Is Nuclear Fusion the Answer to AI’s Insatiable Energy Thirst?

AI has become death, the destroyer of worlds — and environments. AI consumes a lot of water and energy, but there’s hope that nuclear fusion, a type of clean energy, can help satisfy its growing hunger for power. The surge has coincided with the return of nuclear energy — driven by the current energy crisis and increased tax incentives — and recent breakthroughs in fusion.
Last week, the US’ largest steel provider, Nucor (NYSE:NUE), announced a $35M investment into Helion Energy and a partnership to develop a 500MW fusion power plant at their steel manufacturing facility. And unlike other sectors, industrial manufacturing requires enormous energy that’s difficult to fill with wind or solar.
Science 101: Nuclear fission is the method used in traditional nuclear reactors. But nuclear fusion, which powers the sun, is much more difficult to achieve. However, experts believe fusion power plants could arrive by 2050:
And the adoption of nuclear fusion could be getting a little help from tech giants.
AI consumes a massive amount of energy with heavy implications on the environment. UC Riverside and UT Arlington researchers found that it took 700K liters of water to train GPT-3 in Microsoft’s US data centers and that ChatGPT “drinks” a 500ml bottle of water for every ~20-50 questions/answers.
In May, Helion also signed a deal with Microsoft to provide it with energy in five years. And last week, Microsoft posted a job for a nuclear technology program manager — who will help utilize nuclear to power Microsoft’s data centers and AI. In an interview, the head of the Fusion Industry Association said, “The business world is starting to understand that fusion is coming and perhaps sooner than a lot of people thought” (Reuters).