Meta’s AI ambitions ignite capital spend surge

• Alphabet, Microsoft, Amazon, and OpenAI are also scaling up, fueling concerns of an industry-wide AI bubble

Meta Platforms jolted investors with a bold forecast for “notably larger” capital expenditures in 2026, driven by an aggressive push into artificial intelligence.

The tech giant revealed that surging investments—primarily in new AI data centres—would fuel a sharp rise in costs as it races to close the AI gap with rivals Microsoft and Alphabet.

While Meta reported a robust 26 per cent year-on-year increase in third-quarter revenue that exceeded estimates, investor enthusiasm quickly faded as costs outpaced sales, jumping 32 per cent compared to last year. Shares, up 28 per cent year-to-date, tumbled 8 per cent in after-hours trading as Wall Street reacted to CEO Mark Zuckerberg’s expansive capital plans—moves expected to compress profit margins.

Betting big on superintelligence

Meta is now front-loading billions into expanding its AI computing infrastructure, eyeing “superintelligence”: the industry’s aspirational milestone where machines surpass human intellect. “There’s a range of timelines for when people think that we’re going to get superintelligence,” Zuckerberg told analysts.

“I think it’s the right strategy to aggressively front-load building capacity so that we’re prepared for the most optimistic cases.”
If the milestone is delayed, Meta plans to deploy the surplus computing power to accelerate its core businesses, adding flexibility to the strategy. But the immediate pressure is clear: the company upped its capital expenditure outlook for 2025 to $70–72 billion, from a prior floor of $66 billion.

Meta’s bottom line this quarter absorbed a near $16 billion one−time charge tied to President Trump’s “Big Beautiful Bill,” slashing reported net income to $2.71 billion. Strip out the charge, and net income would have reached $18.64 billion—demonstrating continued operating strength beneath headline numbers.

Market competition

Meta has doubled down on AI capabilities after a late start, recently reorganising its AI ambitions under the new “Superintelligence Labs” and snapping up top AI talent and Nvidia’s elite chips.

The push is already yielding returns: Meta’s AI-powered advertising platform allows marketers to automate campaigns, translate and generate content, and target segmented audiences at scale.
The AI data centre spending spree isn’t unique to Meta—Alphabet, Microsoft, Amazon, and OpenAI are also scaling up, fueling concerns of an industry-wide AI bubble.

OpenAI CEO Sam Altman recently set an eye-watering goal of building out 1 gigawatt of compute capacity every week—with each gigawatt costing over $40 billion. Cost pressures and the demands for future results are prompting tech giants to seek partnerships and scrutinise spending.

Meta’s platforms continue to dominate, with more than 3.5 billion people using at least one of its apps daily. The company is expanding monetisation, recently launching ads on WhatsApp and Threads, and challenging TikTok and YouTube Shorts with Instagram Reels.

For the current quarter, Meta forecasts revenue between $56–59 billion, bracketing analysts’ estimates and reinforcing its resilience despite escalating expenses.