Meta just announced it will sell off excess AI computing power to outside customers — and the move sent semiconductor stocks tumbling while Meta’s own shares jumped nearly 9%.

Story Snapshot

• Meta confirmed it will sell unused AI computing power to outside customers, launching a new cloud business to recoup billions in spending.
• Meta stock surged nearly 9% on the news, but chip stocks including Nvidia and Micron dropped sharply in response.
• Meta plans to spend up to $145 billion on infrastructure this year, yet already has computing power left over to sell.
• The move mirrors SpaceX’s strategy of selling unused data center capacity — sometimes at three times the market rate.

Meta Builds a Cloud Business Out of Leftover AI Power

Meta confirmed on July 1, 2026, that it will sell excess computing power to outside customers, creating a new cloud business in the process. The company is still deciding whether to sell raw computing access or offer hosted AI models. Bloomberg broke the story first, and CNBC confirmed it shortly after. Meta’s stock closed up nearly 9% that day, as investors welcomed a way for the company to earn back some of its massive infrastructure spending.

The timing raises eyebrows. Meta said it plans to spend as much as $145 billion on infrastructure in 2026 alone. Yet the company already has computing power it cannot fully use. CEO Mark Zuckerberg has previously said that selling excess capacity is an option “if we get to a point where we feel that we have overbuilt.” That word — overbuilt — is the one making investors nervous.

Chip Stocks Take a Hit as Market Reads the Signal

Not everyone celebrated the news. Semiconductor stocks fell sharply in pre-market trading after Meta’s announcement. Shares of Nvidia, Micron, and others dropped as traders interpreted the move as a sign that the biggest AI spenders may have more computing power than they need. Rival cloud computing firms CoreWeave and Nebius also saw their stocks fall 12 to 15 percent, as investors worried Meta’s entry into the cloud market would cut into their business.

This is the tension at the heart of the story. If the largest AI spenders are sitting on unused computing capacity, that could mean demand has not kept pace with the massive buildout of recent years. The AI hardware market is projected to grow from $151 billion in 2026 to $691 billion by 2033, according to Grand View Research — a 24% annual growth rate. But near-term oversupply and long-term growth can both be true at the same time. Markets are reacting to what is happening right now.

Meta’s Long-Term Plans Have Not Changed

Despite the sell-off in chip stocks, Meta has not pulled back on its big infrastructure ambitions. In January 2026, the company launched a new division called Meta Compute to oversee its global data center buildout. Zuckerberg said Meta plans to build “tens of gigawatts this decade, and hundreds of gigawatts or more over time.” The company has already signed 20-year power deals with nuclear energy facilities and is backing small modular reactor projects to fuel future data centers.

Every AI narrative said compute was scarce. Meta just proved they have enough to sell to others—and punished CoreWeave in the process.https://t.co/Kyjy5TirfP

— CuriousCats (@CuriousCats_US) July 2, 2026

The Meta Compute division is co-led by Santosh Janardhan, Meta’s head of global infrastructure, and Daniel Gross, a co-founder of Safe Superintelligence. Janardhan handles the technical side — running data centers and networks. Gross leads long-term capacity planning and supplier deals. This is not the organizational chart of a company backing away from AI. It looks more like a company that built fast, ended up ahead of its own demand curve, and is now finding a smart way to monetize the gap — just as Amazon, Google, and Microsoft did during earlier tech buildout cycles.

Smart Business Move or Warning Sign?

The honest answer is: it could be both. Selling excess capacity is a proven playbook. SpaceX already does it, reportedly charging companies like Anthropic and Google roughly three times the standard market rate for unused computing power. Meta following that model is not a sign of failure. But the fact that a company spending $145 billion in a single year still has capacity left over is worth watching. For everyday Americans already frustrated by AI hype and runaway corporate spending, this story is a useful reality check. The AI boom is real — but so is the risk of overbuilding.

Sources:

youtube.com, linkedin.com, networkworld.com, reuters.com, mordorintelligence.com