Helion Energy, an energy startup based in Everett, Washington, USA, was founded in 2013. The company's website outlines an ambitious corporate vision: "Fully satisfy the world's need for clean energy, mitigate the adverse effects of climate change, and enhance human welfare." Despite being a startup of less than 11 years, Helion Energy's confidence in achieving the goal of "fully satisfy the world's need" stems from their exploration of technology regarded as the "ultimate energy for humanity"—controlled nuclear fusion technology.
Currently, there are two main technological approaches to achieving nuclear fusion: magnetic confinement fusion (tokamak) and inertial confinement fusion. Helion Energy's core technology, however, follows a "middle path," known as magneto-inertial fusion (MIF). Helion believes this technology can help them achieve nuclear fusion at sufficiently low costs, enabling the successful construction of the first commercial fusion reactor. Currently, Helion is building the world's first fusion power plant.
As a prominent company among nuclear fusion enterprises, Helion Energy's name has increasingly been linked with many tech and AI companies. In October 2023, Microsoft signed a power purchase agreement with Helion Energy, announcing plans to purchase electricity from Helion's first fusion power plant in 2028. Recently, the Wall Street Journal revealed that OpenAI is also in contact with Helion Energy, planning to procure a substantial amount of electricity from its fusion power plant to support its data centers' operations.
Speculations suggest this move is to meet the power demands of the supercomputer "Stargate," co-developed with Microsoft. This supercomputer, expected to cost up to $115 billion and equipped with millions of dedicated server chips, will require 5 gigawatts of power. Earlier this year, at the World Economic Forum 2024 in Davos, Sam Altman had already expressed concerns about a "power crisis," mentioning, "The AI industry is heading towards an energy crisis. The next wave of generative AI systems will consume more energy than anticipated, and the energy system will struggle to cope."
Similarly, Elon Musk predicted a "power crisis," stating, "Over a year ago, I predicted a chip shortage; the next shortage will be power. I believe there won't be enough electricity next year to run all the chips." Last week, the Electric Power Research Institute (EPRI) published a white paper on AI's power demand, noting, "By 2030, the total power consumption of U.S. data centers alone could more than double, reaching 166% of current levels." More specifically, the report found that AI-powered searches consume over ten times the energy of traditional search engines. The New Yorker also reported that ChatGPT's daily operation (responding to inferences, excluding training) could consume over 500,000 kWh daily. Based on EPRI's report, even under low-growth scenarios for future AI power consumption, data center power consumption could reach 19.63 billion kWh annually by 2030.
The significant energy consumption of AI, which has already become a present issue, raises concerns about a potential future energy crisis. Last year, the International Energy Agency's Global Electricity Market Report for 2023-2024 did not provide optimistic estimates for future energy supply. AI giants may have to face an impending AI energy crisis.
Not only OpenAI, but the U.S. government also anticipates potential future energy shortages. On the 6th, the U.S. Department of Energy released the "Fusion Energy Strategy 2024" and, along with the White House Office of Science and Technology Policy, commemorated the second anniversary of the Biden administration's "decadal vision for commercial fusion energy." In the meeting, the Biden administration reiterated their hope to support the development of fusion power plants that could be operational within the next decade. In the "Fusion Energy Strategy 2024," the Department of Energy allocated only $180 million in research funding to the Fusion Energy Innovation Research Engine (FIRE) collaboration, urgently hoping to see results within ten years. This relatively "insignificant" funding for nuclear fusion research seems to indicate the government's hesitation.
The transaction between OpenAI and Helion Energy has also faced scrutiny. Sam Altman, also Helion Energy's board chairman, holds $375 million in shares of this star company. Despite reports that Altman avoided direct involvement in the negotiations, these unclear relationships cast a shadow over the deal.