Meta Wants to Sell Its Spare AI Computing Power. Wall Street Cheered.

Today’s Focus

Meta’s share price climbed close to 10% after news outlets reported the company intends to rent out unused artificial-intelligence computing capacity to other businesses.

Bloomberg first described the plan, writing that Meta would offer access to the vast data-center resources it has been assembling. CNBC and the Wall Street Journal followed with their own accounts, each noting the roughly 9-10% jump in early Q3 trading.

The move would place Meta in a market long dominated by Amazon Web Services, Microsoft Azure, and Google Cloud. Those three companies rent out servers and specialized chips to firms that would rather not build their own.

Meta has spent heavily on graphics processing units and custom silicon to train its Llama models and power recommendation systems across Facebook and Instagram. Selling time on that hardware when it sits idle would turn a cost center into a possible source of income.

Yahoo Finance reported the surge came as some chipmaker stocks fell in the same session, a split investors read as a rotation within the AI trade rather than broad enthusiasm.

Barron’s framed the day as “AI Stocks Hit a Snag,” with Meta rising while several semiconductor names dropped. The reports did not confirm a launch date, pricing, or which customers Meta has approached. Meta had not published a formal announcement detailing the business as the reports circulated.

The Debate

Supporters argue

Backers of the move see a logical way to recover the enormous sums Meta has poured into computing hardware. Renting idle capacity converts a sunk cost into recurring revenue, the argument goes, much as Amazon turned its internal infrastructure into AWS.

Investors signaled approval directly through the roughly 10% jump that Yahoo Finance and the Wall Street Journal reported. That reaction suggests markets believe Meta can diversify beyond advertising, which supplies most of its income today.

Supporters also point to demand. Companies racing to build AI products face shortages of high-end chips and data-center space, so a new supplier could find ready buyers.

CNBC noted the plan targets “excess” compute, meaning Meta would sell capacity it already owns rather than build new facilities solely for rentals. To boosters, that lowers the risk and improves the odds the segment turns a profit quickly. A cloud arm would also give Meta a foothold in a business that has driven steady growth for its rivals.

Critics argue

Skeptics question whether Meta can compete against entrenched cloud giants with a decade head start. Amazon, Microsoft, and Google already operate global networks, sales teams, and enterprise relationships that Meta lacks.

Some analysts cited by Barron’s read the day’s trading as evidence of unease, not confidence, since chipmaker shares slid even as Meta rose. That divergence hints investors are nervous about how much AI spending will actually pay off.

Critics also flag the strategic tension. Renting compute to other firms could mean supplying rivals building competing AI products, and it may distract from Meta’s core advertising and social-media operations.

There is a trust dimension too. Meta has faced years of scrutiny over data practices, and enterprise customers weighing where to run sensitive workloads may hesitate. The reports offered no pricing or service guarantees, leaving open whether Meta can match the reliability that established providers promise.

What the experts say

Independent researchers have documented how large the AI infrastructure buildout has become. McKinsey Global Institute estimated in 2024 that data centers worldwide could require roughly $6.7 trillion in capital spending through 2030 to meet computing demand.

Analysts at Gartner projected worldwide public cloud spending would exceed $700 billion in 2025, underscoring the size of the market Meta is eyeing.

The cloud sector remains concentrated. Synergy Research Group has reported that Amazon, Microsoft, and Google together control roughly two-thirds of global cloud infrastructure spending, a share that has held steady for years.

Historical precedent cuts both ways. Amazon launched AWS in 2006 from its own retail infrastructure, and it became the company’s most profitable division. Other entrants, including IBM and Oracle, have struggled to gain comparable share despite deep resources.

Research from Stanford’s Institute for Human-Centered AI has tracked rising training costs, noting the price of computing power for frontier models has grown sharply, which helps explain why spare capacity carries real market value.

By the Numbers

~10%: approximate rise in Meta’s share price after the cloud reports, according to Yahoo Finance.

9%: the intraday gain CNBC and the Wall Street Journal reported for Meta stock.

~$700 billion: Gartner’s projection for global public cloud spending in 2025.

~$6.7 trillion: McKinsey Global Institute’s 2024 estimate of data-center capital needs through 2030.

~2/3: the combined share of global cloud infrastructure spending held by Amazon, Microsoft, and Google, per Synergy Research Group.

2006: the year Amazon launched AWS, the model Meta’s plan echoes.