Bloomberg Intelligence Meta Signs Multi-Gigawatt Nuclear Deals for AI Data Centers
Jan 9, 2026
Mandeep Singh, Global Tech Research Head at Bloomberg Intelligence, discusses Meta's ambitious nuclear power strategy and its potential impact on their AI data centers. Michael Halen shares insights on the restaurant industry's recovery post-flu and severe weather, projecting a bright 2026 outlook. Craig Trudell analyzes General Motors' financial challenges in the EV sector, highlighting the stark contrasts between U.S. and European EV adoption, and advocates for hybrids as a sustainable interim solution.
AI Snips
Chapters
Transcript
Episode notes
Meta's Massive AI Power Bet
- Meta's multi-gigawatt purchases signal massive long-term AI data center CapEx priorities.
- One gigawatt of AI data center capacity costs about $50 billion, implying multiyear $100B+ spending by hyperscalers.
Why Nuclear Fits AI Scale
- Nuclear is an alternative to gas and renewables for large, continuous AI power needs.
- Companies face long lead times for adding power, making nuclear attractive for gigawatt-scale demand.
Model Efficiency Reduces Power Strain
- Efficiency improvements in AI models increase tokens per unit of power, stretching existing capacity.
- Firms maximize current power usage because adding new supply has long lead times.