Nvidia Backs Lancium With Up to $3 Billion for AI Power Expansion

Key Takeaways

• Nvidia reportedly plans an initial $2 billion investment for roughly 20% of Lancium, with another $1 billion linked to milestones.
• The transaction values Lancium’s land and power-connection portfolio at an enterprise value of around $10 billion.
• Nvidia’s move reflects a broader shift in AI infrastructure strategy, with grid access, land and cooling capacity becoming as important as computing hardware.

Nvidia is preparing to invest up to $3 billion in Lancium, the Blackstone-backed power infrastructure developer whose Texas campus supports the Stargate data center initiative, according to reporting cited by Reuters.

The proposed transaction begins with a $2 billion investment for a stake of roughly 20% in Lancium. Nvidia could provide another $1 billion if Lancium meets specified thresholds, including milestones related to grid hookups, according to people familiar with the matter cited in the original report from The Information.

The transaction structure links one-third of Nvidia’s potential investment directly to the physical infrastructure required to energize new computing capacity.

Advanced processors require immediate access to a sufficiently large and reliable power supply to operate at commercial scale.

Under the reported terms, Lancium and its portfolio of land and power connections have an enterprise value of around $10 billion. The investment would provide Lancium with capital to expand its operations while it explores a potential initial public offering in 2027. Nvidia and Lancium had not responded to Reuters’ requests for comment at the time of publication.

Lancium owns the 1,000-acre Lancium Clean Campus in Abilene, Texas. The site is the first operational campus associated with Stargate, the data center joint venture involving SoftBank, OpenAI and Oracle. U.S. President Donald Trump announced the project in January, saying the participating companies would invest up to $500 billion in artificial intelligence infrastructure.

That puts Lancium in an unusually strategic position. It controls not just acreage, but access to power connections that can take years to secure. Coverage from The Economic Times highlights the link between Nvidia’s proposed stake and Lancium’s role in the first Stargate campus.

Power has become one of the defining constraints on AI expansion. U.S. data centers consumed about 176 terawatt-hours of electricity in 2023, representing roughly 4.4% of total U.S. electricity use, according to research from the U.S. Department of Energy’s Lawrence Berkeley National Laboratory. Consumption could reach between 325 and 580 terawatt-hours by 2028 if AI-related growth continues.

Globally, the International Energy Agency estimated that data center electricity demand could more than double from 2022 levels by 2026, with AI among the major drivers. Meanwhile, long interconnection queues mean proposed generation projects can wait years before gaining access to transmission infrastructure.

Beyond equity, Nvidia is gaining exposure to a scarce portfolio of development-ready land and prospective electrical capacity tied to one of the industry’s largest infrastructure programs.

The structure also gives Nvidia some protection against execution risk. Linking the additional $1 billion to grid hookups and other thresholds means Lancium would receive the full investment only as key infrastructure advances. That approach reflects the reality that land announcements and construction plans do not automatically translate into energized computing capacity.

As Finimize noted in its coverage, the potential investment reaches beyond Nvidia’s traditional role as a chip supplier and places it closer to the power layer supporting AI deployment. It also deepens Nvidia’s relationship with an ecosystem that includes Blackstone, OpenAI, Oracle and SoftBank.

Operational design will remain another consideration. Large AI campuses typically have to address resilience, redundancy and cooling alongside raw electricity supply. Frameworks such as the Uptime Institute Tier classification system and ANSI/TIA-942 provide reference points for designing data center infrastructure, although compliance with a framework does not remove local grid or construction constraints.

For enterprise technology buyers, the proposed deal signals that AI capacity planning is moving further upstream. GPU availability still matters, but so do substations, transmission access, cooling systems and construction schedules. Nvidia’s reported Lancium investment suggests the next stage of competition may be shaped as much by control over megawatts and interconnections as by performance gains inside the chip itself.