Nvidia invests up to $3 billion in Lancium, the Blackstone-backed power company providing electricity for OpenAI's Stargate hyperscale AI data center project.
Nvidia plans to invest up to $3 billion in Blackstone-backed power developer Lancium to lock in scarce electricity resources for AI data centers. An initial $2 billion investment will secure approximately a 20% equity stake, with an additional $1 billion contingent on Lancium achieving specific power interconnection milestones. If fully realized, this follow-on investment would raise Nvidia's ownership to about 30%, valuing Lancium at an enterprise value of roughly $10 billion including debt.
This marks a strategic pivot for Nvidia from chip vendor to core investor in AI infrastructure. Lancium serves as the power infrastructure provider for the "Stargate" AI super-campus being built by OpenAI and Oracle in Abilene, Texas—a landmark project in expanding U.S. AI computing capacity. The company has already secured 4 gigawatts of power on the Texas grid with another 15 gigawatts of pending interconnection projects.
The deal's structure is highly strategic. The first $2 billion represents a direct equity injection, while the remaining $1 billion release is closely tied to Lancium achieving specific business thresholds, including completing grid interconnection and regulatory approval processes. Notably, this investment involves purely an equity stake without credit guarantees for data center construction. However, Nvidia is simultaneously considering providing credit guarantee structures for other hyperscale projects, including a massive 10-gigawatt data center project in Ohio involving OpenAI and SoftBank.
The Lancium investment represents just one part of Nvidia's accelerating investment offensive. During the quarter ending in April 2026, Nvidia deployed $18.6 billion into private companies and infrastructure funds—surpassing its total investments for the entire previous fiscal year. So far this year, Nvidia has issued $2 billion checks to cloud service providers CoreWeave and Nebius, and completed investments in networking equipment companies Lumentum and Coherent as well as semiconductor firm Marvell.
Lancium's power assets within Texas are substantial. The company has secured 4 gigawatts of power resources with another 15 gigawatts of pending interconnection projects awaiting grid permits. Beyond these locked-in assets, Lancium has a land bank potentially capable of hosting 15 gigawatts of power. The company has begun laying groundwork for developing additional power land before 2035 and is positioned to benefit from a proposed ultra-high-voltage transmission network in Texas.
A regulatory variable emerged around the deal announcement. Texas Governor Greg Abbott ordered a halt to new data center grid interconnection applications, requiring further review. This policy shift introduces uncertainty into Lancium's expansion timeline. However, sources indicate Lancium's 15 gigawatt pending projects hold a significant advantage—its partner utility, American Electric Power, has already completed line studies needed to supply power to its campuses, giving these projects a first-mover advantage in the approval process.
Nvidia's capital injection provides strong endorsement for Lancium's market plans. According to sources, Lancium is considering an initial public offering as early as 2027, potentially following data center and infrastructure developers like Switch and CyrusOne. The investment not only provides capital for expansion but enhances the credibility of Lancium's business prospects in investors' eyes.
Nvidia's direct investment in a power developer reflects a deeper shift in AI industry competitive focus. As large language model parameter advancement plateaus, energy supply capacity supporting massive computing clusters has become the new bottleneck determining AI enterprise expansion speed. Analysis indicates that despite strong Nvidia GPU orders and major tech giants announcing additional investments, related companies have experienced stock price volatility. Credit default swaps for Nvidia and Oracle have risen, and CoreWeave's financing costs have climbed—the market's focus has shifted from technological breakthroughs to a more pragmatic question: who will ultimately bear the enormous capital required for this unprecedented AI revolution?
Nvidia's role is undergoing qualitative change. Previously, it provided equity financing and GPU procurement financial support to new cloud service providers like CoreWeave, Nebius, and Lambda. Now it is beginning to engage in financing credit guarantees for customer projects themselves, embedding itself into the financial links of the AI industry chain. If a GPU seller assumes ultimate risk, the semiconductor industry takes on attributes of the financial industry.
Meanwhile, competition for pricing power in AI models adds uncertainty to this capital expansion. Chinese firms like DeepSeek and Kimi are challenging the enterprise market with API token prices far lower than U.S. counterparts, while OpenAI and Anthropic are also continuously cutting prices. For investors, the core question remains: if hundreds of billions of dollars are poured into data center construction while token prices continue falling, how will payback periods for these massive investments be guaranteed? This is the real-world test that projects like Stargate must face. The financial health and financing costs of heavy asset investors like Oracle are becoming more closely watched by the market than the AI models themselves.
Against this backdrop, Nvidia's investment in Lancium represents the company embedding itself deeper into the physical and financial infrastructure underlying artificial intelligence deployment.