Meta is throwing unprecedented sums into artificial intelligence, with a capital expenditure plan that sets the stage for what could be the largest infrastructure buildout in tech history.
In its second-quarter earnings report released Wednesday, the company revealed plans to spend between $66 billion and $72 billion in 2025 on data centers, compute clusters, and technical infrastructure — a staggering increase of nearly $30 billion year-over-year at the midpoint.
The social media giant is not only doubling down — it’s preparing for a multi-year escalation. CFO Susan Li confirmed during the earnings call that 2026 will see an even more aggressive ramp-up as Meta pushes to dominate the AI frontier.
Register for Tekedia Mini-MBA edition 19 (Feb 9 – May 2, 2026): big discounts for early bird.
Tekedia AI in Business Masterclass opens registrations.
Join Tekedia Capital Syndicate and co-invest in great global startups.
Register for Tekedia AI Lab: From Technical Design to Deployment (next edition begins Jan 24 2026).
“We expect that developing leading AI infrastructure will be a core advantage in developing the best AI models and product experiences,” said Li. “So we expect to ramp our investments significantly in 2026 to support that work.”
Meta’s strategy hinges on scale and speed. The company is pouring capital into what it calls “titan clusters” — next-gen AI compute facilities that rival the scale of national power grids.
The first such cluster, Prometheus, located in Ohio, is expected to reach 1 gigawatt of compute power by 2026 — enough to support several large-scale AI models concurrently. Hyperion, the second, will rise in Louisiana and could eventually scale up to 5 gigawatts, with a land footprint comparable to Manhattan, according to CEO Mark Zuckerberg. Several more titan-scale projects are quietly underway.
Energy and Resource Concerns Mount
But Meta’s ambitions are raising red flags in the communities where these clusters are being built. In Newton County, Georgia, where one of Meta’s newer data center projects is under construction, residents have reported water taps running dry — a symptom of the immense stress placed on local infrastructure. Energy consumption is another flashpoint. A single gigawatt can power around 750,000 homes, and Meta is planning multiple clusters of this magnitude.
Critics say Meta’s expansion reflects a Silicon Valley-driven resource grab, one that prioritizes data and compute over sustainability and equity.
While Meta has said it’s exploring external financing models to co-develop some of these data centers, Li admitted there’s “nothing finalized” yet. Still, she noted that the company sees “significant interest from financial partners” and believes shared development models could provide flexibility in case infrastructure needs shift in the future.
AI Talent War and Superintelligence Labs
In tandem with infrastructure, Meta is spending lavishly on AI talent, naming employee compensation as its second-largest driver of growth. Its recently unveiled Superintelligence Labs unit is luring engineers and researchers with premium packages that, according to insiders, are pushing the industry’s pay ceiling.
Zuckerberg has cast the effort as building toward “personal superintelligence” — AI models that interact with people via smart glasses, VR headsets, and other wearables, with the aim of helping them “live their best lives.”
But these ambitions are not without cost. Meta’s Reality Labs, the division responsible for its metaverse and VR work, reported a $4.5 billion loss in Q2 — a reminder that while the company’s future-facing bets may excite Wall Street, they are bleeding money for now.
Stock Surge and Q3 Optimism
Investors shrugged off the Reality Labs loss. Meta’s stock surged more than 10% in after-hours trading, buoyed by $47.5 billion in Q2 revenue and a bullish outlook for the third quarter, where Meta expects $47.5 billion to $50.5 billion in revenue. Advertising, once again, was the engine, thanks in large part to Meta’s AI-powered ad tools, which help marketers auto-generate content and better target audiences.
AI has seeped into every layer of Meta’s business model — from user-facing products to ad infrastructure — but now, the company is making it clear that compute power is the next battleground. Meta believes the company that owns the most servers, data centers, and models will shape the future of not just advertising, but entertainment, communication, and knowledge.
With its $72 billion moonshot, Meta is betting it will be the one.