Geothermal startup Mazama Energy raised $135 million in an oversubscribed Series B, the startup said Thursday.

The fresh funding will help the company develop horizontal wells in super-hot rocks capable of generating 15 megawatts of electricity per well.

Mazama plans to start generating electricity sometime next year and said its first site in Oregon is capable of producing 10 gigawatts of electricity. That’s a sharp upward revision from the company’s estimate last year, when investor Vinod Khosla said the site had the potential to produce 5 gigawatts.

Enhanced geothermal startups are shaping up to be the dark horse in the battle to power AI data centers and other large loads on the grid. While tech companies and data center developers have been throwing their weight behind natural gas, geothermal startups have been quickly advancing their technology.

Unlike previous geothermal projects, enhanced geothermal startups are targeting deeper and hotter rock. Mazama, for example, has drilled past 10,000 feet in 15 days on its way to about 15,000 feet deep. At those depths, the company can access 750˚ F (400˚ C) heat, which at high pressures turns water into a “supercritical” fluid that’s neither fully liquid nor fully gas — a state that lets it absorb far more energy than ordinary steam. Mazama said that will allow each well to produce up to 10 times more power than traditional geothermal technologies.

Given the depths Mazama is targeting, the potential payoff is significant. Tapping just 1% of super-hot rock worldwide could generate more than 63 terawatts of electricity, according to the University of Twente and the Clean Air Task Force.

While work at its Oregon site is still in the early stages, the startup has already secured land for a second development. Mazama aims to complete development of the first site by 2030, at which point it says it should generate 200 megawatts of electricity.

The new round was led by Centaurus Capital and Doerr Capital, with participation from ConocoPhillips and Shell Ventures. Existing investors including Khosla Ventures and Gates Frontier returned, and new investors SiteGround Capital, H. Barton Asset Management, and the Jeffrey and Marieke Rothschild Foundation also joined the round.

Mazama was incubated at Khosla Ventures.

Google and Google DeepMind researchers launched the DeepMind Institute on Wednesday to advance the conversation around artificial general intelligence. The institute lists DeepMind co-founder Shane Legg, Google executive James Manyika, and Google DeepMind chair Demis Hassabis as directors, with Legg serving as managing editor. 

The new institute aims to surface differing views between Google, Google DeepMind, and the broader global research community around AGI. “They will not always agree, and they will likely change their minds, as more data and information comes to light at the fast-moving frontier,” the announcement read.

The inaugural collection of four essays covers a range of topics: economic policies for managing potential AGI disruption, preserving human-readable model reasoning, principles for human flourishing, and a framework for evaluating frontier AI models. 

One essay, by DeepMind safety researchers Rohin Shah and Anca Dragan, argues that AI’s shrinking window of transparency—the ability to see and check a model’s step-by-step reasoning — is not inevitable. As new architectures make the most powerful models harder to monitor, the authors say developers and regulators should confront the safety trade-offs directly. That could mean limiting “opaque serial depth”—the amount of sequential computation a model can perform without producing a readable reasoning trace—or requiring developers to demonstrate that less transparent systems remain just as monitorable.

In another essay, Hassabis proposes a U.S.-led frontier AI standards body to evaluate the most advanced AI models. Under his framework, developers would initially submit models voluntarily for review up to 30 days before release. Once the evaluation system has proved effective, passing its tests could become a requirement for deploying frontier models in the United States.

The body would at first design assessments in consultation with AI companies, but would eventually develop independent, undisclosed evaluations — what the essay calls “held-out” tests —to prevent labs from tailoring their models to known evaluations. Hassabis said the framework could be “ratcheted up if the seriousness of the situation demands,” potentially including a coordinated slowdown among frontier AI developers. 

The essays arrive as the industry’s safety debate shifts from broad statements of concern toward concrete proposals for disclosure, outside scrutiny and, if safeguards fall behind, coordinated slowdowns. That shift accelerated this week as industry leaders endorsed elements of Anthropic CEO Dario Amodei’s call to “pace” frontier AI development.

A regulatory cap limiting Zoox to 100 robotaxis in Nevada is set to expire this month, clearing the way for the Amazon-owned company to expand its commercial fleet of custom-built robotaxis in Las Vegas just as competition heats up.

The Nevada Transportation Authority permit that imposed the 100-vehicle cap is set to expire September 25, TechCrunch has learned. An NTA official and a Zoox spokesperson confirmed the change, which is also outlined in an updated permit.

This doesn’t mean Las Vegas is about to be inundated with hundreds of Zoox robotaxis. But it does give Zoox more freedom to grow — and at a critical moment. Waymo recently launched a commercial service in Las Vegas, while Tesla and Uber have secured their own driverless ride-hailing permits.

Today, Zoox has about 100 custom-built robotaxis — cube-like vehicles that lack steering wheels and pedals — spread across four U.S. cities, including Austin and Miami. The bulk of the fleet is in Las Vegas and San Francisco. And for now, Las Vegas remains the only city where Zoox is charging for rides. Zoox must still obtain another permit from California regulators before it can charge for passengers there.

Zoox began charging for robotaxi rides in Las Vegas last month after receiving a temporary exemption from certain federal motor vehicle safety standards from the National Highway Traffic Safety Administration (NHTSA).

That exemption was crucial for Zoox because unlike Waymo, its robotaxis were designed without a steering wheel, pedals or other traditional, and federally mandated controls. The commercial exemption, allows Zoox to deploy up to 2,500 vehicles per year for the next two years.

It’s unclear how quickly Zoox will take advantage of its newfound freedom in Nevada. A Zoox spokesperson declined to say how many robotaxis the company plans to add to its Las Vegas fleet or even how many are in operation there today. Robotaxi Tracker, which collects data on autonomous vehicles, estimates Zoox has more than 30 vehicles operating in the city.

The spokesperson told TechCrunch that Zoox plans to steadily increase the size of its robotaxi fleet over the next several months “to meet the high demand in Las Vegas and other markets.” 

Zoox will have lots of company, no matter how fast it expands. In August, the NTA approved three permits that will allow Tesla, Uber, and Waymo to operate commercial robotaxi services in Clark County, home to Las Vegas. Together, the permits could bring up to 7,000 robotaxis to the area over the next 12 months.

Waymo has wasted little time acting on that permit. The company opened its robotaxi service to customers in Las Vegas earlier this month. Waymo told TechCrunch at the time that it would start with “dozens” of its new Ojai minivan robotaxis in a 23-square-mile area that includes the Strip south of Highway 589 and into Boulder Junction.