DeepHorus is an innovative tech media and intelligence platform focused on AI, blockchain, startups, and emerging technologies.
Published on

Nvidia reportedly plans to invest up to $3 billion in Lancium, a power infrastructure developer

Authors
  • avatar
    Name
    DeepHorus
    Twitter

Nvidia is accelerating its efforts in the foundational energy sector of AI infrastructure. According to The Information on August 8, the chip giant plans to invest up to $3 billion (approximately 20.279 billion yuan) in power infrastructure developer Lancium to secure electricity resources needed for AI data centers. People familiar with the matter revealed that Nvidia will initially invest $2 billion for about a 20% stake in Lancium; if Lancium meets specific milestones, including grid interconnection, Nvidia will add another $1 billion in investment.

Lancium is a power infrastructure company backed by Blackstone, headquartered in Houston, Texas. It was co-founded by Michael McNamara, who also serves as CEO, in 2017. The company initially focused on energy transition technologies. Its core product, Smart Response software, enables clean campuses to act as controllable load resources for the grid, helping balance the variability of renewable energy generation and providing grid reliability services traditionally handled by fossil fuel power plants. Lancium's early business included energy-intensive applications such as Bitcoin mining, before gradually transitioning to focus on power infrastructure for AI data centers.

In this deal, Lancium and its land and power resource assets are valued at approximately $10 billion (about 67.597 billion yuan). The company is also evaluating the possibility of an IPO in 2027. The key to this investment is Lancium's 'Clean Campus' in Abilene, Texas, which is the site of the first phase of the 'Stargate' AI supercomputing data center project.

Stargate is a joint venture project to build data centers, co-founded by SoftBank, OpenAI, Oracle, and others. It plans to invest up to $500 billion (approximately 33.8 trillion yuan) in the U.S. by 2029 to build AI data centers and supercomputing infrastructure. In July 2025, OpenAI announced a 4.5 GW data center cooperation agreement with Oracle. The first phase site, Stargate I in Abilene, has been put into operation, with some facilities already using Nvidia GB200 racks for early training and inference workloads. In February 2026, according to The Information, OpenAI adjusted its project strategy, shelving plans to build its own data centers, and instead partnering with Oracle to build a 4.5 GW data center and with SoftBank to develop a 1 GW campus in Texas. It also hired former Intel CTO Sachin Katti to lead the infrastructure team.

Nvidia's investment is in line with the AI 'five-layer cake' theory proposed by its CEO Jensen Huang. The theory divides the AI industry from bottom to top into five layers: energy, chips, infrastructure and system software, AI models, and applications. Energy is defined as the 'first principle' of AI infrastructure. In 2025, Nvidia made more than 10 moves in areas such as clean energy development, grid management, and data center construction, including investments in nuclear fusion company Commonwealth Fusion Systems (CFS) and nuclear energy company TerraPower, founded by Bill Gates.

Analysts believe this investment marks the chip giant's transformation from simply selling hardware to deeply integrating energy resources. As AI computing demand grows exponentially, power supply has become a core bottleneck for data center expansion. Lancium's power assets and grid management technology can provide stable energy security for Nvidia's customers, while also helping the smooth progress of the Stargate project. If Lancium succeeds in its IPO, Nvidia will gain substantial financial returns and further consolidate its dominant position in the AI industrial chain. In the future, Nvidia's initiatives in the energy sector may continue to expand, from traditional power to emerging technologies such as nuclear energy and energy storage, building a complete AI energy ecosystem.