Meta Is Playing a Completely Different Game Than the Coverage Suggests

Meta's announcement of 8,000 job cuts — roughly 10% of its total workforce — is dominating headlines. But the full picture is far different: Meta is simultaneously making hardware investments that dwarf the tech budgets of most countries.

The Numbers Are Staggering

Meta signed a $60 billion deal with AMD to purchase AI chips over five years, including both GPUs and customized CPUs, according to The Guardian. AMD CEO Lisa Su confirmed AMD will supply 6 gigawatts worth of chips to Meta, starting with 1 gigawatt of forthcoming MI400-series hardware in the second half of this year.

Separately, Meta extended its custom chip partnership with Broadcom through 2029, committing to 1 gigawatt of custom silicon in just the opening phase, according to Financial News. Broadcom CEO Hock Tan is reportedly moving into an advisory role on custom chip strategy — which Financial News notes removes the optics of any board-level conflict while maintaining real influence.

Meta's in-house chip program is equally aggressive. According to GSMGotech, Meta has committed to rolling out four new generations of its Meta Training and Inference Accelerator (MTIA) chip — the MTIA 300, 400, 450, and 500 — over the next two years. The MTIA 300 is already in production. That's a new chip generation roughly every six months, compared to the industry standard of 18-24 months.

Meanwhile, Meta is one of the founding partners of a $125 million Semiconductor Hub at UCLA's Samueli School of Engineering, alongside Broadcom, Applied Materials, GlobalFoundries, and Synopsys, according to CNBC. The five-year commitment includes doctoral internships. Applied Materials CEO Gary Dickerson called it essential for bringing "technology breakthroughs to market faster."

What's Actually Happening Here

Meta is not trying to out-ChatGPT OpenAI. That strategy is over.

Forrester analyst Alvin Nguyen told The Guardian that Meta appears to be pivoting away from competing directly with OpenAI and Anthropic, and toward becoming an AI infrastructure provider — data centers, chips, the physical backbone of the AI economy. Meta's $27 billion data center under construction in Louisiana fits that picture.

The AMD deal mirrors what OpenAI did when it reached scale. "OpenAI had to go multi-vendor because they got to a size where being locked in with just Nvidia limits their growth," Nguyen said. "Meta are already big enough where they need multiple options."

The custom chip program extends this strategy. By owning the silicon that runs inference workloads for Facebook, Instagram, and WhatsApp at scale — billions of users — Meta can dramatically cut the cost of serving AI features without paying Nvidia's margins every time.

Broadcom doesn't need to win the AI race. It just needs to keep selling shovels to everyone digging for gold.

The Layoffs and Infrastructure Strategy

Meta is spending tens of billions on chips and data center infrastructure. It's funding academic research hubs. It quietly launched a new Reddit-style app called Forum — spotted in the App Store by newsletter writer Matt Navarra — that nobody officially announced, according to Engadget.

And it's cutting 8,000 people from payroll.

This is not a company in financial distress. This is a company that has decided human labor is the variable cost and silicon infrastructure is the fixed investment worth doubling down on. WIRED's podcast team reported that employee morale at Meta is "increasingly grim" — and the layoffs are not the only reason. Record profits and record low morale: that's the internal story.

What Mainstream Coverage Is Missing

Most outlets are treating the layoffs and the chip spending as separate stories. They are not.

The layoffs fund the chip strategy. Both reflect the same strategic decision: bet on AI infrastructure, reduce human headcount, own the physical stack.

Coverage is also under-reporting the UCLA research hub as a workforce story. This isn't philanthropy. These companies are locking in a pipeline of PhD-level chip engineers before the talent war intensifies. Year-long doctoral internships with Meta, Broadcom, Applied Materials, GlobalFoundries, and Synopsys — that's a recruiting funnel wrapped in academic language. UCLA Engineering Dean Ah-Hyung "Alissa" Park told CNBC directly: the semiconductor industry in 10 years is unrecognizable, and the hub is designed to answer the hardest questions now.

The funding comes privately. No taxpayer dollars. Industry solving an industry problem.

The Wager

Meta is making a generational bet that the company which owns the chips, the data centers, and the talent pipeline will dominate the AI era — not the company with the flashiest chatbot.

8,000 people lost their jobs this week to help fund that bet.

Whether that trade-off is worth it depends entirely on whether you're a shareholder or a laid-off engineer. Zuckerberg has made his choice clear. The rest of us are watching it play out.