Overview
Meta is pouring large sums into the computing backbone required for advanced artificial intelligence - chips, servers, energy and data centers - even as that investment has driven free cash flow to just $784 million in the April-June quarter, a 91% decline from the same period a year earlier. The resulting strain on cash generation sent the company’s stock down about 9% in premarket trading the following day.
Management stance: keep and build
CEO Mark Zuckerberg framed compute as a strategically scarce asset that Meta should retain and expand around, rather than treating it primarily as an asset to be sold for short-term gains. He said the company had received offers from third parties to use Meta’s computing capacity, with some proposals coming in "at a meaningful premium" to Meta’s own cost of construction. That, he suggested, highlights the value of capacity but does not resolve the core strategic choice facing the company.
Zuckerberg argued the long-term prize for Meta lies in AI-driven consumer products and business services. He described potential mass-market applications such as AI-powered personal assistants for billions of users and business agents to support customer service, sales and marketing. Beyond those broad objectives, management provided limited detail on how subscription or enterprise services would, in practice, monetize the substantial AI spending.
Investor concerns and comparisons
Some investors reacted nervously, drawing parallels to the company’s earlier large-scale investment in the metaverse, which produced tens of billions of dollars in losses without evolving into a major revenue stream. The free cash flow slump in the latest quarter represents the steepest decline since late 2022, a period when investor scrutiny intensified around that same metaverse investment.
Analysts pressed leadership for specifics. When asked why Meta is both buying capacity from third parties and receiving offers to sell compute, Zuckerberg said the company is intentionally investing ahead of anticipated demand. "There is a lead time where we’re investing in building out these data centers now. They come online at some point in the future. You obviously are not getting value out of them until they’re online," he said.
Market commentators noted the asymmetry between Meta’s approach and that of large cloud providers. Josh Gilbert, lead APAC analyst at online investing platform eToro, summed up this dynamic concisely: "Meta is spending like a hyperscaler without a hyperscaler’s business model." The point is that companies such as Microsoft, Alphabet and Amazon have cloud businesses that can directly monetize data center investments by offering compute to enterprise customers, an outlet Meta does not currently possess to the same extent.
Competitive backdrop
The article highlighted Microsoft as an example of a company whose early AI investments are yielding commercial benefits. Microsoft reported a 23% fall in free cash flow but nevertheless beat expectations for growth in its Azure cloud unit and its Copilot assistant, a performance that pushed its stock up roughly 8% after the results. That contrast illustrates how entrenched enterprise relationships and an early move into AI infrastructure can translate into revenue and market enthusiasm.
Capital spending and industry capacity
Meta raised the low end of its capital expenditure forecast by $5 billion, setting a new range of $130 billion to $145 billion. Management tied the elevated spending level to what CFO Susan Li described as an industry-wide underbuild for AI demand. Li said that because capacity has historically been underprovided, existing compute is "extremely valuable," and she expects capacity to remain tight "for the foreseeable future." She suggested that tight capacity creates openings for Meta to generate returns via product development, enterprise services and compute sales, but did not pinpoint where the highest returns would come from.
When asked by a Wells Fargo analyst whether Meta expects to fully use its compute capacity internally through 2027, Li reiterated the company’s view that profitable uses exist both inside and outside Meta without specifying which would dominate. That mirrored the comments from Zuckerberg in emphasizing flexibility but providing limited clarity on near-term monetization.
Analyst reactions
Observers described the earnings discussion as exploratory in tone. Bernstein analyst Mark Shmulik said: "The earnings call felt a lot like a good old-fashioned brainstorming session." The remark reflected a broader sense among some market participants that Meta is still articulating how its heavy investment in compute will translate into sustainably profitable businesses beyond advertising.
Implications
Meta’s current approach to AI infrastructure raises a set of strategic trade-offs: retaining compute supports internal development of AI products and potential future enterprise services, while monetizing that same capacity could ease the immediate pressure on cash flows but divert resources away from proprietary model development. The company’s public statements emphasize both paths without committing exclusively to either.