Meta shares fall 3.1% on AI spending doubts, 2026 capex raised to $125B–$145B

Meta Platforms shares fell 3.1% in the afternoon session following a broad sell-off in AI-related technology stocks and mounting investor concerns about the company's capital spending on artificial intelligence. The primary catalyst was Meta's upward revision of projected capital expenditures for 2026 to a range of $125 billion（約20兆円） to $145 billion（約23兆円）. According to BMO Capital, Meta carries the "least visible return on investment story among peers" for its AI projects—a stark signal of investor doubt about the payout from such massive outlays. The concerns were compounded by an internal note from CEO Mark Zuckerberg indicating that AI agent development had not accelerated as expected, suggesting that progress may be falling short of earlier assumptions.

Macroeconomic headwinds and antitrust pressure also weighed on the stock. The Federal Reserve's restrictive policy stance and renewed global scrutiny over data privacy and online age verification added to the downward momentum. By the afternoon close, Meta shares were trading at $643.18, down 3.2% from the previous close.

One week earlier, the market had told a different story. Meta shares had gained 5.2% when the company framed its AI spending not as pure expense but as a path to lower costs, new revenue, and faster growth. That narrative rested on three pillars. First, cost reduction: Meta announced it would begin manufacturing its custom "Iris" AI chip in September with Broadcom and TSMC partners, a move designed to cut reliance on expensive Nvidia and AMD GPUs and reduce inference and training costs as the company increased ownership of its silicon stack. Second, monetization: Meta is building a $9.1 billion（約1.5兆円） (CAD 13 billion) data center in Canada and targeting 14GW of total compute by 2027, with the intent to rent excess capacity as an "AI Cloud" service—effectively converting idle or surplus infrastructure into a revenue stream. The company's build costs sit below the roughly $45 billion（約7.2兆円）-per-GW industry average, giving it an advantage in unit economics. Third, growth: Meta's advanced coding model, Muse Spark 1.1, is priced at about one-quarter of comparable offerings from OpenAI and Anthropic, a deliberate strategy to pull developers into Meta's ecosystem and expand the funnel for tools, ads, and cloud demand later.

However, the risk is execution. Custom silicon, multi-gigawatt data center buildouts, and an aggressive pricing model only generate returns if utilization rises, margins hold, and developer adoption materializes. Until those outcomes appear in financial results, markets retain the power to punish capex as overbuild rather than investment. Today's decline suggests that investor confidence in Meta's ability to convert $125 billion（約20兆円）–$145 billion（約23兆円） in spending into tangible returns has eroded.

Meta's share decline reflects a market recalibration around the company's AI capex strategy. Seven days prior, the stock had gained 5.2% when investors viewed Meta's heavy spending through a more optimistic lens—as an investment in manufacturing efficiency, new revenue streams, and platform growth rather than pure expense. That narrative centered on three levers: cost reduction through custom silicon ownership, monetization via the AI Cloud service targeting 14GW by 2027, and growth through aggressively priced offerings like Muse Spark 1.1, which is priced at about one-quarter of comparable OpenAI and Anthropic offerings to pull developers into Meta's ecosystem.

Today's sell-off indicates that confidence in this turnaround story has weakened. BMO Capital's assessment—that Meta has the least visible return on investment among peers—suggests that despite the detailed capex rationale, investors lack visibility into concrete milestones or financial proof points. Zuckerberg's internal note that AI agent development had not accelerated as expected compounds this doubt. The broader market context matters too: Federal Reserve restrictiveness and renewed antitrust scrutiny (data privacy and age verification) add headwinds unrelated to Meta's AI spending itself, but they clearly weigh on sentiment at a moment when the company's valuation already hinges on successful execution of an unproven capex-to-revenue formula.