There’s an terrible lot of hype surrounding AI, and much more capital behind it. AI information facilities are propping up the useconomy proper now, accounting for over 1% of GDP. So when somebody suggests the race might not be being received, or worse, that the AI race could also be over, it makes folks sit up of their seats.
That’s what Adam Livingston, writer of The Bitcoin Age, claims: it’s already recreation over: China has pulled far forward, not by out-coding the U.S., however by quietly cornering the one useful resource frontier AI wants most—power, particularly nuclear energy.
However how a lot fact is there to this narrative, and are issues actually so black and white?
Nuclear scoreboard: reality vs. fiction
Livingston highlights a placing disparity. China is at present constructing 16 nuclear energy vegetation, whereas america has zero. He’s not far off together with his numbers. As of late 2025, China has about 30 reactors below building, with repeated yearly approval for extra, making up practically half the world’s new builds.
Some analysts say China goals to achieve 65 gigawatts of nuclear capability by the top of this 12 months and 200 gigawatts by 2040 (roughly a tenfold development).
Against this, the U.S. accomplished its Vogtle 3 and 4 reactors after prolonged delays and price overruns. At present, there aren’t any brand-new large-scale nuclear initiatives on the ground-breaking stage.
But, this isn’t the entire image. For the primary time in years, there are new plans for U.S. nuclear. Following latest government orders and coverage reforms, Westinghouse introduced intentions to assemble 10 huge reactors by 2030. Work is predicted to start within the subsequent few years.
Nonetheless, regulatory hurdles, public skepticism, and the sheer complexity of nuclear buildouts imply execution is much from assured, and precise new building is just not but underway.
Vitality: the true AI bottleneck?
Livingston poses an necessary query: Are we underestimating the function of pure power in AI progress? Mannequin coaching and inference have turn into ravenous for electrical energy.
Coaching frontier fashions like GPT-4 requires tens of megawatts, and data-center energy demand within the U.S. is projected to greater than double over the subsequent decade (as a lot as 78 gigawatts by 2035).
World information heart power consumption hit 415 terawatt-hours in 2024, forecast to double by 2030, with AI accounting for a rising share. So, in idea, nations that may deploy probably the most regular, carbon-free energy will certainly have a bonus within the AI race.
China’s strategy to industrial coverage is direct, top-down, and aggressive. It has allowed it to ramp up nuclear building quickly, whereas American utilities have relied extra on upgrades, license extensions, and sluggish, market-based exercise.
However whereas China is advancing quick, the U.S. can also be centered on enhancing effectivity and leveraging new applied sciences akin to Small Modular Reactors (SMRs) and renewables to complement its base.
Is the AI race already over?
Has “the funeral already occurred,” as Livingston claims? The reply appears much less definitive. China’s nuclear enlargement is actual and spectacular, and its linkage to AI infrastructure is just not far-fetched. AI does critically rely upon steady, inexpensive energy.
Nonetheless, U.S. leaders and firms usually are not precisely standing nonetheless. New initiatives, coverage strikes, and elevated funding in each power and AI are selecting up, however thus far, they don’t seem to be matching China’s scale or velocity.
The American benefit in foundational AI analysis, chip design, cloud infrastructure, and enterprise funding stays vital. Even when data-center energy turns into a constraint, innovation in effectivity, good grids, and distributed computation might slim the hole.
Notably, “power wars” might turn into as necessary as software program or information, however the final result will rely upon far more than the variety of nuclear vegetation alone. Livingston’s arguments spotlight an neglected facet of the worldwide tech wrestle, however declare the funeral untimely. The scoreboard is altering, however the AI race isn’t over but.
