In One Week, Nvidia Signed a $50 Billion Lease and Entered Talks to Guarantee $250 Billion for OpenAI
Last week was already a big one for AI news. Sam Altman declared we’re living through the singularity. Anthropic’s CEO clarified the company’s position on open-weights models. Google announced its “Beyond Zero” security paradigm.
But while those headlines grabbed the clicks, two quieter stories broke that I think matter more for where this industry is actually headed. The singularity debate, the open-weights controversy — and they both involve Nvidia.
On Monday, the Financial Times reported that Nvidia signed a 15-year lease for a 1-gigawatt data center in Texas — a deal worth up to $50 billion if renewed for 30 years. Two days earlier, the Wall Street Journal broke the news that Nvidia is in talks to provide a $250 billion financial backstop for OpenAI’s planned 10-gigawatt data center project in Ohio.
Put those two numbers together: Nvidia is putting somewhere between $50 billion and $300 billion on the line for AI infrastructure. In one week. That’s not a chip company anymore. That’s a shadow bank for the AI economy.
The Two Deals That Redefined Nvidia’s Role
Deal 1: The $50 Billion Texas Lease
Nvidia signed a contract to lease a massive data center that Hut 8 is building in Texas. The facility runs at 1 gigawatt — that’s roughly the capacity of a nuclear reactor. Nvidia committed to a 15-year lease valued at $19.6 billion, which could climb to $50 billion over 30 years if renewed.
The data center will house hundreds of thousands of Nvidia GPUs. After completion, Nvidia plans to sublease parts of the facility to “neocloud” partners — smaller cloud operators that need access to Nvidia hardware but can’t afford to build their own infrastructure.
This isn’t Nvidia’s first infrastructure play. The company has been making substantial investments to help neocloud companies purchase and operate its chips. But this lease goes a step further — Nvidia is now underwriting the data center itself, not just the silicon inside it.
The effect on Hut 8’s cost of capital was immediate. The company issued about $4.3 billion in corporate bonds in June at 6.129% — priced roughly 100 basis points above Nvidia’s own 30-year bonds. Nvidia’s credit rating effectively became Hut 8’s credit rating.
Deal 2: The $250 Billion OpenAI Guarantee
This one is still in negotiation and could fall apart, so take the numbers with appropriate skepticism. But the Wall Street Journal and New York Times both reported that Nvidia is discussing a roughly $250 billion financial backstop for OpenAI.
The backstop would help OpenAI lease a 10-gigawatt data center being developed by SoftBank’s energy subsidiary in southern Ohio. The project’s power is controlled by the U.S. government and funded separately by Japan under a recent trade deal, with Commerce Secretary Howard Lutnick reportedly involved in access decisions.
Ten gigawatts is an almost incomprehensible amount of power. For context, the largest data centers today run at 100-500 megawatts. A 10-gigawatt facility is equivalent to 8-10 nuclear power plants running full-time to power one AI campus.
If the deal goes through, OpenAI gets its infrastructure without needing investment-grade credit. Nvidia gets a guaranteed customer for its most expensive products for years to come.
What “Circular Financing” Actually Means
If you’re seeing a pattern here, you’re not alone. The Seoul Economic Daily called it “circular financing,” and the term is fitting.
Here’s how the cycle works:
- Nvidia manufactures the most advanced AI chips in the world
- Nvidia either leases data centers directly (Texas) or guarantees loans for customers to lease them (Ohio)
- Those data centers are filled with — you guessed it — Nvidia GPUs
- Customers pay Nvidia for the chips, which Nvidia uses to finance more facilities and more guarantees
The circle closes on itself. Nvidia puts up capital, which pays for Nvidia hardware, which generates revenue that justifies more capital deployment.
It’s brilliant. And it’s risky.
Nvidia isn’t the first company to pull this off. Apple did something similar with its $100 billion stock buyback program — using borrowed money to buy its own shares, which increased the stock price, which made the debt cheaper. But the scale here is different. Nvidia is essentially acting as the financial backbone of the entire AI industry.
Why Nvidia Is Doing This
The straightforward answer: Nvidia has an insatiable demand problem. Every AI company, every cloud provider, every startup with a GPU cluster wants Nvidia’s chips. But building data centers takes years, and financing them takes even longer.
By stepping in as a guarantor and lessee, Nvidia unblocks the bottleneck. Data centers get built faster. Chips get installed sooner. Nvidia’s revenue curve gets steeper.
There’s also a competitive angle. AMD, Intel, and a growing list of AI chip startups like Etched and the team behind AMD’s Helios are all trying to eat Nvidia’s lunch. If Nvidia can lock in infrastructure commitments at this scale, it creates an enormous moat. A startup that builds its entire business around Nvidia-backed data centers isn’t going to switch to AMD chips overnight, no matter how competitive their specs look on paper.
And then there’s the “keep Jensen happy” factor. Nvidia CEO Jensen Huang has been vocal about AI infrastructure being the new oil and gas. In a world where AI demand could double or triple every year, the winners are the ones who build the most infrastructure the fastest. Nvidia is placing a massive bet that this growth curve continues.
The Risks Nobody’s Talking About
This is the part I can’t stop thinking about. What happens if AI demand doesn’t materialize as expected?
Nvidia could be holding $300 billion or more in lease obligations and loan guarantees. If OpenAI runs into financial trouble (which, let’s be honest, a company spending $500 billion on infrastructure before earning real revenue from its most expensive models is not impossible), Nvidia would be on the hook.
The circular financing model works brilliantly in an up market. In a down market, it turns into a chain of dominos.
There’s also a concentration risk that should concern every developer and startup founder reading this. If Nvidia controls both the chips AND the infrastructure financing, the entire AI industry becomes dependent on one company’s strategic decisions. We’ve already seen how fragile single-supplier dependencies can be — the pandemic-era chip shortage is still fresh in everyone’s memory.
AMD, Intel, and startups like Etched and Groq are trying to build alternatives — AMD’s ROCm.AI is already telling models to write their own GPU kernels. But they’re not just competing with Nvidia’s technology anymore. They’re competing with Nvidia’s balance sheet.
What This Means for Developers and Businesses
If you’re building on top of AI models — whether you’re running local LLMs like I do or calling APIs from providers — this concentration should inform your strategy.
I’ve been saying this for a while, but it’s worth repeating: don’t bet everything on one provider. The more the AI infrastructure market consolidates, the more important it becomes to maintain optionality. Use multi-model strategies. The AI model marketplace is already consolidating fast. Keep local inference running. Pay attention to where the bottlenecks are — because right now, they’re all flowing through Nvidia.
On the flip side, if you’re in the data center or energy business, these deals are a massive signal. The market for AI infrastructure isn’t just growing — it’s attracting capital at a scale that would have been unthinkable even two years ago. The companies that figure out how to build faster and cheaper are going to be the unsung heroes of this boom.
The Bottom Line
Nvidia just became a $3 trillion company by selling chips. These two deals suggest it’s aiming to become something much bigger — the financial and infrastructural backbone of the AI industry.
The $50 billion Texas lease shows Nvidia is willing to put its own balance sheet on the line to accelerate infrastructure deployment. The $250 billion OpenAI guarantee, if it goes through, would be the single largest financial commitment to AI infrastructure ever made. Combined, they represent a vision where Nvidia isn’t just a supplier — it’s the bank.
Whether this works out depends on whether AI demand keeps growing at its current trajectory. If it does, Nvidia will have positioned itself at the center of the most valuable industry on Earth. If it doesn’t, we’ll be studying this as one of the most spectacular cases of financial engineering — and overreach — in tech history.
As someone who runs AI models on local hardware and watches these numbers with a mix of awe and concern, I’d say the smart money is on continued growth. But the smartest money is on staying diversified, staying flexible, and never betting the farm on any single company’s vision of the future.