The artificial intelligence boom has an inconvenient secret: it is hungry, and it is getting hungrier. Every chatbot query, every generated image, every model training run draws power from a grid already straining under the weight of surging demand. As data center operators scramble to secure electricity, a growing cohort of startups is asking a deceptively simple question — is there a better way to make power? One small California-based company believes the answer may lie in ceramics.
Petra Power, founded in 2017, is betting that solid oxide fuel cells — compact ceramic devices that convert fuels such as natural gas directly into electricity without combustion — can transform how data centers and even military vehicles consume energy. The company's pitch is bold: lower fuel costs, fewer emissions, and a dramatically smaller physical footprint than conventional generation.
To understand Petra's proposition, it helps to understand how most electricity is made in the first place. Traditional power generation, whether in a gas plant or a vehicle engine, relies on burning fuel. Aaron Goodman, Petra's founder, describes the process as an ugly chain of conversions.
"Combustion takes fuel and creates explosions," Goodman said. "Those explosions create heat. That heat creates motion. The motion powers an electromotive force, which eventually creates electricity. It's a lot of steps, and they all have so much loss in them that you end up with a really inefficient system."
Every stage of that chain — heat, motion, electromagnetic induction — leaks energy. By the time electricity reaches its destination, much of the original fuel's potential has been lost along the way.
Fuel cells take a fundamentally different route. Rather than burning fuel to create mechanical motion, solid oxide fuel cells perform an electrochemical reaction, extracting electrons directly from the fuel itself.
"What fuel cells do is they take the electrons directly off of the fuel, so they strip the fuel of its electrons, which creates electricity," Goodman explained. "It's one step, no loss, and subsequently, in theory, at least, they're much, much more efficient."
That theoretical efficiency translates, in practice, into lower fuel bills and reduced emissions — two qualities that carry enormous weight in an industry where energy is rapidly becoming the dominant operating expense.
Petra's near-term focus is data centers, the sprawling facilities that house the servers behind AI and cloud computing. The company is not yet working with the hyperscale giants — the Amazons, Microsofts, and Googles of the world — but with so-called neoclouds and smaller infrastructure providers a few steps down the ladder. Goodman declined to disclose specific agreements but said first deployments with these customers are targeted for 2028, with full-scale production hoped for 2029.
"Hyperscalers are an ICP [ideal customer profile] for us, and we talk a lot to them, but we don't have anything firm with them yet," he said. "They're the largest consumers of power. They're the ones that we can make the biggest impact with, so that's where we'd like to go."
The timing is not accidental. Utilities across the United States have warned that data center demand is outpacing grid expansion, with some regions imposing moratoriums on new interconnections. Tech companies have responded by exploring everything from small modular nuclear reactors to geothermal and on-site gas generation. Fuel cells, which can be installed where the power is needed, offer a form of energy independence that grid-tied operators increasingly covet.
A Second Front: The Military
Petra's other target customer is the United States Department of Defense, and here the use case looks quite different. The company envisions its fuel cells providing auxiliary power for land vehicles — supplying electricity for onboard equipment such as sensors, communications gear, and climate systems while the main engine is switched off. Silent, low-heat auxiliary power can reduce a vehicle's fuel consumption and its thermal and acoustic signature, qualities that matter on the modern battlefield.
The technology has not yet been deployed on a live vehicle. Goodman clarified that the fuel cells are currently in a testing phase, with the government weighing whether to proceed. Even so, the relationship is substantial: Petra has received nearly $9 million in contracts from the Defense Department to date, and Goodman hopes to scale that work significantly in the coming years.
Small Team, High Stakes
Petra remains a lean operation — roughly 15 people — which underscores both the promise and the peril of its position. It is a small company with an emerging customer base, competing in an arena where well-funded rivals and established energy giants are also circling. Fuel cell technology itself has endured cycles of hype and disappointment over the past two decades, and the industry's credibility rests on delivering real-world performance at competitive cost.
Yet the macroeconomic forces behind Petra's thesis are undeniable. AI's energy appetite shows no sign of abating, grid constraints are tightening, and pressure to reduce emissions — even for gas-based power — continues to mount. If solid oxide fuel cells can deliver on their efficiency promise at scale, they could carve out a meaningful niche in a market worth tens of billions of dollars.
For now, Petra's roadmap is measured: prove the technology with smaller cloud operators and government testers by 2028, reach full-scale production by 2029, and keep knocking on the doors of the hyperscalers. Whether a 15-person team with ceramic cells can help power the age of AI remains to be seen — but the industry's desperate search for new electrons means companies like Petra have never had a more attentive audience.
