Mar 12, 2024 - Technology

Startup taps AI to bolster nuclear power development

Illustration of a computer screen with a pixelated atom on it

Illustration: Sarah Grillo/Axios

A new startup called Atomic Canyon is harnessing artificial intelligence to help nuclear developers navigate complex federal licensing processes.

Why it matters: The firm's emergence from stealth comes as developers — including small modular reactor companies — will increasingly seek Nuclear Regulatory Commission approvals.

The big picture: Their tech "uses AI-powered search, trained on all 52 million pages from the NRC's Agency-wide Documents Access and Management System (ADAMS) database," the California company said.

Quick take: AI plays an emerging role in energy transition.

  • Backers see uses like permitting help and developing battery materials, to name just two.
  • Yet generative AI sucks up lots of power, which complicates decarbonization.

The intrigue: A WSJ piece on Microsoft's AI work explores this dynamic.

  • The tech giant is exploring nuclear energy to power data centers and is toying with tech "to see if AI could help streamline the approval process," the report notes.

State of play: Atomic says it's self-funded for now, via prior exits from founders Trey Lauderdale and Kristian Kielhofner.

  • Current users include power plants, SMR manufacturers and research groups, they said.

Look no further than the EQT-Equitrans merger for evidence that AI computing power discussions are cropping up everywhere — with efforts to capitalize on them in tow.

What they're saying: EQT CEO Toby Rice cited AI when touting the market for Equitrans Mountain Valley Pipeline (MVP) project, which will provide natural gas to markets in the Southeast.

  • On a call with analysts, Rice cited power generation markets and added, "throw on top of that, the power demand from AI...MVP is servicing data center alley, and it's going to create even more opportunity," Bloomberg reports.

The bottom line: AI's climate role is an emerging storyline.

Go deeper