Mazama’s $135M Bet Is Really About AI’s Power Problem

AI companies have spent billions trying to secure more chips. Increasingly, the harder thing to find is enough electricity to run them.

Geothermal startup Mazama Energy has raised $135 million in an oversubscribed Series B to develop technology that drills into extremely hot rock deep underground.

The round was led by Centaurus Capital and Doerr Capital, with investors including ConocoPhillips, Shell Ventures, Khosla Ventures and Gates Frontier.

The startup is chasing an unusually ambitious idea:

Use superhot geothermal energy to produce enormous amounts of round-the-clock electricity from a relatively small number of wells.

AI data centers may become one of its most attractive customers.

The geothermal playbook is changing

Traditional geothermal power works best in places where heat, water and underground geology naturally line up.

That limits where projects can be built.

Enhanced geothermal companies take a different approach.

They drill deeper and create engineered systems capable of extracting heat from rock that would otherwise be inaccessible.

Mazama is going further still.

The company has drilled beyond 10,000 feet and is targeting depths of roughly 15,000 feet, where temperatures can reach about 750°F, or 400°C.

At those temperatures and pressures, water can reach a supercritical state that carries substantially more energy than ordinary steam.

Mazama says this could allow each well to produce as much as 10 times the power of conventional geothermal technology.

One well could generate 15 megawatts

The startup says its horizontal-well design could generate around 15 megawatts per well.

Its first Oregon site is now estimated to have the potential for 10 gigawatts of power, twice the 5-gigawatt estimate associated with the project last year.

Mazama plans to begin producing electricity next year and aims for around 200 megawatts of generation from its first development by 2030.

Those numbers matter because the economics of AI infrastructure increasingly depend on access to huge blocks of electricity.

A hyperscale data center can consume as much power as a small city.

And unlike some traditional industrial demand, AI infrastructure is growing extremely quickly.

AI has turned energy startups into tech infrastructure

For years, the AI funding conversation was dominated by software and semiconductors.

Now venture investors increasingly treat energy as part of the AI stack.

Chips need power.

Power systems need transmission.

Data centers need cooling.

The grid needs new capacity.

That has opened the door for nuclear, geothermal, energy-storage and grid-software startups to raise rounds that would once have looked unusually large for climate technology.

Mazama's $135 million Series B fits directly into that shift.

The startup is nominally an energy company.

Its growth opportunity is increasingly tied to technology companies.

Geothermal has one important advantage

Solar and wind power can be built relatively quickly, but their production changes with weather and time of day.

Data centers prefer consistent electricity.

Geothermal can theoretically operate around the clock.

That makes it attractive for AI workloads requiring continuous power.

Superhot geothermal could make the economics even more compelling by generating substantially more electricity from each drilled well.

That is the bet.

The engineering challenge remains significant.

Drilling deep into extremely hot rock puts equipment under intense temperature and pressure. Costs need to fall enough for projects to compete with established power sources.

The global resource is enormous

Research cited by TechCrunch estimates that accessing just 1% of the world's superhot-rock resources could theoretically produce more than 63 terawatts of electricity.

That is a resource estimate, not a prediction of economically recoverable power.

But it helps explain why investors are interested.

If new drilling technology makes even a fraction of those resources commercially viable, geothermal could expand far beyond the relatively limited footprint it has today.

The funding race around AI is moving underground

There is a broader startup lesson here.

Whenever a technology boom hits a physical constraint, investors start looking for companies that can remove it.

For AI, the first major bottleneck was compute.

Then networking.

Then data-center construction.

Electricity is now becoming equally important.

That means the AI boom can create billion-dollar opportunities in industries that do not look like AI companies at all.

What happens next?

Mazama still has to prove that superhot geothermal can move from successful drilling to reliable commercial electricity.

That will matter more than the size of its latest funding round.

But the capital flowing into the company is another sign that venture investors believe AI's power problem will create an entirely new infrastructure market.

The biggest beneficiaries of the AI boom may not all build models.

Some may simply build the power plants that keep those models running.