Xeal Seeks $2 Billion to Turn Underused EV Chargers into Power for AI Computing

EV charging startup Xeal is betting that idle charging stations can help meet surging power needs from AI computing. CoinDesk reported the New York-based company plans to raise about $2 billion in debt financing over the next 18 months to build and roll out modular GPU compute units powered by existing charger connections. Xeal operates thousands of chargers across roughly 500 cities in the U.S. and Canada, concentrated in upscale apartment buildings, high-end retail malls, and office properties. CEO Nihal Balladwaj said contracts with utilities allow 24/7 charging, yet average utilization across its network is under 10%, leaving substantial unused electrical capacity. Xeal estimates its idle power could continuously support about 130,000 households or roughly 110,000 NVIDIA AI chips. The company's new business line aims to sell that unused power to AI workloads by colocating compute next to charger sites. In December, Xeal plans to deploy NVIDIA AI server racks beside charging stations at parking facilities in Chicago, Houston, and Dallas. Each rack will be placed inside a custom-protected metal container about the size of a single parking space and will draw power directly from the existing charging-station electrical lines. Each container will hold about 48 NVIDIA AI chips. Balladur said Xeal has already secured orders for 100 compute containers but declined to name customers. Over the next 18 months, the company targets raising about $2 billion to manufacture and deploy 1,000 containers. The build-out would amount to about 50,000 GPUs in total and roughly 60 megawatts of combined power capacity. Xeal has previously raised $540 million in equity across two rounds, including a $400 million Series B in 2022. Balakrishnan said the company intends to pursue $2 billion in low-interest asset-backed loans and project financing to fund the expansion, arguing Xeal's ability to build a large charging network with $540 million in venture capital strengthens its case with lenders. The initiative highlights how companies outside the traditional AI sector are trying to monetize the electricity demand created by the data-center boom. Still, the publication noted fundraising has become more difficult for some AI infrastructure developers, and Xeal may not secure its full target even if early deployments perform well. Xeal's main edge is grid access. AI data-center developers often face multi-year waits for interconnection, pushing some to consider alternatives such as gas turbines or nuclear power, each with long permitting and construction timelines. Batteries can serve as backup but cannot support continuous operations and must be recharged after only a few hours of use. Baldevi said, "We can launch our reasoning and computing services in just a few weeks, rather than waiting years. Funding is the catalyst that helps us build a sustainable computing infrastructure." Balladur co-founded Xeal in 2019. The company sells chargers to property owners, who typically recover their investment within five years when EV usage is high. With U.S. EV sales falling short of expectations for reasons including vehicle costs, Xeal says its stations average a 92% idle rate, making many property owners hesitant to add more chargers. Baldevagi described the dynamic as a chicken-and-egg problem: "Property clients say, 'I don't know if there are EV owners in my community, so I'm unsure whether to install more chargers.'" Xeal says AI demand changes the economics. Ballardwaj is offering building owners site-rental payments to host a GPU container in parking garages served by Xeal chargers. He said the payback period improves from five years to about two and a half years, with no capital expenditure or operating costs for the property owner because Xeal covers electricity costs. He added that compute demand is strong enough to justify additional charger deployments. Xeal said its total power capacity across all sites is about 200 megawatts and is growing by roughly 10 megawatts per quarter. The company claims that by using existing infrastructure and monetizing power through computing, it can expand charging projects across the U.S. without federal subsidies or utility incentives, helping make the broader ecosystem self-sustaining.