Photo by Suppanuch Wongpasklang on Unsplash
OpenAI’s CEO Sam Altman is on a deal-making streak that’s caught the entire tech industry’s attention. After signing massive contracts this year with giants like Oracle, Nvidia, and AMD, Altman says there’s more coming — and soon.
Let’s unpack what’s going on here.
OpenAI’s $1 Trillion Ambition
If you’ve been wondering how serious OpenAI is about building the future of artificial intelligence, consider this: in 2025 alone, the company has already signed deals worth an estimated $1 trillion.
That’s trillion — with a T.
Here’s a quick snapshot of what’s on the table so far:
- A $500 billion infrastructure deal with Oracle and SoftBank (called “Stargate”) to build 10 gigawatts’ worth of facilities in the U.S.
- A separate $300 billion cloud service agreement with Oracle.
- A partnership with Nvidia to directly supply AI hardware — including GPUs and networking systems — for another 10 gigawatts of AI data centers.
- A one-of-a-kind deal with AMD for 6 gigawatts, structured in a way that makes OpenAI a major shareholder in AMD itself.
- A “Stargate UK” initiative to expand AI data centers in Europe.
And that’s just what we know so far.
Deals For Chips, Deals For Equity
Photo by Aperture Vintage on Unsplash
What’s fascinating — and a little controversial — is how these deals are being structured.
Let’s take AMD, for example. OpenAI isn’t just buying chips from them. Instead, AMD is issuing OpenAI large amounts of AMD stock, potentially giving the AI company up to 10% ownership over time. In return, OpenAI gets access to AMD’s next-gen AI chips and helps develop them along the way.
It’s clever. It’s unusual. And it makes OpenAI a direct stakeholder in its supplier.
Flip the script to Nvidia, and it’s the reverse. There, Nvidia is investing up to $100 billion into OpenAI itself. That move made Nvidia a shareholder of OpenAI.
These are not your average vendor contracts. They’re investments with long-term leverage baked in — and they’re fueling OpenAI’s rapid march toward becoming what Nvidia’s CEO Jensen Huang called a “self-hosted hyperscaler,” meaning OpenAI wants to run its own global-scale data centers instead of renting cloud space from others.
The Money Problem (And the Bet Altman’s Making)
The challenge? These facilities aren’t cheap. According to Huang, just one gigawatt’s worth of AI-ready data center space could cost between $50 billion and $60 billion. OpenAI’s infrastructure appetite clocks in at nearly 30 gigawatts already.
That’s a huge mismatch with OpenAI’s current revenue, which hit about $4.5 billion in the first half of 2025 — impressive, but nowhere near the trillion-dollar commitments it’s stacking up.
So how does this all work?
As Altman put it, OpenAI is making an “aggressive infrastructure bet” because they expect future AI models and tools to create massive demand and economic value.
“I’ve never been more confident in the research road map in front of us,” Altman said on the a16z Podcast. But to build that future, he says, “we kind of need the whole industry… from the level of electrons to model distribution.”
In other words, OpenAI can’t do it alone — and that’s why it’s pulling in partners from every corner of the tech world. From chips to data centers to cloud access, it’s building its own vertically integrated AI empire, one contract at a time.
Stay Tuned: More Deals Are Coming
Photo by Masjid MABA on Unsplash
When asked whether we should expect more blockbuster deals in the months ahead, Altman didn’t hesitate.
“Yes,” he said. “You should expect much more from us.”
So, if you thought OpenAI was done after partnering with Oracle, Nvidia, and AMD — think again. The company’s just warming up.
And honestly? I can’t look away.
Keywords: OpenAI, Sam Altman, Nvidia, AMD, Oracle, Stargate deal, AI infrastructure, data centers, hyperscaler, AI GPUs, trillion-dollar investment, tech partnerships, a16z podcast
