In the high desert outside Reno, Nevada, rows of solar panels track the sun, feeding clean energy into a novel storage system. Beneath white plastic covers sit stacks of used electric vehicle batteries, repurposed to keep two Crusoe AI cloud data centers operating continuously. The setup reflects a growing urgency in the technology sector: securing reliable, scalable power as artificial intelligence workloads surge.
“The whole AI industry is struggling with how to access more reliable power on a fast time scale,” said Cully Cavness, Crusoe’s co-founder, president, and chief operating officer. That urgency is underscored by a 2024 U.S. Department of Energy study projecting that data centers could consume up to 12% of national electricity by 2028—triple their current share. In Texas alone, developers have filed plans for more than 100 new gas-fired plants, largely to serve server farms.
Cavness acknowledged the appeal of gas generation for its rapid deployment. “Gas power is a great solution that data centers are turning to for speed, for speed to megawatts. This is an alternative way to accomplish that speed, but with a renewable power source,” he said, pointing to the potential of recycled EV batteries to deliver both responsiveness and sustainability.
The batteries in Crusoe’s installation originate from Redwood Materials, a Nevada-based recycler founded in 2017 by JB Straubel, former chief technology officer at Tesla. In 2024, Redwood recovered more than 20 gigawatt-hours of lithium-ion cells—enough to equip roughly 250,000 new electric vehicles. General Motors recently announced a partnership to supply Redwood with end-of-life EV packs, strengthening the supply chain for such second-life applications.
Colin Campbell, Redwood’s chief technology officer, emphasized the residual value in these components. “There’s really very little wrong with them,” he said. “Like, maybe they have lost 20% of their capacity. Maybe your electric vehicle’s a little bit slower. And so that’s appropriate that you don’t want it in your car anymore. But it still works great. So we just looked at that and we were like, ‘Hey, why don’t we use it to store energy for the grid?'”
From an engineering standpoint, Redwood’s approach favors simplicity. The storage arrays require no pipes or pumps, reducing mechanical complexity and installation time. “That’s been a fun engineering effort for the engineering team to make something that is robust, that is incredibly useful, but also very cheap and quick to put in the field,” Campbell explained. Such design choices not only cut costs but also improve reliability in remote environments.
Straubel sees vast potential for replication. “There’s no practical limit that we see on how we can scale this,” he noted. “It’s one thing that makes us so excited. This is very modular.” Modularity allows identical units to be deployed rapidly across diverse sites, from rural solar farms to urban microgrids, without extensive redesign.
Beyond technical efficiency, Straubel frames the initiative as a way to avoid the environmental downsides of conventional AI power sourcing. “This is a different way to power the AI revolution,” he said. “We’re showing, you know, that AI doesn’t have to be in conflict with the existing grid.” By decoupling data center growth from fossil-fuel expansion, second-life battery systems could reduce strain on transmission infrastructure and cut carbon emissions.
The Reno project exemplifies how aerospace-grade thinking—modular design, resource reuse, and robust field performance—can migrate into the digital infrastructure space. For engineers and enthusiasts, it offers a tangible case study in bridging clean energy generation with high-demand computing. As AI workloads intensify, such solutions may define the next frontier in sustainable power engineering.