Meta Platforms raised its 2026 capital spending forecast to as much as $145 billion on April 29, a jump of more than $25 billion from its prior guidance, and the stock promptly lost roughly 9% in after-hours trading. JPMorgan downgraded the shares the same evening, cutting its rating from Overweight to Neutral and warning that the ballooning AI budget could squeeze margins even as revenue keeps accelerating.
The reaction overshadowed a first quarter that, by any conventional measure, was excellent. Revenue rose 33% year over year to $56.3 billion, net income reached $26.8 billion, and diluted earnings per share of $10.44 cleared Wall Street estimates, according to Associated Press reporting on the results. Before the after-hours drop, Meta shares had gained roughly 25% year to date in 2026, making the stock one of the stronger performers among mega-cap technology names. But investors were not focused on the rearview mirror.
The numbers behind the quarter
Meta’s Q1 2026 results, disclosed in a Form 8-K filed with the SEC, confirmed that the core advertising business is running hot. The $56.3 billion revenue figure marked the company’s fastest year-over-year growth rate in several quarters. Net income of $26.8 billion and diluted EPS of $10.44 both topped consensus expectations.
Capital expenditures for the quarter alone hit $19.84 billion. Annualized, that already implied a full-year total near $80 billion. Free cash flow stayed strong enough to fund the buildout without significant new debt. Yet the forward guidance, not the quarterly results, is what sent shares lower.
Why the spending spike rattled Wall Street
On the earnings call, CFO Susan Li attributed the higher capex range “primarily” to expanded AI infrastructure investments, including new data center campuses, networking equipment, and custom silicon, according to Bloomberg’s account of the call. CEO Mark Zuckerberg was blunt about the rationale, arguing that the companies willing to pour money into AI infrastructure now will define the next generation of technology products.
The revised range of $125 billion to $145 billion is staggering by any measure. Meta’s total capital expenditures for all of 2024 were approximately $39.2 billion, per its annual 10-K filing. The company is now guiding toward roughly three and a half times that figure in a single year. For comparison, Microsoft guided toward approximately $80 billion in capex for its fiscal year 2025, and Alphabet announced a $75 billion target for calendar year 2025; neither company had publicly updated those figures for 2026 as of late May 2026. Meta’s upper bound blows past both prior-year targets.
JPMorgan’s downgrade sharpened the sell-off. The bank’s analysts, in a note reported by multiple outlets including CNBC and Bloomberg, flagged the risk that escalating costs could compress profit margins in the near term, even if the long-term AI strategy ultimately delivers. The move was notable because JPMorgan had been among Meta’s more vocal supporters over the prior year, during which the stock had been one of the best-performing names in the S&P 500.
Not every analyst shared JPMorgan’s caution. Evercore ISI maintained its Outperform rating on Meta in a note published the following day, arguing that the company’s revenue trajectory and dominant position in digital advertising justified the elevated spending. The contrasting views underscored how divided Wall Street had become over the right framework for valuing AI-heavy capital plans.
Where Meta says the money is going
Meta has outlined the broad categories of its AI investment without disclosing project-level dollar allocations. Its investor relations materials and earnings call commentary point to several major buckets: expanding data center capacity across the United States and internationally, developing custom AI training and inference chips to reduce dependence on Nvidia, and scaling the compute required to train increasingly large versions of its open-source Llama language models.
On the product side, Meta has tied the infrastructure push to specific business applications. Its Meta AI assistant, which Zuckerberg described on the Q1 2026 earnings call as attracting hundreds of millions of monthly users, requires enormous inference compute to run at scale. AI-powered improvements to ad-targeting algorithms have already contributed to the revenue acceleration visible in recent quarters. And the recommendation systems across Facebook, Instagram, and Threads are being rebuilt with larger, more sophisticated models that demand substantially more processing power.
What the company has not provided is a granular breakdown showing how the $125 billion to $145 billion splits across those initiatives, or a timeline for when specific investments are expected to produce measurable financial returns. Meta’s board has not publicly disclosed whether it formally approved the revised capex range or what governance guardrails, such as stage-gate reviews or spending caps tied to return thresholds, are in place to oversee outlays of this magnitude. That lack of detail is a central source of investor unease.
The gap between conviction and patience
The after-hours sell-off captures a tension that has been building across Big Tech for more than a year: management teams are convinced that massive AI spending is existential, while shareholders want proof that the money will come back. Meta is generating enough profit to self-fund one of the largest corporate infrastructure buildouts in history. But it is asking investors to accept a capex target that jumped by more than $25 billion in a single guidance update, with limited visibility into the return profile.
Meta’s 2025 annual report includes risk-factor disclosures warning that large-scale infrastructure projects can exceed budgets, face supply-chain bottlenecks, or fail to deliver expected performance improvements. Those boilerplate warnings carry more weight when the spending plan grows this fast.
There is also a timing mismatch. Even if Meta’s AI investments ultimately lift ad revenue, power new consumer products, and strengthen its competitive moat, the payoff could take years to show up in reported results. Quarterly earnings will reflect the costs immediately while the benefits accrue gradually, creating a stretch where margins could narrow and the stock could face sustained selling pressure.
The signposts that will shape Meta’s next move
Meta’s stock is now trading less as a bet on advertising strength and more as a referendum on AI spending discipline. The Q1 print proved the ad engine is performing. The open question is whether a company already deploying nearly $20 billion per quarter in capital projects can scale that pace toward $30 billion or more without running into execution problems, supply constraints, or diminishing returns on each incremental dollar.
Several markers will shape sentiment in the quarters ahead: whether Meta offers a more detailed capex breakdown at its next earnings call or investor day, how quickly AI-driven ad improvements translate into higher revenue per user, and whether rivals like Alphabet and Microsoft revise their own 2026 spending plans upward in ways that validate Meta’s aggressive posture. Until those data points arrive, the stock sits in an uncomfortable middle ground, powered by a business that keeps beating expectations but weighed down by a spending trajectory that keeps outpacing Wall Street’s tolerance.
