Here’s a scenario that would make any CEO jealous: your company reports $20 billion in quarterly profit, revenue grows 26%, and you beat every analyst expectation. Your reward? A 12% stock plunge that vaporizes $200 billion in market value overnight.
Welcome to Meta’s paradoxical reality, where success isn’t measured in profits but in promises, specifically, promises about artificial intelligence that investors aren’t buying anymore.
The Great AI Money Pit
During Meta’s recent earnings call, Mark Zuckerberg delivered numbers that should have sent champagne corks popping. Instead, he triggered a mass investor exituation when he revealed the company’s capital expenditure forecast would hit $70-72 billion for 2025, with next year “notably larger.” Reports quickly emerged that Meta’s planning $600 billion in AI infrastructure spending over the next three years, more than the GDP of most countries.
For context, that’s enough money to buy every Nvidia H100 GPU expected to be produced through 2026, with billions left over. Yet when pressed for details about what exactly this mountain of cash would produce, Zuckerberg’s answer boiled down to “trust me bro we need the compute for superintelligence.”
The Ghost of Metaverse Past
The parallels to Meta’s metaverse disaster are impossible to ignore. Between 2021-2022, the company burned $36 billion on Reality Labs, watched its stock crater 77% from peak to bottom, and lost over $600 billion in market value. Investors endured years of Zuckerberg’s virtual reality vision while asking the same question they’re asking now: where’s the return?
As one analyst from Oppenheimer downgraded Meta’s stock, writing that the “significant investment in Superintelligence despite unknown revenue opportunity mirrors 2021/2022 metaverse spending.” The company’s return on invested capital dropped from a record 32% last quarter to 25% this quarter, a clear signal that money is pouring in faster than value is coming out.
The Competitive Context Problem
What makes Meta’s situation particularly awkward is watching its peers succeed where it’s struggling. Microsoft can point to Azure’s explosive growth from enterprise AI services. Google has AI integrated into its core search and advertising products. OpenAI pulls in $20 billion annually from ChatGPT’s 300 million weekly users.
Meta? 98% of its revenue still comes from ads on Facebook, Instagram, and WhatsApp, the same business model it’s had for a decade. The company’s AI assistant has “more than a billion active users”, but as TechCrunch’s Russell Brandom notes, those numbers are “surely juiced by the three billion active users on Facebook and Instagram.”
What $600 Billion Actually Buys
So what does this astronomical spending actually purchase? Meta’s throwing money at three primary targets:
Nvidia chips by the hundreds of thousands, with H100s and new Blackwell chips costing $30-40k each. Then there are the massive data centers, each consuming as much electricity as a small city. Finally, there’s the talent war, paying top AI researchers to compete with OpenAI, Google, and Anthropic.
The irony? Much of this spending flows directly to Meta’s competitors. When they need extra compute capacity, they rent from AWS, Google Cloud, and Azure. They buy chips from Nvidia. The entire AI infrastructure race is creating a circular economy where tech giants pass money between themselves while calling it economic growth.
The Superintelligence Gambit
Zuckerberg’s grand vision centers around Meta’s “Superintelligence team”, created four months ago when he restructured the company’s AI division and hired Alexandr Wang from Scale AI for a reported $14.3 billion. The goal: build AI smarter than humans.
But when analysts asked what products would emerge from this moonshot, Zuckerberg could only offer vague promises about “novel models and novel products” with “more to share in coming months.” For investors sitting through an earnings call, “coming months” sounds suspiciously like “trust the process.”
Even Meta’s formal $600 billion investment announcement came with awkwardness. During a White House dinner, Zuckerberg was caught on a hot mic telling President Trump, “I wasn’t sure what number you wanted to go with” regarding the spending figure. The informal nature of the commitment, now formalized in a blog post, did little to reassure Wall Street.
The Backup Plan That Terrifies Investors
Perhaps most concerning was Zuckerberg’s contingency plan if superintelligence takes longer than expected: “If it takes longer then we’ll use the extra compute to accelerate our core business which continues to be able to profitably use much more compute than we’ve been able to throw at it.”
Translation: if the moonshot fails, we’ll make ads slightly better. You’re spending $600 billion over three years and the fallback is marginally improved ad targeting?
Why This Matters Beyond Meta
As Steve Eisman, the investor famed for predicting the 2008 collapse, warned, Meta “cannot bear the burden” of the AI spending race against Google and Microsoft. He pointed to Meta’s cash reserves plummeting 43% in 2025 (from $77.8 billion to $44.4 billion) while competitors saw their war chests grow.
The company’s position as one of the Magnificent 7 stocks that make up 37% of the S&P 500 means when Meta sneezes, everyone’s 401k catches a cold. This isn’t just about one company’s questionable spending, it’s a warning shot for the entire AI investment frenzy.
If Wall Street starts questioning whether these massive AI investments will actually pay off, we could see a broader sell-off. Microsoft, Amazon, Alphabet are all spending similar amounts. If Meta, with its billions of users and dominant ad business, can’t justify the spending, what hope do others have?
The Growing Skepticism
The prevailing sentiment among developers and tech observers reflects deepening skepticism about where all this is headed. Many believe the best AI models are being kept in labs “for the elite”, while public-facing versions serve primarily as data harvesting tools. This perception undermines the narrative of AI as a democratizing force and raises questions about who ultimately benefits from these massive infrastructure investments.
As one critic summarized the situation: “You’re spending $600 billion over three years and the contingency is maybe your ad targeting gets 20% more efficient. Investors looked at that math and said this doesn’t add up.”
Meta finds itself at a critical inflection point. The company must either deliver groundbreaking AI products that justify this spending spree, something beyond slightly better ad recommendations, or face continued investor skepticism. The “if we build it, they will come” approach worked for infrastructure plays like AWS and Azure, but Meta lacks the enterprise customer base to monetize AI infrastructure directly.
Zuckerberg’s bet hinges on superintelligence arriving soon enough to justify the upfront costs. But in a market that’s grown skeptical of tech moonshots after the metaverse debacle, patience is wearing thin. The company’s challenge isn’t just building AI, it’s convincing Wall Street that this time really is different.
The coming months will determine whether Meta’s $600 billion gamble becomes the foundation of AI dominance or another expensive lesson in the dangers of betting big without a clear path to returns. For now, investors are voting with their wallets, and the message is clear: show us the products, not just the promises.
