This is one outlet's own report from Forbes — the article as it was filed.
AIPROPX ReportForbes · 3h ago
Data Centers Need Power Fast. The Auto Industry Already Has It.
Energy Data Centers Need Power Fast. The Auto Industry Already Has It. By Anna Demeo ,
Forbes contributors publish independent expert analyses and insights. Anna Demeo is an executive and advisor on energy infrastructure and AI Follow Author Jul 30, 2026, 06:56pm EDT --:-- / --:-- This voice experience is generated by AI. Learn more . This voice experience is generated by AI. Learn more . Summary Hyperscalers face a critical "speed to power" challenge, as surging AI demand outpaces the slow timelines of traditional utility infrastructure for data centers. While some explore exotic solutions like fusion or space-based data centers, a more immediate answer may lie in electric vehicles. Vehicle-to-Grid (V2G) technology allows parked EVs to discharge stored energy back into the grid, effectively turning millions of cars into a vast, distributed power plant. These aggregated EV fleets could provide significant, dispatchable power, utilizing abundant off-peak generation. Despite concerns about mobility, EVs are mostly idle, making their contribution predictable. V2G offers a practical, near-term solution to meet escalating data center electricity needs faster than conventional methods.
Access to reliable electricity constrains hyperscalers trying to build data centers. Speed to power, the phrase coined to capture this challenge, drives the industry to leave no stone unturned in its quest for reliable power to keep up with AI demand. In this pursuit, hyperscalers took a crash course in the utility sector, known for slow timelines and heavy regulation. Ironically, the auto industry may be what delivers speed to power in the form of the millions of vehicles people drive every day.
U.S. data center power demand could reach 176 gigawatts by 2035 , five times current usage. Delivering that much power demands a massive lift from a utility industry that has seen little growth in decades and runs on infrastructure well past its intended lifespan.
For data center developers, securing enough power poses one challenge. Securing it fast enough to keep pace with AI poses another entirely. When hyperscalers first engaged with utilities at scale, the differences between the industries became immediately apparent. Companies like Google and Meta, used to solving problems by throwing money and engineering at them, ran into a regulated system built around ratepayer impact studies and multiyear reliability reviews that does not move faster no matter how much capital shows up. Utility power delivery timelines running seven to nine years are a nonstarter for companies trying to bring new data centers online within a year or two. The result: the data center industry started investigating and investing in every alternative available.
Some of those efforts verge on the extraordinary. Google signed an offtake agreement for fusion energy, a technology still years from commercial viability and unlikely to solve the speed problem regardless of its long-term promise. Nvidia's venture arm invested $650 million in TerraPower's Natrium reactor technology. A few are betting on leaving the grid behind entirely: SpaceX and xAI have filed with federal regulators for a constellation of AI data center satellites , chasing unlimited solar power in orbit rather than waiting on power lines on Earth. In an "all of the above" strategy, finding a solution that is less exotic, more proven, and more near term may require looking no further than what already exists and is parked in a driveway.
While hyperscalers test unconventional fixes, electric vehicles could offer a more immediate one: a vast, largely untapped pool of storage already spread across school buses, delivery fleets, and personal cars. Vehicle-to-grid technology, known as V2G, lets those batteries discharge stored power back into the grid, turning millions of parked vehicles into a distributed power plant. Aggregated at scale, EV fleets could rival the output of traditional utility-scale plants, delivering the kind of dispatchable power hyperscalers need.
Car batteries dwarf residential batteries such as the Tesla Powerwall. A Tesla Model
Y carries roughly 78 kilowatt-hours of battery capacity, nearly six times the Powerwall's 13.5 kilowatt-hours. The potential scale is enormous. A 2025 Union of Concerned Scientists and Evolved Energy Research study found that 2 million EVs enrolled in vehicle-grid-integration programs, a realistic share of eligible vehicles, could supply a cumulative 17.2 gigawatts to California's grid alone.
Cars, unlike stationary batteries, can drive away, raising questions about reliability. But Department of Energy data shows household vehicles sit parked about 95% of the time, driven an average of roughly 65 minutes a day, leaving most of a car's battery doing nothing for the vast majority of its life.
With a large enough fleet, that idle time becomes predictable rather than random. Utilities already rely on the same logic with aggregated pools of controlled water heaters that deliver hundreds of megawatts of curtailable power today. With the right program structure and the technology-driven intelligence to manage it, a fleet of V2G-enabled vehicles can commit that same kind of firm, modeled capacity, even as individual cars come and go.
Battery storage can shift and shape demand peaks but it still needs electricity to charge. Off-peak hours offer plenty of it: excess solar and wind go to waste at scale, with curtailment running roughly 20 million megawatt-hours a year across the U.S. In Texas, ERCOT curtailed more than 8 terawatt-hours of wind and solar in 2024. Traditional power plants can provide even more, with off-peak fossil generation alone offering 500 to 700 gigawatts of available capacity. Intelligent charging paired with optimized EV battery dispatch turns that idle supply into distributed power for the grid, smoothing out the peaks that strain it most.
Data centers will define one of the century's biggest infrastructure challe...
AIPROPX is an independent multi-source news index — we track, compare, and connect coverage from across the web into one place you won't find anywhere else.
