Meta Joins Big Tech Nuclear Push for AI Power, Following Google’s Atomic Energy Deal

Key Takeaways

• Meta has put out a call for proposals to develop new nuclear power stations.
• The firm is looking to partner with developers that can help it meet its soaring electricity requirements.
• AI processing has significantly increased data center power consumption.

Amid surging demand for electricity to fuel AI training and inference, data center operators are actively exploring nuclear power options.

Following Google’s recent partnership with atomic energy startup Kairos Power, Meta is the latest Big Tech firm to express an interest in building nuclear reactors.

Meta Seeking Nuclear Partners

On Tuesday, Dec. 3, Meta invited proposals to build new nuclear infrastructure projects that could power its data centers.

“We believe that nuclear energy can help provide firm, baseload power to support the growth needs of the electric grids that power both our data centers” and “the communities around them.”

The latest proposal represents Meta’s rising prevalence as a partner in grid development projects. Since 2020, the company says it has helped add 12,000 megawatts of wind, solar and geothermal energy to networks around the world.

However, the firm acknowledges that building nuclear power stations is a far more ambitious undertaking.

Compared to renewables, “nuclear energy projects are more capital intensive, take longer to develop, are subject to more regulatory requirements, and have a longer expected operational life,” it said.

As such, Meta is seeking a partner it can work with “across projects and locations” as it embarks on its nuclear journey.

The company said it is targeting one to four gigawatts of power and hopes to get generation up and running by the early 2030s. According to the Department of Energy, the average nuclear power station in the U.S. outputs 1 gigawatt.

In terms of its potential for generating electricity, the strategy is more ambitious than Google’s, which will generate 500 megawatts from Kairos’ small reactors.

Surging AI Electricity Demand

Meta’s interest in nuclear power reflects the AI industry’s soaring energy needs, driven by the higher electricity consumption of power-hungry GPUs.

As AI is integrated into everyday life, more and more of the world’s digital activity will require high-energy inference.

In 2023, Google CEO Sundar Pichai said that the average chatbot interaction costs the company “ten times more than a standard keyword search.” But that hasn’t prevented Google from doubling down on AI search.

As data centers have shifted to meet the sector’s requirements, their electricity bills have skyrocketed, and in some cases, so have consumers’ bills.

In Virginia, which is one of the world’s preeminent data center hubs, researchers estimate that unless action is taken now, energy prices could increase by as much as 70 percent over the next decade due to demand from data centers.

A Nuclear Renaissance

Having risen consistently for years, nuclear energy generation in the U.S. has plateaued since the mid-2000s.

However, more than a decade after the Fukushima meltdown, there are signs that the industry is picking up again.

In August, Donald Trump vowed to accelerate nuclear-powered electricity generation to help ensure the U.S. maintains its AI leadership.

“For AI,” he observed, “you’ll need to double the energy that we produce right now just for that one industry if we’re going to be the big player and dominant.”

If the U.S. doesn’t secure its leadership of the sector, “then China’s going to take over,” he warned.

Since winning the election, Trump’s anticipated appointment of Chris Wright , a board member at nuclear energy firm Oklo, as head of the Department of Energy has further fueled the bullish sentiment around the technology.

Meanwhile, advances in reactor technology proposed by companies like Kairos promise to drive down the cost of development with smaller plants that don’t require the giant cooling towers traditionally associated with nuclear power.