Meta CEO Mark Zuckerberg announced plans to invest hundreds of billions of dollars in building multiple AI data centers and research superclusters as part of a bold shift toward superintelligence. The new initiative, part of the recently launched Superintelligence Labs division, represents a strategic pivot following challenges with Llama 4 and key staff turnovers.
Under the plan, two core superclusters are set for launch—Prometheus in 2026 and Hyperion scale-out capacity to five gigawatts over time. Meta has upped its capital expenditure forecast to $64–72 billion for 2025, largely directed toward AI infrastructure, upgraded adtech, smart glasses, and AI-powered applications.
To bolster its AI capabilities, Meta is aggressively recruiting top-tier talent. CEO Zuckerberg has personally been involved in staffing decisions, offering nine-figure compensation packages to lure leaders such as former Scale AI CEO Alexandr Wang and ex-GitHub chief Nat Friedman. Meta reportedly invested $14.3 billion in Scale AI to reinforce its research base.
The announcement sent Meta shares up 1% amid wider tech sector momentum. The strategy follows mixed reception to Llama 4, prompting organizational restructuring to accelerate AI development and turbocharge future revenue from ad tools, smart devices, and AI-enabled services.
However, the ambitious plan may trigger investor concerns. Billions in long-term capex increase execution risk, and the market must judge if returns on superintelligence justify the near-term outlay. Meta will need to balance immediate profitability with expansive future commitments—a pattern mirrored by other AI-driven companies like Google and OpenAI.
Zuckerberg’s superintelligence commitment signals a pivotal inflection point for Meta and the tech industry. Its capacity to become an AI frontrunner depends on aligning technical achievement with financial discipline amid stakeholder expectations. The coming years will reveal whether this grand strategy delivers transformative innovation or proves overly aspirational.