Meta commits $115-145 billion to superintelligence push with new Muse Spark model
Meta is making an unprecedented bet on artificial superintelligence, investing between 115 and 145 billion dollars in 2026 alone on computing infrastructure and AI talent. CEO Mark Zuckerberg launched Meta Superintelligence Labs last year and recently unveiled Muse Spark, the division's first major AI model, in April 2026. The company is constructing five massive data center clusters, including the Prometheus facility in Ohio, to power advanced AI training. Zuckerberg frames Meta's vision as developing personal superintelligence to empower individuals rather than automate work globally. However, the massive capital expenditure has sparked debate about whether such advancement warrants heightened regulatory oversight. Safety researchers and watchdog groups have raised concerns about insufficient control mechanisms and the lack of credible plans to prevent catastrophic misuse as companies race toward artificial general intelligence and beyond.
Key facts
- Meta's 2026 AI capital expenditures target between $115 billion and $135 billion, nearly double 2025 spending
- Muse Spark, released in April 2026, is Meta's first major model from Meta Superintelligence Labs, led by Chief AI Officer Alexandr Wang
- Meta is simultaneously building five gigawatt-scale data centers including Prometheus in Ohio and Hyperion in Louisiana
- A December 2025 Future of Life Institute study found that Meta and seven other major AI companies lack credible plans to prevent catastrophic AI risks
- Zuckerberg stated Meta would be careful about open-sourcing superintelligence models due to novel safety concerns
- Meta is the only major U.S. AI developer not to reach a voluntary agreement to share models with the federal government for review
How different outlets framed it
Left-leaning coverage: Progressive critics and safety advocates highlight Meta's insufficient safety measures and resistance to regulation. They point to past harms involving minors, criticize the company's shift from open-source to proprietary AI, and argue that massive AI investment diverts attention from current algorithmic harms. Left-leaning outlets emphasize that tech giants lobby against binding safety standards and that Meta should face stricter government oversight comparable to other industries.
Right-leaning coverage: Right-leaning commentators express concern about regulatory overreach while also pointing to Meta's content moderation practices as problematic. Some libertarian voices criticize government-mandated safety standards as stifling innovation, while conservative lawmakers have investigated whether Meta's AI systems enable harm to children. The right frames Meta's recent moves toward less restrictive content policies more favorably, though Republican figures have raised separate concerns about Meta's past fact-checking practices.
Limits of this briefing
Coverage reflects announcements and projections rather than delivery; superintelligence timeline and feasibility remain speculative. Most sources lack detailed technical safety research independent of industry narratives.
Sources
- SemiAnalysis Newsletter: The Future of Meta Superintelligence: A 1 Year Progress Update Centre
- Meta Official Blog: Introducing Muse Spark: Meta's Most Powerful Model Yet Right
- Built In: Meta Superintelligence Labs: What We Know So Far Centre
- Euronews: AI 'less regulated than sandwiches' and no tech firm has superintelligence safety plan, study Left
- Reason Magazine: The Age of Stupid—and Self-Serving—AI Regulation Is Here Right