Meta posted the most profitable quarter in its history and the stock dropped 9 per cent. Revenue rose 33 per cent to $56.31 billion, beating Wall Street’s estimate of $55.49 billion. Net income reached $26.8 billion, up 61 per cent year over year. And for the first time since Meta began reporting its “family” of apps as a single unit, the number of people using Facebook, Instagram, WhatsApp, and Messenger on a daily basis declined. Daily active people fell more than 5 per cent sequentially to 3.56 billion, missing the 3.62 billion that analysts expected. The company that built the largest social network in history is now growing its profits faster than its audience, spending more on artificial intelligence infrastructure than any other company on earth, and cutting 8,000 workers whose jobs it has decided machines will do better. The market looked at all of this and concluded that the best quarter Meta has ever reported is not good enough for what Meta is becoming.

The numbers

The headline figures are unambiguous. Revenue of $56.31 billion in the first quarter represents Meta’s strongest top-line performance ever, driven by a 19 per cent increase in ad impressions and a 12 per cent rise in the average price per ad. The advertising machine that funds the entire operation is running at full capacity. But the net income figure requires an asterisk. Of the $26.8 billion Meta reported, $8.03 billion came from a one-time tax benefit resulting from the One Big Beautiful Bill Act, the Trump administration’s tax reform legislation that adjusted the treatment of previously capitalised research and development costs under the Corporate Alternative Minimum Tax. Without that benefit, net income was $18.7 billion and earnings per share were $7.31 rather than the reported $10.44. The quarter was still excellent by any historical standard. But a third of the profit improvement was a tax windfall, not an operational gain, and that distinction matters when investors are being asked to fund $145 billion in capital spending.

The user decline is what spooked the market. Meta reported 3.56 billion daily active people across its family of apps in March 2026, up 4 per cent year over year but down more than 5 per cent from the fourth quarter of 2025. The company attributed the sequential drop to internet disruptions caused by the Iran war and a government-imposed restriction on WhatsApp access in Russia. Chief financial officer Susan Li said on the earnings call that absent these two factors, daily active people would have grown quarter over quarter. That may well be true. But it is also the first time Meta has had to explain away a decline in its user base, and the explanation itself is revealing: the company’s growth is now exposed to geopolitical events in markets where it has no ability to influence outcomes. A war and a government ban erased the equivalent of roughly 190 million daily users in a single quarter. Whether those users return depends on events in Tehran and Moscow, not Menlo Park.

The spending

The 💜 of EU tech

The latest rumblings from the EU tech scene, a story from our wise ol’ founder Boris, and some questionable AI art. It’s free, every week, in your inbox. Sign up now!

Meta raised its full-year 2026 capital expenditure guidance to between $125 billion and $145 billion, up from the $115 billion to $135 billion range it had guided just three months earlier. The increase, which the company attributed to higher component pricing and expanded data centre capacity, means Meta will spend more on infrastructure this year than the gross domestic product of more than 130 countries. The money is going into Nvidia GPUs, custom chips from Amazon and Broadcom, data centres that now require their own power plants, and the compute foundation for what Mark Zuckerberg has called “personal superintelligence.” Meta has committed to building several generations of custom MTIA processors with Broadcom on a 2-nanometre process, signed a $27 billion infrastructure deal with Nebius, and is exploring space-based solar energy to power facilities whose electricity demands exceed what terrestrial grids can reliably supply.

Reality Labs, the division responsible for the Quest headsets, Ray-Ban Meta glasses, and the company’s augmented and virtual reality ambitions, posted $402 million in revenue and a $4.03 billion operating loss. Cumulative Reality Labs losses since Meta began breaking out the unit now exceed $90 billion. The division has absorbed rolling layoffs throughout 2026, with its budget cut by 30 per cent in earlier rounds before the company announced the broader 8,000-person reduction that will take effect on 20 May. Zuckerberg has not abandoned the metaverse thesis, but the capital allocation tells the story more clearly than any earnings call rhetoric: AI infrastructure is getting $145 billion, and Reality Labs is getting budget cuts.

The conversion

The layoffs are the clearest signal of where Meta believes its value will come from. The company announced on 23 April that it would cut approximately 8,000 employees, roughly 10 per cent of its 78,865-person workforce, with additional reductions planned for the second half of 2026. The cuts span Reality Labs, the Facebook social division, recruiting, sales, and global operations. Combined with earlier rounds in January and March that eliminated around 2,200 positions, and the 21,000 employees cut in 2022 and 2023, Zuckerberg has now reduced Meta’s headcount by approximately 25,000 people since the post-pandemic contraction began. The pattern is consistent: every quarter, the company reports higher revenue, higher profits, and fewer employees. The savings are redirected into AI infrastructure that Meta believes will generate more value than the workers it replaced.

The strategic logic is embodied in Meta Superintelligence Labs, the unit Zuckerberg created after spending $14.3 billion to acquire a 49 per cent stake in Scale AI and installing its founder, Alexandr Wang, as chief AI officer. The lab’s first model, Muse Spark, launched on 8 April as a closed-source, natively multimodal reasoning system that now powers Meta AI across Facebook, Instagram, WhatsApp, Messenger, and the Ray-Ban glasses. The decision to make Muse Spark closed-source, breaking from Meta’s Llama open-source tradition, reflects a calculation that the company’s most capable AI systems are too valuable to give away. Wang’s team, which includes five founders hired from Thinking Machines Lab at costs reportedly exceeding $1.5 billion for a single engineer, is building toward what Zuckerberg described as “delivering superintelligence to billions of people.” The ambition is not incremental. It is a bet that Meta’s future revenues will come not from showing more ads to more people, but from providing AI services so capable that they become essential infrastructure in their own right.

The question

The market’s reaction to Meta’s first quarter was not about the quarter itself. Revenue beat expectations. Advertising growth was strong. Even the user decline came with a plausible geopolitical explanation. What drove the stock down 9 per cent after hours was the forward-looking picture: a company that is increasing its spending commitments by $10 billion per quarter while its user base contracts, its metaverse division bleeds $4 billion every three months, and its most profitable line item was a one-time tax benefit that will not recur. Meta guided second-quarter revenue of $58 billion to $61 billion, which at the midpoint would represent 25 per cent growth. That is still exceptional for a company of Meta’s size. But the capex guidance of $125 billion to $145 billion means Meta expects to spend roughly $2.30 on infrastructure for every dollar of revenue it generates this year. The question investors are asking is not whether Meta can afford this. It clearly can, with $70.2 billion in cash and equivalents on its balance sheet. The question is whether the AI systems being built with that capital will produce returns that justify the expenditure before the advertising business, which funds everything, begins to slow.

Zuckerberg’s answer is that the AI systems will become the advertising business. Meta AI, powered by Muse Spark, is already being integrated into product surfaces across the family of apps, and the company has said that AI-generated content recommendations now drive a significant share of engagement on Facebook and Instagram. The thesis is that better AI produces better content recommendations, which produce higher engagement, which produces more ad impressions at higher prices, which funds more AI. It is a closed loop, and the first-quarter results suggest it is working: ad impressions up 19 per cent, price per ad up 12 per cent, total ad revenue up 33 per cent. But the user decline introduces a variable that the loop does not account for. If the number of people seeing those ads is shrinking, even temporarily, the growth has to come entirely from showing more ads to fewer people at higher prices. That model has a ceiling, and Meta has not yet demonstrated that AI can raise it. The best quarter in the company’s history was also the first in which it had to explain why fewer people are using its products. For a company spending $145 billion on the premise that AI will make its products indispensable, that is a contradiction it will need to resolve with more than a tax benefit and a geopolitical footnote.