Fusion is a 2030s story, not relevant to the current AI power cycle — but the capital bet is worth tracking

Confidence: 7/10 Supporting evidence:

• No commercial fusion has ever operated. CFS SPARC is 60% built but first plasma is 2027; ARC (commercial) is 2030s. Helion 2028 deadline is almost certainly delayed. Evidence: strong
• Private fusion investment is $10B — this is a real bet, not hype, but the timelines are 10-20 years. Evidence: strong

Challenging evidence:

• If Helion actually delivers 50 MW to Microsoft in 2028, the entire fusion timeline resets dramatically. Very low probability, but outcome is massive. Evidence: weak (highly uncertain)

Evolution:

• Jun 15, 2026 — Initial thesis at 7/10 (the "2028 Helion hedge" keeps it below 9/10)

Depends on: fusion Would change if: Helion achieves measurable net-energy-gain at Orion scale with an actual grid connection