Nvidia Plans $3 Billion Investment in SB Energy to Expand into AI Data Center Energy Supply Chain
Nashnova编辑部
Nvidia is negotiating an investment of up to $3 billion in SoftBank's SB Energy, pushing its role beyond chip sales into the power and construction bottleneck — because if the electricity doesn't arrive, GPU orders don't convert to revenue.
What is Nvidia actually buying here?
Nvidia is in talks to invest up to $3 billion in SB Energy, per The Information. SB Energy is developing a large data-center project in Ohio for OpenAI.
The deal is still being negotiated. Investment share, use of funds, project scale, and chip-procurement terms are all undisclosed.
This means → Nvidia isn't buying equity in an energy company for its own sake. It is locking down the upstream links — land, power, cooling, grid connection — that determine whether GPU orders actually get fulfilled.
How deep does the Nvidia–OpenAI relationship go?
The partnership dates to 2016. In January 2025, the Stargate initiative announced plans to spend $500 billion over four years on U.S. AI infrastructure, with SoftBank handling finance, OpenAI handling operations, and Nvidia as lead technology partner.
In September 2025, Nvidia and OpenAI signed an MOU to deploy at least 10 gigawatts of Nvidia systems — millions of GPUs. Nvidia would invest progressively with each gigawatt deployed, up to $100 billion total.
The first 1-gigawatt milestone targets the second half of 2026, launching on the Vera Rubin platform.
In plain terms = the SB Energy deal extends the loop one step further upstream — from "sell chips" to "invest in chip buyers" to "invest in the people who build and power the chip buyers' facilities."
What role does SoftBank play in this?
SoftBank is the financial lead of Stargate, responsible for breaking a $500 billion vision into projects that can be financed, built, and connected to the grid.
Nvidia taking a stake in the energy or campus platform brings in industrial anchor capital and boosts lender confidence in follow-on equipment purchases and long-term leases.
This reflects a broader pattern: on August 10, 2026, Nvidia announced a standalone compute-financing platform with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR, targeting over $500 billion in third-party capital.
This means → Nvidia's strategy has shifted from "sell high-value hardware" to "make the entire AI factory an investable asset class."
What could go wrong?
Circular-growth concern: Nvidia funding its own customers' infrastructure chain raises the question of whether "capital in → chip orders out" is self-reinforcing growth.
Execution risk: A 10-gigawatt project is highly sensitive to grid capacity, transformer supply, and financing costs. Timelines could slip.
Customer diversification: OpenAI is not betting solely on Nvidia. It may still balance supply risk with alternative hardware and cloud partnerships.
Regulatory scrutiny: As ties tighten among the chip leader, model companies, cloud platforms, and energy assets, regulators may take a closer look.
What is the real thing to watch?
Analysts note that the $3 billion figure is not the point. The point is whether Nvidia is pushing the competitive frontier toward "power delivery and data-center construction capability."
Put simply = chip companies used to compete on whose chip is faster. Nvidia now wants to compete on who can get the power connected and the facility built first.
The answer will become clear only when the deal either closes or falls through.
Content is for reference only, not financial advice.