As the race to secure electricity for artificial intelligence data centers intensifies, nuclear fusion, long dubbed the technology that is always 20 years away, is becoming a target of big-tech power purchase deals and private investment. Many hurdles remain before commercial power.
Chosun Biz reported on the 17th that fusion is stepping out of the laboratory and emerging as an investment asset, a theme that echoes what Korean and foreign business media have pointed to in recent months. We could not check the original article directly, so this piece lays out the background from earlier reports such as Etoday and from publicly disclosed contract terms.
광고 문의 · 300×250According to Bloomberg and U.S. Energy Information Administration analysis cited by Etoday in July, U.S. electricity consumption is expected to hit record highs for two straight years, this year and next. As large data centers used for generative AI training and inference strain the grid, big tech is looking beyond gas, nuclear and renewables to fusion, which is not yet commercial, as a long-term power source.
Etoday reported that Microsoft signed a power purchase agreement with U.S. fusion startup Helion to receive electricity from 2028, and that Google signed a 200-megawatt supply deal with Commonwealth Fusion Systems (CFS). Helion was reported to have been valued at about 15.5 billion dollars (roughly 23 trillion won) in its latest fundraising.
OpenAI's moves are also cited. U.S. outlet Axios reported that OpenAI is negotiating with Helion to buy 5 gigawatts by 2030 and 50 gigawatts by 2035. This is a report of talks, not a finalized contract, and that distinction matters.
The fusion industry long relied heavily on government research budgets. With private capital and big-tech advance purchases now joining, observers see a new phase. Etoday noted that AI underpins the commercial economics of fusion, while fusion is eyed as a long-term source of the large-scale power AI demands.
Even so, the risks investors face are large. Delivery dates in the contracts depend on actual generation succeeding, and technical hurdles, regulation and heavy capital needs remain. A high valuation does not guarantee commercial success. Industry experts expect the next decade to decide energy supremacy in the AI era, but that is an outlook only, and schedules could slip.
