AI is giving us a nuclear renaissance: Can it help solve the climate crisis too?

Capturing this zeitgeist, US president Dwight D. Eisenhower delivered his famous ‘Atoms for Peace’ speech at the United Nations in 1953, promising nuclear power for clean and abundant energy rather than for death and destruction. The US was particularly excited, as it rolled out nuclear power plants and submarines and even considered nuclear aircraft.

But US euphoria gave way to panic when in March 1979 the reactor core of Three Mile Island Nuclear Generating Station at Pennsylvania suffered a partial meltdown. The ensuing alarm meant that this technology hit a big speed-bump. Nuclear tech never recovered.

Multiple reactors closed and the share of nuclear power shrank. Today, most plants in the US are 30-50 years old and the cost of setting up new ones is prohibitive, with a recent one costing $35 billion.

Also read: Nuclear energy’s AI boom blew a fuse—here’s what could happen next

Cut to 60 years later, however, and AI is the new nuclear. There is extraordinary excitement as trillion-dollar companies and hot startups race to build bigger and better AI models. But even as investment flows and enthusiasm reach a fever pitch, there are dark clouds ahead, especially around the unquenchable thirst for energy that AI models and data centres have.

For instance, the Electric Power Institute (EPI) expects data centres to consume up to 9% of US electricity generation by 2030, more than double what they currently use. This translates to 50 gigawatts (GW) of additional power needed by 2030 in the US alone, a colossal amount of energy which could power 10 New Yorks.

Most of the power generated across the world is with coal and other fossil fuels, making the sector one of the top contributors to CO2 emissions and global warming. Due to AI’s unprecedented need for energy, the global goal of reaching net zero emissions by 2050 will likely be missed, reports the EPI.

This harsh reality has set off a mad scramble for clean energy. While companies are scrounging around for wind and solar power, the most noticeable action is around nuclear. Amazon has struck three deals in this area, including the purchase of a $650 million nuclear-powered data centre outright.

Also read: Nuclear-powered AI: Big Tech’s bold solution or a pipedream?

Google signed a power purchase agreement with a California nuclear energy startup, Kairos Power. But the biggest eyeball-grabbing move was the one that took us back to the Three Mile Island reactor, with its owner Constellation Energy announcing last week that it will restart the nuclear reactor and Microsoft would buy the energy it generates for 20 years.

Microsoft has set stiff net-zero targets for itself, and has made other moves—it tied up with Brookfield Asset Management to generate 10.4GW of renewable energy capacity across the US and Europe. It has even invested in the distant hope of nuclear fusion energy, with an agreement with Helion Energy, a fusion startup.

Even in India, Microsoft has struck a clean energy deal with ReNew Power. Microsoft president Brad Smith has highlighted the potential of repurposing existing coal-fired power plants with advanced nuclear technologies, saying: “You can cut the cost of constructing a new nuclear plant by a third by converting these plants, and we can use the power of cloud computing, AI and data to accelerate all of that."

Founder Bill Gates is a big proponent too, having co-founded TerraPower, a company focused on developing advanced nuclear reactors, including designs that address traditional safety concerns around this source.

It is an interesting conundrum. While many experts believe that AI’s insatiable energy demands will accelerate climate change, optimists like Gates and others believe the reverse could be true. Big Tech now has an incentive to invest hugely in clean energy, led by nuclear power, and has an equally large need to get energy for its AI dream.

Also read: Inside the audacious plan to reopen Three Mile Island’s nuclear plant

Gates argues that Big Tech firms would pay a ‘green premium’ for clean energy, thus incentivizing its development and deployment over fossil fuel, and so the net result would be positive for the planet. AI techno-optimists believe that when artificial general intelligence (AGI) is achieved, it will solve big hairy problems like global warming for humanity. In a roundabout way, it may already have started solving our energy problem.