Meta’s AI spending just hit a new milestone. So did investor anxiety.
The Facebook and Instagram parent said its free cash flow plunged 91% in the second quarter after pouring billions into AI infrastructure, leaving the company with just $784 million in free cash flow. The sharp decline came as Meta raised the low end of its capital spending forecast to as much as $145 billion, a figure that underscores Chief Executive Mark Zuckerberg’s belief that compute will define the next era of technology.
Meta stock slides 9% as investors question Zuckerberg’s $145B AI spending spree
Wall Street wasn’t convinced. Meta shares fell about 9% in premarket trading on Thursday as investors questioned how long the company can sustain one of the largest AI investment programs in corporate history before it produces meaningful financial returns, CNBC reported.
The earnings report exposed a question hanging over Meta’s AI strategy: Should the company keep its scarce computing capacity for its own AI ambitions, or rent part of it to outside customers willing to pay a premium?
That debate surfaced repeatedly during Meta’s earnings call.
Zuckerberg described compute as a scarce strategic asset that will become increasingly valuable as demand for AI continues to outpace supply. He revealed that companies have approached Meta with offers to buy access to its computing capacity “at a meaningful premium” over what the company spent to build it.
Meta has no shortage of potential customers. Demand for AI chips and data center capacity has outstripped supply across the industry, leaving many startups and enterprises scrambling for compute to train and run AI models.
Still, Zuckerberg signaled that Meta views those resources as far more valuable inside the company than as a simple cloud business.
“We believe that there will continue to be a significantly higher margin on selling intelligence rather than selling compute directly, but we think that there’s a big opportunity obviously to sell compute as well,” he said.
The comment offers one of the clearest looks yet at how Meta sees its long-term AI business.
Meta weighs selling AI compute as investors worry about soaring AI spending
Unlike Microsoft, Amazon, and Google, Meta does not operate a large enterprise cloud platform capable of selling excess computing power at scale. Its business has long relied on advertising across Facebook and Instagram, leaving investors searching for a clearer path from enormous AI spending to future revenue.
Microsoft offered a sharp contrast this week. The software giant reported that Azure cloud growth exceeded expectations, fueled by enterprise AI demand, sending its shares sharply higher after earnings. Its cloud business gives Microsoft a direct way to monetize the billions it spends on AI infrastructure.
Meta is making a different bet.
Zuckerberg argued that AI assistants serving billions of consumers and AI agents helping businesses with customer support, marketing, and sales could eventually become large businesses in their own right. Outside of broad references to subscriptions and enterprise services, Meta offered few details on how those products will translate into revenue.
Investors appeared to want more certainty.
Josh Gilbert, market analyst at online investing platform eToro, summed up the concern: “Meta is spending like a hyperscaler without a hyperscaler’s business model.”
His observation captures the challenge facing Meta. Microsoft, Amazon, and Google can point new data centers directly at cloud customers. Meta must rely largely on advertising today while betting that tomorrow’s AI products will justify today’s spending.
The comparison has revived memories of another expensive chapter in Meta’s history.
Several analysts drew parallels with the company’s multibillion-dollar push into the metaverse, a strategy that consumed tens of billions of dollars before producing a meaningful business. The latest quarter marked Meta’s steepest free cash flow decline since late 2022, the period when investors were raising similar concerns about metaverse spending.
Meta executives pushed back against those fears.
Chief Financial Officer Susan Li said the industry has historically underbuilt AI infrastructure, leaving compute in short supply and existing capacity “extremely valuable.” She added that demand is expected to remain tight for “the foreseeable future,” creating opportunities to generate returns through AI products, enterprise services, and compute sales.
Yet the company stopped short of explaining where those returns will come from first.
When asked whether Meta expects to consume all of its computing capacity internally through 2027, Li echoed Zuckerberg’s position that the company sees profitable opportunities both inside and outside Meta, without committing to one path, Reuters reported.
That lack of specificity left analysts looking for answers.
“The earnings call felt a lot like a good old-fashioned brainstorming session,” Bernstein analyst Mark Shmulik said.
For now, Meta is asking investors to accept a simple proposition: spend first, monetize later.
That approach has worked before in Silicon Valley. AI may prove to be another success story. Wall Street, for now, wants more than ambition. It wants evidence that a $145 billion AI buildout can become a business capable of generating returns on the same scale as the investment.
