Going Nuclear: Why AI Will Lead the Next Energy Transition
As policies and regulations are adjusted to reflect a new presidential administration, the U.S. needs to unleash its full energy potential by embracing nuclear power. With opinions and regulations historically slow to adjust, the impetus for nuclear power may depend on the tech companies whose race for artificial intelligence (AI) may at last solve the green energy crisis. Even if AI never achieves the world altering potential its advocates promise, it may yet still hold the key for our future.
COMMENTARY
The energy demanded by AI development and use is enormous, and the source of much handwringing among those concerned about the environment. Never before has there been such an apparent and dichotomous split in the road between our technological future and our environmental one. AI needs energy—and its growing needs are, to some extent, being met through coal-fired plants and fossil fuels that are accelerating already dangerous trends in global climate change. An AI future would seem to be one that comes at the expense of our environment. [caption id="attachment_228781" align="alignleft" width="216"]
Chris Yoko[/caption] Certainly, reliance on legacy energy sources could be the path the new Trump administration takes the country down. His nominee for Secretary of Energy, Chris Wright, who was confirmed on Feb. 3, seems to be willing to embrace all forms of energy production, including natural gas as well as nuclear. In order to maintain the nation's competitive advantage in the AI race, the new Energy Secretary could fire up every coal plant in the country, especially if he's stymied in his attempt to embrace nuclear power. The issue: it still wouldn’t be enough. Simply training these AI models consumes enormous amounts of energy; one estimate states that the energy required to train GPT-3 is the same as is consumed annually by 130 U.S. homes. And that’s not even the latest model. Getting to the next stage of AI development—AGI or ASI—is going to require more energy than the U.S. currently has access to. The Biden administration's recent executive order on AI energy production highlights the extent of the issues facing scalable AI infrastructure. The standard renewable energy forms can't measure up. Even coated in solar panels, the world wouldn't be able to provide enough solar energy to satiate the eventual demands of growing AI. It's why nuclear is being looked at as the obvious next step of the energy evolution—and in many ways, it should be the consensus choice for America. The capitalists get their energy, and the environmentalists get a clean source. This emerging dynamic is every free-market politician’s dream: the market is, at long last, forcing the energy transition. But nuclear-curious companies can only get so far on their own. As the Trump administration takes power, it's increasingly clear that building nuclear energy capabilities is going to be a matter of national security, as well as competitiveness, over the next 10 years. China has ramped up its investment in nuclear power, and more nations are likely to follow suit, as the entire world wakes up to the clear energy demands of AI. Right now, Big Tech is leading the way; they're in an arms race with one another, and their enormous size and deep pockets mean they're the only ones with the necessary capital to make the switch on their own. The last year has been littered with headlines about Big Tech companies investing in and buying out contracts with nuclear energy providers. Microsoft wants to re-open Three Mile Island while Amazon and Google are investing in their own smaller reactors. Meta, meanwhile, has an active "request for proposals" open to source and produce its own nuclear energy. Nuclear is no longer strictly in the domain of government investment; much in the same way space exploration has been commercialized and expanded to the masses, so too is nuclear becoming a go-to solution for companies looking to power their AI innovations. Of course, that doesn't mean that government support wouldn’t be a relief. Especially as AI becomes more vital for everything from medicine to banking to the military, the government needs to have some chips on the table. This could be the big push that the environmental movement has been waiting for. With so much riding on the development of AI, the government can step in, backing up its commitment to Big Tech companies by supporting them in their nuclear energy efforts. The energy issue is, in many ways, a continuation of the landmark CHIPs act, which invested billions of dollars to help bring chip manufacturing into the U.S. Nuclear energy is part and parcel of that conversation. Having the U.S. government as a partner will be especially important as nuclear energy works to overcome perception issues and regulatory concerns. Microsoft's Three Mile Island push has certainly raised eyebrows from those that still remember the initial disaster, though safety of nuclear plants has increased significantly over the decades. There's also the issue of cost. In its attempts to reinvest in nuclear, the state of Georgia spent billions of dollars building two new plants, expanding Plant Vogtle at a cost that's certainly unsustainable as the world charges ahead. Of course, nuclear energy isn't renewable; eventually the Earth's stores of uranium will run out unless experts figure out a way to make nuclear energy more efficient—a goal that certainly seems possible with greater investment and discovery. Frankly, there aren't any other good options. With AI growing and expanding at its current rate, energy production simply isn't going to keep up. More effective energy sources are vital, and balancing energy investment with Big Tech's AI ambitions is going to be a key issue for the next presidential administration. Nuclear energy is a critical part of the AI future—one that we have the power to lead. —Chris Yoko is the Founder/CEO of Yoko Consulting and Carbon Off.