🧠 Food for thought

Implications, context, and why it matters.

Helion’s concrete plans are far smaller than OpenAI’s goals

• Helion has shared a power purchase agreement (PPA) with Microsoft, which is a long-term electricity buying contract, for power from its first fusion plant set for 2028 1.
• That first site, Orion, includes financial penalties if Helion misses its delivery commitment 2.
• A 5-gigawatt OpenAI target by 2030 would be 100 times Helion’s 50-megawatt promise to Microsoft 1.
• That scale would arrive before Helion clears a technical bar since it has not produced more energy than its fusion reactions use 3.
• Helion expects its seventh prototype, Polaris, to generate electricity in 2024 1.

AI’s energy crunch is pushing a hunt for moonshot power sources

• The potential OpenAI deal answers rising AI power needs that are stretching current infrastructure 4.
• U.S. data centers could reach 426 terawatt-hours of electricity use by 2030, which is more than double 2024 levels, driven largely by AI 5.
• Sam Altman has reportedly said his AI plans require 250 gigawatts of new power by 2033, which experts call almost inconceivable with today’s grids 6.
• Traditional generation cannot expand fast enough, so large tech firms are backing high-risk options such as fusion to keep growth on track 4.
• Microsoft has also pursued conventional nuclear power and signed long-term PPAs for large volumes of carbon-free electricity starting in 2028 7.