Voxel Energy Powers AI Data Centers With Solar and Second-Life EV Batteries

Voxel Energy, a Y Combinator-backed startup in the Winter 2026 batch, is taking aim at one of the most urgent bottlenecks in the AI boom: getting electricity to data centers fast enough to keep up with compute demand. Co-founded by Casey Spencer, Max Pfeiffer, and Evan Schmidt, Voxel is building off-grid data center power systems that let operators bypass utility connection queues entirely – pairing on-site solar with a novel DC microgrid architecture built around repurposed EV batteries. The company’s founding premise is captured in a single line on its website: the best grid connection is no grid connection.

The energy problem in AI infrastructure has become impossible to ignore. Hyperscale and colocation operators are signing power purchase agreements years in advance, and interconnection queues for new utility-scale loads routinely stretch from 24 months to more than five years. Meanwhile, model training and inference workloads keep growing, and new AI clusters are being sited in markets where the local grid was never planned for gigawatt-class loads. The result is that the rate-limiting factor for AI capacity is no longer chips, capital, or real estate – it is kilowatts delivered on time. Voxel’s thesis is that the fastest path to new capacity is off-grid data center power – stop waiting for the utility and start generating and storing electricity on the data center’s own footprint.

Voxel’s system pairs on-site solar generation with a DC microgrid architecture designed around second-life EV batteries. The world now has an immense and rapidly growing supply of used electric vehicle batteries coming off the road with 70-80% of their original capacity still intact — more than enough for stationary storage applications, and often an order of magnitude cheaper than newly manufactured cells. By building its own vertically integrated stack rather than stitching together off-the-shelf commercial microgrid products, Voxel can squeeze more round-trip efficiency, higher density, and lower cost per kilowatt-hour out of those batteries, and can deliver a system engineered specifically for the always-on demands of AI workloads.

The founding team’s pedigree is notable. Casey Spencer is an award-winning engineer who previously served as Project Manager on Tesla’s Infotainment and Autopilot team, where he oversaw distribution of prototype hardware and played a key role during Tesla’s famously intense “production hell” period in 2018, leading teams that built ad-hoc assembly lines and then worked on them. That background is precisely the kind required to industrialize a novel energy stack. Co-founders Max Pfeiffer and Evan Schmidt bring complementary engineering and operating depth, giving Voxel an unusually strong operational backbone for a company building hardware, software, and physical infrastructure at once.

The market opportunity is enormous, if also notoriously hard to execute on. Global data center power demand is projected to more than double over the coming decade, largely driven by AI, and analysts routinely cite tens of gigawatts of incremental load that will need to come online in the United States alone. Utilities, for structural and regulatory reasons, cannot keep pace. Behind-the-meter generation and storage – especially approaches that make creative use of second-life batteries and pair them with solar – represent one of the most credible paths to closing that gap while also improving the environmental profile of AI infrastructure.