Mark Zuckerberg got most of what he wanted out of Q1 2026: revenue of $56.31 billion, up 33% year over year, operating income up 30% to $22.9 billion, and a 61% jump in net income to $26.77 billion. The one number Wall Street actually cared about, however, was the one he raised after the bell. Meta now expects to spend between $125 billion and $145 billion on capital expenditure this year, up from a prior ceiling of $135 billion. The reaction was swift. Meta shares fell as much as 10% in early trading Thursday. JPMorgan analyst Doug Anmuth pulled his Buy rating. The narrative on the stock flipped from AI winner to AI question mark in less than 18 hours.
Underneath the noise, this is the moment investors have been waiting for. Big Tech has been spending an unprecedented amount of capital on artificial intelligence, with combined hyperscaler outlays for 2026 now projected to exceed $650 billion. The market has been broadly willing to underwrite the bet because the alternative was being told the cycle was over. Now, with Meta’s number leaking past $145 billion and a credible Wall Street analyst projecting it can keep going to $202 billion in 2027, investors are doing what investors are supposed to do: they are demanding to see the cash.
The Capex Number That Broke The Stock
Meta’s prior guidance had capex landing between $115 billion and $135 billion. The new range, $125 billion to $145 billion, looks small in percentage terms but is the third consecutive raise, and it took the high end well beyond what most sell-side models had penciled in. CFO Susan Li told analysts the increase reflected higher component pricing and additional data center costs to support future capacity, language that managed to sound both reassuring and ominous depending on the listener. Reassuring because demand for compute is real. Ominous because the math on returns gets harder every time the denominator gets bigger.
The user growth miss made things worse. Meta’s Family of Apps daily active users came in below the most aggressive Street targets, undermining the central case that the company can keep monetizing more attention from the same audience. Reality Labs again posted a multibillion-dollar operating loss. There was no clean place to look for the offset.
The JPMorgan Note That Lit The Fire
By Thursday morning, Doug Anmuth’s research note was the most-circulated piece of paper in tech. JPMorgan cut Meta to Neutral, citing what he called Meta’s “more challenging path to returns on heavy AI capex beyond advertising.” His base case, in plain English: full-stack AI competition is intensifying, the marginal AI dollar at Meta does not have an obvious revenue surface to land on, and the spending curve is bending the wrong way relative to free cash flow. Anmuth’s modeled capex peaks at $202 billion in 2027, which would push Meta into negative free cash flow of roughly $4 billion in 2026 and $24 billion in 2027.
Investors who have ridden Meta from the 2022 bottom know how those numbers will land in a portfolio review. Negative free cash flow is the line you never want to cross at a mature mega-cap, particularly when the buyback story has been load-bearing for the bull case.
The Hyperscaler Comparison Meta Cannot Win
Here is the part that should worry Meta shareholders most. Three of the four big AI capex spenders, namely Amazon, Microsoft, and Alphabet, all have public cloud businesses that can absorb excess capacity by renting compute to enterprise customers. AWS grew 28% in Q1, the fastest pace in 15 quarters. Google Cloud crossed $20 billion in quarterly revenue and grew 63%. Microsoft’s Azure remains the second-largest cloud business on Earth and a genuine monetization vehicle for capacity. Each of those companies can credibly argue that AI capex is feeding a paying customer base, even if the unit economics are still being negotiated.
Meta has none of that. Its monetization story for AI compute remains advertising, recommendations, and Reality Labs, with a long tail of internal productivity gains that are real but harder to map to revenue. That works at $50 billion of capex. It gets uncomfortable at $100 billion. It strains belief at $145 billion. And the JPMorgan model says we are not done.
The Silver Lining Bulls Are Pointing To
To be fair, the bull case has not gone away. Meta’s core advertising business is still growing 33% with operating leverage, an accomplishment that almost no other mega-cap can match. The company’s open-source Llama models continue to set distribution benchmarks across the developer ecosystem and lower the long-term cost of compute by giving Meta optionality on inference. The AI infrastructure being built in places like Louisiana and Wyoming will eventually compress operating expenses across recommendation, search, and ads ranking. There is a world in which Zuckerberg looks correct in 18 months and the stock more than reclaims today’s losses.
The problem is that the market does not give 18 months of patience to a company swinging into negative free cash flow. The question is not whether the long-term thesis works. It is whether shareholders will sit still long enough to find out.
What This Means For The Broader AI Trade
Meta is not the only company being repriced this week. The Wall Street Journal reported on Monday that OpenAI missed internal revenue and user targets, knocking Oracle, CoreWeave, and SoftBank lower. Microsoft slipped 5% Thursday on its own capex disclosures. The pattern is unmistakable. Investors are no longer underwriting AI capital expenditure as an undifferentiated bet on the future. They are underwriting specific revenue lines, specific customers, and specific operating leverage stories. Companies that can produce that math are being rewarded. Companies that cannot are getting their multiples compressed.
For Zuckerberg, the next two quarters become a referendum. Meta needs to either start showing AI-driven revenue acceleration in advertising or articulate a credible monetization path beyond ads, the way Google has articulated cloud. Without one of those, the JPMorgan downgrade will not be the last.
The Bottom Line For META Investors
Meta is no longer the best-loved name in the Magnificent Seven. The Q1 print was strong on every metric except the one Wall Street decided to make decisive. A $145 billion capex forecast in a year when the company will earn roughly $90 billion in operating income is a math problem, and it is one Zuckerberg has chosen to embrace publicly. Either AI delivers the productivity and engagement gains he is betting on, or this becomes the most expensive strategic pivot in technology history. There is not much in between. The stock now reflects, finally, that binary.
Investors should watch three things over the next six months: signs of advertising revenue acceleration tied directly to AI ranking improvements, any indication that Meta is exploring a third-party compute or platform business, and the absolute trajectory of free cash flow. The Q2 print will tell you most of what you need to know. For broader context on Meta’s earnings release, the official numbers are available via the company’s investor relations site. JPMorgan’s view is reflected in the broader research coverage tracked at Reuters’ Meta news hub.