The ChatGPT maker closed the largest private tech funding round in history, with cash from Amazon, Nvidia, and SoftBank, while its valuation climbed past the GDP of many sovereign nations.
OpenAI, a company that didn’t exist a few years ago, announced that it closed a $110 billion funding round, obviously the largest private financing event in tech history. The cash injection vaults the company to a staggering $730 billion pre-money valuation (or $840 billion post-money). In October 2025, secondary markets pegged OpenAI at $500 billion, which was already double its valuation from a then record $40 billion raise the year prior.
But if you look past the sheer volume of zeroes, this is a geopolitical scale bartering system for the world’s most constrained resources: electricity and advanced silicon.
The money comes from a trio of backers who don’t want to miss the AI train. Amazon invests $50 billion (though $35 billion of that is with unnamed conditions), Nvidia contributes $30 billion, and SoftBank rounds out the top three with another $30 billion. More investors are expected to join as the round closes out, because $110 billion has room to grow.
Although Amazon has invested in OpenAI, they’ve maintained a degree of control. Only $15 billion is being released today, with the remaining $35 billion contingent on meeting undisclosed performance criteria over the coming months.
OpenAI is practicing cloud polygamy, and that’s the main thing to watch. For years, Microsoft was the undisputed, exclusive sugar daddy of OpenAI’s infrastructure. Now, ChatGPT’s head organization is aggressively diversifying, and the fine print of these new deals reveals a company desperate to lock down compute power before the grid taps out.
What Nvidia Wants Out of This
NVIDIA’s $30 billion deal has an important purchase commitment attached to it. OpenAI will deploy 3 gigawatts of inference capacity and 2 gigawatts of training capacity on Vera Rubin, Nvidia’s next-gen GPU architecture that succeeds the current Blackwell chipsets (Yes, we’ll now measure software infrastructure in gigawatts, a metric historically reserved for nuclear power plants and time travel DeLoreans). So Jensen Huang is investing in his own best customer, a fine trick if you can pull it off.
OpenAI already uses Nvidia’s old Hopper and Blackwell hardware across Microsoft Azure, Oracle Cloud, and CoreWeave, and the Vera Rubin commitment is on top of all that, so Team Green will not suffer.
Amazon’s Very Expensive Cloud Deal
Then there’s the Amazon deal, which is a masterclass in playing your partners against each other. Along with the capital, OpenAI is committed to spending $100 billion on AWS compute over eight years. Crucially, they have agreed to use at least two gigawatts of Amazon’s Trainium AI chips that hedge against NVIDIA‘s hardware monopoly and are built to break Team Green’s stranglehold on AI hardware.
What does Amazon get? Exclusive rights to distribute Frontier, OpenAI’s enterprise platform, to third parties. Both companies have also agreed to set a custom “stateful runtime environment,” so OpenAI models can run natively on Amazon’s Bedrock. Amazon CEO Andy Jassy has called this a paradigm shift for AI developers. Maybe. What it actually does is carve straight into Microsoft’s turf.
The Infrastructure Breakdown: Nvidia vs. Amazon
| Feature | Nvidia Commitment | Amazon (AWS) Commitment |
| Direct Investment | $30 Billion | $50 Billion ($15B upfront, $35B conditional) |
| Primary Hardware | Vera Rubin (Next-gen GPU architecture) | Trainium (Amazon’s proprietary AI chips) |
| Power Allocation | 5 Gigawatts (3GW Inference / 2GW Training) | 2 Gigawatts (Dedicated Trainium capacity) |
| Cloud Ecosystem | Multi-cloud (Azure, OCI, CoreWeave) | AWS (Exclusive 3rd-party Frontier channel) |
| Software Integration | Successor to Blackwell/Hopper systems | New “Stateful Runtime” for Amazon Bedrock |
| Long-term Play | Securing the GPU monopoly | Breaking the Microsoft/Azure monopoly |
Given that Microsoft has been OpenAI’s primary financial backer and cloud partner since before anyone outside of the industry knew what a large language model was, the Amazon deal required some careful messaging. OpenAI and Microsoft said in a joint statement that Azure will remain the exclusive cloud provider for OpenAI’s APIs and their own products, and Microsoft will keep its ironclad intellectual property licenses. But the walls of exclusivity are clearly cracking as OpenAI’s compute appetite outgrows what any single mega-cap can provide.
How does a private company justify asking for $110 billion and a small country’s worth of electricity? By pointing to consumer traction that borders on the absurd.
OpenAI currently has over 900 million weekly active users on ChatGPT. More importantly, they have crossed 50 million paid subscribers. Do the napkin math: at roughly $20 a head, that’s $1 billion in recurring revenue every single month, just from consumers. Add in 1.6 million weekly users on Codex (three times higher since January), and the enterprise revenue can match the huge infrastructure costs.
In the press release, OpenAI stated, “We are entering a new phase where frontier AI moves from research into daily use at a global scale. Leadership will be defined by who can scale infrastructure fast enough to meet demand, and turn that capacity into products people rely on.” It’s still up in the air if AGI will help humanity, but in the short term, NVIDIA, Amazon, and the bank accounts of whoever happens to own the local power grid are definitely going to make serious money.
