The AI party not only continues... it’s starting to get completely out of control! Because OpenAI has just pulled off one of the biggest tech deals in history. And the question is no longer “who will dominate,” but rather... “who can afford the cost.”
If you think we’ve already seen everything in the AI race — think again. What’s happening in recent weeks looks more like an arms race.
Everyone’s scrambling to secure computing power, GPUs, partnerships, and data access — and no one wants to be left behind.
The pie keeps growing, but the slices are getting very expensive.
THE DEAL WITH AMAZON
We start with the mammoth OpenAI–Amazon deal. OpenAI has signed a $38 billion agreement with AWS, giving it access to hundreds of thousands of Nvidia GPUs and the ability to scale to tens of millions of CPUs in the future. The infrastructure will be used for both training and inference — meaning everything from model training to live ChatGPT responses.
The 7-year deal also includes access to Amazon’s UltraServers. According to Sam Altman:
“Scaling frontier AI requires massive, reliable compute power — and AWS is here to meet that need.”
From Amazon’s side, the company emphasizes that this partnership reinforces AWS’s position as the cloud leader, offering unmatched scale and availability in AI infrastructure. And don’t forget — Amazon has also invested billions in OpenAI’s rival, Anthropic. So the fact that “there’s room for both” says a lot about the sheer magnitude of demand.
THE TRILLION-DOLLAR DEALS
But that was just the beginning. In less than a month, OpenAI has announced partnerships worth over $1.4 trillion(!) — involving Nvidia, Broadcom, AMD, Oracle, Google ($GOOG), Microsoft, and Salesforce.
Specifically:
- With Broadcom → collaboration on 10GW of custom AI accelerators.
- With AMD → building out 6GW of Instinct GPUs.
- With Arm → designing CPUs that plug directly into AI servers.
- With Microsoft → a $250 billion contract for Azure services.
- Ongoing partnerships with PayPal, Shopify, Salesforce, and Thermo Fisher.
And of course, Nvidia ($NVDA) is the clear winner in all of this. Almost every major project — from training GPUs to inference chips — runs on Nvidia hardware. OpenAI alone has locked in massive quantities of the new Blackwell chips, which are seen as game changers for speed and energy efficiency.
But here’s the big question: Where will OpenAI find all the money for this?
The answer isn’t clear yet. Rumors are already swirling that the company is preparing for an IPO, possibly by late 2026, with a potential valuation near $1 trillion. If that happens, it would be the second-largest IPO in history, after Saudi Aramco. Until then, though, the cash burn rate is enormous, and the pressure to generate revenue keeps increasing.
MICROSOFT’S MOVE
While OpenAI is doing its thing, Microsoft isn’t sitting still. It just signed a $9.7 billion deal with IREN, a data-center company, to gain indirect access to Nvidia GB300 chips through Dell hardware. This deal allows Microsoft to expand its AI capacity without having to build new data centers or secure additional power — the two biggest bottlenecks in AI growth right now.
At the same time, it signed another $17.4 billion deal with Nebius, further strengthening its infrastructure to meet skyrocketing AI demand. And as if that weren’t enough, just yesterday it was revealed that Microsoft also struck a multi-billion-dollar deal with Lambda, adding even more Nvidia-based AI infrastructure.
In other words, Microsoft is now operating more like a broker of AI compute power than a traditional cloud provider. The model is changing — and everyone will have to adapt.
Posted Using INLEO