Old Oil Wells to Power AI: Geothermal Tech Taps LA's Past for the Future

DENVER, CO – February 25, 2026 – In a move that fuses the legacy of the fossil fuel era with the demands of the digital age, a new partnership is set to transform dormant oil and gas infrastructure in the Los Angeles Basin into a power source for the burgeoning artificial intelligence industry.

Denver-based Gradient Geothermal announced it has signed a Memorandum of Understanding (MOU) with GEOT.Ai, an infrastructure firm incubated by Muir Global Holdings. The collaboration aims to develop dedicated geothermal power for so-called "AI factories," starting with a proposed pilot project in Southern California. This initiative seeks to solve one of the most pressing challenges of the 21st century: how to fuel the voracious energy appetite of AI without crippling power grids or compromising climate goals.

The AI Power Paradox

The rapid ascent of artificial intelligence has created an unprecedented demand for computational power, and by extension, electricity. The energy required for training cutting-edge AI models has been increasing by approximately five times every year since 2020. More than 30 AI models have already been trained at a scale exceeding 10²⁵ floating-point operations (FLOP), a measure of computational intensity. This hardware arms race, driven by a supply of high-performance chips that doubles roughly every 10 months, is putting immense pressure on traditional energy infrastructure.

AI compute clusters and the data centers that house them require a constant, uninterrupted flow of massive amounts of electricity—a stable "baseload" power that the existing grid, often strained and reliant on intermittent renewables, struggles to guarantee. The prospect of connecting entire fleets of new, power-hungry AI factories to this public infrastructure raises serious concerns about grid stability, cost, and environmental impact. This partnership aims to sidestep the problem entirely by building a self-contained, sustainable energy and computing ecosystem.

"This collaboration represents the convergence of geothermal energy and AI infrastructure," said Benjamin Burke, CEO of Gradient Geothermal, in the announcement. "We are building a resilient, secure, and net-zero foundation for next-generation AI Factories empower both on-premises power solutions and local power generation within an urban community."

From Black Gold to Green Kilowatts

The partnership's proposed solution is as innovative as it is pragmatic: tap into the heat beneath California's old oil fields. The Los Angeles Basin is home to numerous mature oil fields characterized by historically high volumes of hot water brought to the surface during extraction—a byproduct that is also a sign of significant geothermal potential at depth.

Instead of drilling new wells, which is costly, time-consuming, and carries exploration risk, Gradient Geothermal plans to repurpose existing, and often abandoned, wellbores. By employing its field-proven Organic Rankine Cycle (ORC) technology—a closed-loop system that uses heat from the underground water to vaporize a fluid, which then drives a turbine to generate electricity—the company can transform these brownfield sites into productive, emissions-free power plants.

This "low- to zero-drill" model dramatically accelerates deployment. The company claims it can reduce project timelines by up to 94%, bringing scalable, baseload power online in a matter of months, not the years typically required for new energy projects. The initial pilot project being evaluated by the two companies aims for up to 5 megawatts of geothermal power, enough to support a significant AI compute facility.

Furthermore, the model being advanced by GEOT.Ai addresses another critical resource constraint: water. Traditional data center cooling systems are notoriously water-intensive. GEOT.Ai's platform incorporates a proprietary cooling system that reportedly uses zero water, a crucial advantage in drought-prone regions like Southern California.

Redefining AI Infrastructure

The strategic vision extends beyond a simple power contract. The collaboration is built on GEOT.Ai's vertically integrated "Power-to-Compute" model, which co-locates energy generation and AI manufacturing in a single, behind-the-meter facility. This creates a decentralized and sovereign infrastructure stack, independent of grid failures or price volatility.

By anchoring energy-intensive operations directly to a reliable, 24/7 power source, the model offers the predictability and stability that AI development demands. This approach stands in contrast to other green solutions for data centers, which often rely on purchasing renewable energy credits or pairing intermittent sources like solar and wind with large, expensive battery storage systems.

While geothermal energy provides the constant power, the repurposing of industrial land offers a path for economic and environmental regeneration within urban areas. It presents a potential playbook for transforming thousands of legacy oil and gas sites across the country from environmental liabilities into assets for the new tech economy.

Jenifer Muir, Founder & CEO of Muir Global Holdings and GEOT.Ai, framed the initiative as a fundamental shift in thinking. "By anchoring AI factories directly to sovereign, regenerative energy," she stated, "we are redefining the infrastructure of AI."

As the first pilot project moves forward for evaluation in Los Angeles, the energy and tech industries will be watching closely. The success of this venture could prove that the solution to powering the future of technology may lie buried in the remnants of its industrial past.