

The absent: Governments. Companies are moving at breakneck speed. Comprehensive workforce transition policies? Nowhere close.

The Quiet Part, Said Loud

Here’s what’s actually happening: the largest companies in history are firing humans specifically to build machines that do those humans’ jobs. And they’re barely pretending otherwise.

Dorsey said it plainly. Zuckerberg implied it clearly. Morgan Stanley put it in a research note.

The “Year of Efficiency” wasn’t a one-time correction. It was the opening act of a permanent restructuring of how companies think about labor. Sixteen thousand jobs at one company, in one quarter, to fund servers.

The question isn’t whether AI will transform the workforce — it already has. The question is whether anyone is building the safety nets, retraining programs, and economic models to catch the people falling through. So far, the answer is a resounding not yet.

And Meta is just one domino.