Tech Titans Turn To Nuclear Energy To Try And Meet AI's Insatiable Power Hunger

Big tech is going nuclear. In just the past two months, industry giants Google, Microsoft, OpenAI, and Amazon have signed up for advanced nuclear reactors to power their insatiable artificial intelligence (AI) developments.

On 14 October, Google made headlines by signing the world's first corporate agreement to purchase nuclear energy from a consortium of small modular reactors (SMRs), which will be built by a startup called Kairos Power.

Just two days later, Amazon unveiled its ambitious plan, partnering with Energy Northwest to develop four advanced SMRs. The company also picked up a strategic stake in X-energy, a leading SMR and fuel developer.

In September, Microsoft solidified its commitment to nuclear energy by signing a 20-year power purchase agreement with Constellation, which will facilitate the restart of the Three Mile Island Unit 1 in Pennsylvania — a facility that has remained dormant since 2019.

Furthermore, Sam Altman, OpenAI's chief executive officer (CEO), has invested in Oklo, a nuclear startup aiming to construct a commercial microreactor in Idaho by 2027.

Altman’s investment in the nuclear fusion company Helion in 2021 further underscores a broader commitment to harnessing advanced nuclear technologies.

This string of announcements raises critical questions about the future of energy consumption in a world increasingly driven by data. To truly grasp this nuclear push, we must first explore the economics of SMRs.

The Promise Of SMRs

SMRs are an advanced kind of nuclear reactor, with a power capacity of 300 MW or less — about one-third that of standard plants. Their smaller physical footprint allows them to be built closer to the grid, and their accelerated construction timelines promise to bring them online far more swiftly than conventional reactors.

Economically, SMRs are more cost-effective to build, making them an attractive option compared to the bespoke, large-scale plants notorious for budget excesses and missed deadlines.

However, it’s important to note that this technology is still in its early stages of development. Currently, no SMRs are operational in the United States (US), with only NuScale Power’s design having received regulatory approval to date.

The rigorous safety standards in the nuclear sector pose formidable challenges for gaining regulatory approvals.

The tech industry’s dash for nuclear energy is, in part, fuelled by the explosive growth of AI. For instance, a single ChatGPT query requires 2.9 watt-hours of electricity, in stark contrast to just 0.3 watt-hours for a typical Google search, according to the International Energy Agency (IEA).

In that difference lies the coming sea change in how the US, Europe, and the world at large will consume power — and how much that will cost.

Goldman Sachs Research projects that power demand from data centres will grow 160 per cent by 2030, with AI projected to account for 19 per cent of data centre power demand by 2028.

The anticipated rise in data centre energy consumption from AI alone could reach an astonishing 200 terawatt-hours per year between 2023 and 2030.

Geographical concentration plays a vital role in this narrative. Regions like Northern Virginia in the US, famously dubbed the "Data Center Alley," as it's home to the largest number of data centres in the world, underscore the necessity for tech companies to secure reliable power sources for their operations.

Their net-zero pledges require the sources to be green, and they have already invested heavily in wind and solar. While diversifying into nuclear energy makes sense, it also carries inherent risks.

What Lies Ahead

Amazon and Microsoft have already struck deals to procure energy from legacy nuclear plants, while other tech giants are now stepping into building their own nuclear infrastructure, giving SMRs newfound momentum. These AI developers see these plants as scalable power sources capable of providing reliable energy around the clock.

As J Clay Sell, CEO of X-Energy, observes, Amazon’s recent agreement marks a shift in the tech industry’s historical reluctance to buy nuclear power unless someone else pays to build the infrastructure.

The willingness of tech companies to pay above-market prices for reliable, carbon-free energy could be the key for SMRs to establish themselves in the energy landscape, according to Jacopo Buongiorno, a professor at the Massachusetts Institute of Technology.

“You start by selling electricity to a customer that is willing to pay a little bit more,” he says. “Then as your own costs come down, you start penetrating markets where the competition is a little bit tougher — for example, the grid.”

The surge in AI-driven data demand means we are already in the new age of electricity. As we navigate this reality, we face a pivotal choice: will AI remain just another energy-hungry consumer, or will it evolve into a central player in our green energy future?