Tiangong Ultra ran 400 meters in Beijing on Sunday in 38.15 seconds, nearly five seconds inside Wayde van Niekerk’s world record. It competes in the World Humanoid Robot Games, so its time will not replace Wayde’s in the record book. Impressive, as is the 57% improvement on last year’s winning 1:28.03.
Energy technologies can look goofy, even laughably primitive, for much of their lives. Then one day they aren’t. Wind and utility-scale solar went from less than 1% of U.S. generation in 2005 to 17% in 2025. Fusion may eventually do the same.
Yes, the same fusion that gets snickers from the crowd. Private fusion companies report $15.2 billion of cumulative funding, so somebody believes. And 42% of developers surveyed by the Fusion Industry Association expect the first commercial fusion plant to deliver electricity by 2035. Comically soon for a technology seventy years in the works. After watching the robots cut 57% off the winning time in a year, and seeing AI speed up work across the sciences, I am less inclined to snicker.
The robots race again next year. Fusion gets one run a decade. Even taking developers at their word, the 4 GW to 6.5 GW disclosed by U.S. companies is small beside the 79 GW of grid-served data center load we expect by 2035. Fusion may eventually stop being funny, but not in time to carry this build cycle.
Existing fission is already doing useful work. Our nuclear power purchase agreement screen identifies roughly 2.1 GW of uprate potential across 14 plants, including 340 MW at Limerick. No new site or reactor required. META’s Vistra agreement supports 2,176 MW of operating generation and another 433 MW of uprates in the early 2030s.
Across six independent power producer nuclear deals, we estimate an average first-year price of $86/MWh, a $34/MWh premium to zonal forwards. AMZN’s premium over merchant power rose 73% between Susquehanna and Comanche Peak. Buyers are paying for time.
Our IREN work explains why. Initial build costs run from $31 million to $50 million/MW, and at full build we model about $9 million/MW of annual revenue at an 83% EBITDA margin. At a 95% capacity factor, twenty years at that nuclear premium costs about $5.7 million/MW before escalation. One year of modeled EBITDA is worth more.
Holtec Nuclear’s planned IPO brings both bets together: an operating nuclear-services business and the Palisades restart, alongside the SMR-300. The market will decide how much to pay for useful work and how much for the possibility that one day the robot stops looking goofy.
Comments, questions or things I missed? Send me a note (or hit reply) - I would love to hear from you. Thanks for reading!
Morning Energy is a syndicated note published through Enverus Intelligence. My contributions will also be distributed here. Please note that links frequently lead to content available only to subscribers of Enverus solutions. Please reach out if you have any questions. Thanks! - Ian.