Nvidia Corp (NASDAQ: NVDA) announced a partnership with six major U.S. energy companies to build a new class of AI data centers designed to tap up to 100 gigawatts of underutilized capacity across the U.S. power grid. The chipmaker announced the collaboration at CERAWeek 2026 in Houston alongside Emerald AI and six energy partners: AES Corp (NYSE: AES), Constellation Energy Corp (NASDAQ: CEG), Invenergy, NextEra Energy Inc (NYSE: NEE), Nscale Energy & Power and Vistra Corp (NYSE: VST).
What's Happening
For the four publicly traded names in the deal, the partnership may offer a new revenue stream: selling flexible power services to AI data centers, on top of the demand tailwind that has already pushed Constellation and NextEra to Big Tech-level valuations.
The facilities will run on Nvidia's Vera Rubin DSX AI Factory reference design, unveiled at GTC earlier this month, which includes DSX Flex software that lets data centers interact with power grids in real time.
The group says the approach could unlock up to 100 gigawatts of stranded capacity across the U.S. power system.
The conceptual flip matters: these AI factories can use co-located generation and battery storage as bridge power at launch, then supply electricity back to the grid during peak demand.
Vistra CEO Jim Burke said AI factories with flexible power use are a faster solution for grid utilization, especially with co-located generation.
Nscale's Daniel Shapiro said the company's West Virginia campus may scale from 2 to 8 gigawatts of onsite generation, positioning it as a power asset rather than a load.
"AI factories are the engines of the intelligence era, and like any great engine, every system must be designed together-energy, compute, networking and cooling as one architecture," Jensen Huang said.
Why Does it Matter for Nvidia
Nvidia's biggest problem right now isn't demand; it's that data centers can't get built fast enough.
Grid interconnection queues are backing up projects for years.
Every delayed facility is a facility that isn't buying GPUs.
If DSX Flex lets AI factories come online faster by using bridge power before full grid connection, Nvidia moves more chips, faster.
The moat isn't unchallenged.
Tesla Inc (NASDAQ: TSLA) CEO Elon Musk is building Terafab, a joint chip fab with SpaceX designed to bring AI silicon production in-house.
If Terafab delivers, it represents a major customer potentially walking away from Nvidia's ecosystem.
What The Prediction Markets Say
Polymarket bettors give Nvidia a 66% chance of finishing 2026 as the world's largest company by market cap.
The more interesting contract may be the AI Bubble Burst market, which prices a 21% chance of an AI industry downturn by year-end on $2.2 million in volume.
One trigger: Nvidia's stock falling 50% from its all-time high.
Deals like this one, locking in real energy infrastructure commitments behind AI capex, are exactly the kind of thing that keeps that number from climbing higher.