techcrunch.com web signal

Bedrock Energy Raises $24M to Scale AI-Guided Geothermal Drilling

TL;DR

  • Bedrock Energy closed a $24 million Series B led by Clean Energy Ventures and Clearvision Ventures to scale AI-guided geothermal drilling.
  • Co-founder and CEO Joselyn Lai says the company's AI-directed rigs can cut drilling times by up to 80% versus conventional methods.
  • Bedrock is pitching data centers on installing geothermal heat pumps at neighboring buildings to free up constrained grid capacity.

Bedrock Energy, a startup drilling shallow geothermal boreholes for commercial buildings, has closed a $24 million Series B led by Clean Energy Ventures and Clearvision Ventures, TechCrunch reported. The round brought in a long list of climate-oriented backers, including GVP Climate, the Autodesk Foundation, Energy Impact Partners, Titanium Ventures, Wireframe Ventures, Elemental Impact, and Catalytic Capital for Climate and Health by Temasek Trust.

The pitch is an oilfield trick aimed at a new problem. Bedrock uses directional drilling techniques borrowed from the oil and gas industry, guiding them with its own models of the subsurface, to install heat pump wells up to 1,000 feet deep. The company's claim is that "AI-directed rigs can slash drilling times by up to 80%, significantly reducing costs."

Where this intersects AI is the grid. Co-founder and CEO Joselyn Lai told TechCrunch some data center operators are drawn to geothermal's flat load profile because installing it on neighboring buildings opens up capacity on constrained grids. "We had someone in Michigan say, if we could free up a few megawatts, I think we might be able to operate that many months or even a year earlier," Lai said. "A year of revenue for a data center is very meaningful."

Bedrock now operates in Utah, Colorado, New York, Alberta, and Texas, with planned expansion into New England, New Jersey, the upper Midwest, and additional Western states. Prior raises were a $12 million Series A in 2025 and an $8.5 million seed in 2023.

The 80% figure is the company's own, and the article does not include a third-party benchmark. The Michigan customer is also unnamed. It lands in a stretch when the grid math around AI buildouts keeps surfacing: a recent Harvard study pegged US hyperscale data center consumption at 68 to 99 TWh.