Meta CEO Mark Zuckerberg acknowledged Wednesday that the company faces a fundamental tension as it weighs whether to lease excess AI computing capacity to outside customers or preserve it for its own model development — a decision that carries significant financial consequences in either direction.
The comments came during Meta's second-quarter earnings call, where the company also reported that free cash flow fell 90% from a year earlier, driven by soaring capital expenditures, and issued a weaker-than-expected revenue forecast for the third quarter.
"We're getting a lot of offers for compute at a significant premium over what we paid for it," Zuckerberg said. "It would be foolish to basically just sell all of the compute and take a short-term profit."
Meta is the only one among the four major U.S. hyperscalers — alongside Microsoft, Alphabet, and Amazon — that does not currently operate a cloud infrastructure business, even as its capital expenditure levels rival those of its peers. Zuckerberg has in recent months floated the possibility of launching such a business to capitalize on its compute stockpile amid tight market supply. Anthropic is in preliminary talks to lease computing power from Meta, according to earlier reporting.
"In terms of running the business, obviously, a common trade-off that we need to make is around how much do you monetize something today versus develop future assets," Zuckerberg said. "I think that it's always a portfolio."
Beyond raw compute leasing, Zuckerberg noted that a potential enterprise offering would also encompass API access, productivity services, and AI agents the company is actively building. He acknowledged that competing in the enterprise market — where Meta has historically had little presence — will require building new institutional capabilities.
"That's going to be somewhat a new muscle that we build as a company," he said. "But I think it's a very important one that we build."
Meta bumped up the low end of its 2026 capital expenditure guidance by $5 billion, setting the new range at between $130 billion and $145 billion. The revision arrives as the industry's hyperscaler spending race continues to accelerate: Alphabet last week raised the top end of its guidance to $205 billion while reporting negative free cash flow for the first time, and Microsoft said Wednesday that its capex for the year would reach approximately $175 billion.
Meta's stock fell more than 7% in after-hours trading following the earnings release, extending a decline that had already pushed shares down 11% for the year as of Wednesday's close.
The pressure on Zuckerberg to articulate a coherent AI strategy is mounting. "I think everyone wants clarity into what he wants to do in the compute business," Brent Thill, an analyst at Jefferies, told CNBC's "Closing Bell Overtime."
Meta's Reality Labs division, which develops virtual reality hardware and wearables, reported a loss of $4.62 billion in the quarter on revenue of just $431 million, underscoring the financial weight of the company's longer-term hardware bets. Digital advertising still accounts for 98% of Meta's total revenue.
The company has moved to bolster its AI leadership team. Dave Brown, a former senior executive at Amazon Web Services, is set to join Meta, and AI chief Alexandr Wang debuted the Muse Spark 1.1 model earlier this month, which Wang described as the "strongest model for agentic and coding work yet" at a lower cost than competing offerings from OpenAI and Anthropic.
Zuckerberg, for his part, expressed confidence that the capital commitment will prove justified. "My personal bet is that the people who invest in this are going to be rewarded and feel very good over time," he said.
With Amazon set to report earnings Thursday, investors will be watching whether another hyperscaler mirrors Meta's spending escalation or follows Microsoft's approach of holding its capital expenditure outlook steady — a discipline that drove Microsoft's stock up roughly 9% in after-hours trading Wednesday.
Disclaimer