Meta Platforms shares fell sharply in after-hours trading after the company raised its 2026 capital expenditure forecast to $125 billion to $145 billion, primarily for AI infrastructure.
Investors reacted to the disclosure despite first-quarter revenue of $56.3 billion and earnings per share of $10.44, both exceeding Wall Street estimates. The stock declined more than 8% as punters questioned timelines for AI payoff from the spending surge.
Meta confirmed plans for 10% workforce reductions in May to offset rising compute needs. CFO Susan Li noted the firm continued to underestimate AI requirements, prompting dynamic planning for capacity.
Chief Executive Mark Zuckerberg touted Muse Spark as a key milestone, with double-digit rises in Meta AI user sessions. Automated ad tools reached a $60 billion annual run rate, fueling revenue growth.
Daily active users increased year-over-year but dipped from the prior quarter due to internet disruptions in Iran and WhatsApp limits in Russia. Advantage+ campaigns drove outcomes like sales and leads.
The capex hike follows $72 billion spent in 2025, doubling down on data centers and chips from Nvidia, AMD, and Broadcom. Meta signed a $10 billion cloud deal with Google to supplement capacity.
Analysts trimmed price targets, citing risks from homegrown AI models trailing rivals. Bloomberg Intelligence flagged lower engagement versus frontier labs.
Big Tech peers like Alphabet and Microsoft also lifted spending projections, but Meta faced harsher punishment. Alphabet shares rose on cloud strength, underscoring divides in investor tolerance.
Layoffs tie directly to efficiency amid AI splurge, with internal directives targeting headcount cuts. The moves aim to accelerate operations while funding superintelligence pursuits.
Meta's Q2 revenue guidance spans $58 billion to $61 billion. Costs hold steady as capex balloons, pressuring margins short-term.
Traders positioned for volatility around payrolls and policy signals. The reaction signals broader scrutiny on AI economics across hyperscalers.