The 'silver bullet' of the AI industry, relieving the energy thirst of technology tycoons: Elon Mu
As Silicon Valley spends huge amounts of money on the artificial intelligence (AI) race, tech giants are running into a major bottleneck. Without abundant energy sources, they will not be able to build giant supercomputing centers that can host the next generation of AI.
CEO Mark Zuckerberg of Meta - Facebook's parent company, has purchased a large number of AI chips from Nvidia. In a podcast last month, Zuckerburg said it would take a gigawatt of power to run all those chips, potentially requiring a dedicated nuclear power supply. "One gigawatt for a data center will be equivalent to the capacity of a nuclear power plant," the Facebook boss said.
Mark Zuckerburg's comparison isn't exactly ridiculous. The fierce race for energy makes technology tycoons around the world increasingly interested in nuclear energy. Venture capitalists are becoming drivers of nuclear energy in the belief that this will be the energy source to sustain the AI boom.
Marc Andreessen, a prominent technology investor, strongly supports advances in areas such as nuclear energy and AI. He even called nuclear energy the "silver bullet" of artificial intelligence.
Elon Musk once said closing nuclear plants is "anti-humanity". Billionaires Bill Gates and Jeff Bezos have invested in many nuclear startups, ranging from small modular to fusion reactors.
OpenAI CEO Sam Altman - developer of ChatGPT, is a strong supporter of nuclear energy. In 2015, Altman invested in and agreed to serve as president of Oklo – a company that builds futuristic microreactors. Altman is also the largest outside investor in Helion when the company is looking for nuclear fusion, or thermonuclear, solutions.
Last year, Microsoft reached an agreement to buy electricity from Helion's fusion generator. It is expected that power will be supplied to Microsoft in 2028. In March, Amazon acquired a nuclear-powered data center in Pennsylvania. Meanwhile, DeepMind - Google's AI laboratory, is researching technology that uses AI to better control fusion reactions.
Last Energy, a Washington DC-based company developing small modular reactors, said half of its orders now come from companies building data centers, compared with the number was only 1/4 a few years ago.
The biggest key point is that AI requires a huge amount of energy. The International Energy Agency estimates that the amount of electricity used by data centers and AI could double by 2026. The amount of electricity needed for 1 question command to ChatGPT is estimated to be equal to 15 Google searches .
Although technology companies have invested heavily in wind and solar power, their intermittent nature may not be suitable for data centers that operate all day. Meanwhile, nuclear energy is always stable
