Meta reported first-quarter revenue up 33 per cent, its biggest year-on-year quarterly jump in nearly five years, and then warned it will spend even more on AI data centres than expected.
The outfit posted $56.3 billion in sales and net income of $26.8 billion, both ahead of analyst expectations, though $8 billion of that profit came from a first-quarter tax benefit.
Meta shoved its capital spending forecast up by $10 billion to a $125 billion to $145 billion range, blaming “expectations for higher component pricing” and “additional data centre costs.”
Daily active people, Meta’s user activity yardstick, dipped slightly quarter-on-quarter due to internet disruptions in Iran and a restriction on WhatsApp access in Russia, marking the first DAP decline since the company started reporting it in 2019.
For the current quarter, Meta reckons sales will land between $58 billion and $61 billion, roughly matching analysts’ expectations, while its shares sank more than five per cent in after-hours trading.
Meta chief executive Mark Zuckerberg pinned the capex jump on pricier memory, saying, “Every sign that we’re seeing in our own work and across the industry gives us confidence in this investment. That said, we are very focused on increasing the efficiency of our investments.”
Meta has already been stretching the life of some servers to cope with a memory chip shortage.
The spending fever is not confined to one company. Meta, Microsoft, Amazon, and Alphabet’s Google have announced plans to collectively invest more than $650 billion in AI infrastructure this year.
Meta’s earlier 2026 spending outlook was already a lurch up from $72 billion in capital expenditures last year, with the whole sector now moaning about a computing power crunch.
Pressed on the investor call about 2027 spending, Meta’s finance chief, Susan Li, kept it vague, saying, “Our experience so far has been that we have underestimated our compute needs, even as we have been ramping capacity significantly. So our expectation is that computing will become even more central to the business.”
Meta is cranking its AI push on two fronts, training fresh models while prodding its 70,000-plus workforce to wedge AI into daily work.
The company shipped its first new model since last summer’s AI team shake-up and the appointment of chief AI officer Alexandr Wang. The model, Muse Spark, launched in early April, and Meta claims it stacks up against other leading labs based on internal checks and external testing.
The frontier model race is getting messy. Meta has announced plans to lay off 10 per cent of staff, partly to bankroll the infrastructure spree.
In a recent memo, Meta technology chief Andrew Bosworth said AI agents will do most of today’s employee work in future, leaving humans to supervise, direct and help the bots improve.
