Nvidia (NASDAQ: NVDA) is investing up to $3 billion in Lancium, the power infrastructure company at the center of the Stargate AI data center initiative, according to Reuters.
The deal carries the potential to expand by an additional $1 billion if Lancium clears specific benchmarks, including milestones tied to connecting to the electrical grid.
The transaction assigns Lancium, together with its land holdings and power assets, an enterprise value of roughly $10 billion, underscoring the scale of capital now flowing into AI infrastructure.
Lancium plans to deploy the fresh capital to expand its operations while also evaluating whether to pursue a public listing in 2027.
Neither Nvidia nor Lancium offered a response when Reuters sought comment on the details of the deal.
Lancium’s flagship property, the 1,000-acre Lancium Clean Campus in Abilene, Texas, is where the Stargate project first became operational.
Stargate brings together SoftBank, OpenAI, and Oracle (NYSE: ORCL) with the goal of constructing a network of AI data centers across the United States.
President Donald Trump unveiled the Stargate initiative in January, at which point the three founding partners committed to channeling as much as $500 billion into AI infrastructure development.
Nvidia CEO Jensen Huang has been outspoken about the investment opportunity in AI infrastructure, arguing at a private gathering of family offices and financial institutions in Taipei that returns had reset over the prior six months and were now profitable at a scale that had silenced earlier skeptics.
“You need land, power, you need energy, but you also need financing,” Huang said, framing wealthy family offices as an emerging source of capital alongside more traditional institutional investors.
Nvidia has also extended its infrastructure footprint through other means, including a broader partnership with chip designer Marvell Technology that included a $2 billion investment.
The Lancium deal signals that Nvidia is moving beyond chip manufacturing to take direct stakes in the physical infrastructure required to run the AI systems its hardware powers.
