OpenAI just got a step closer to its blockbuster IPO
OpenAI CEO Sam Altman has been clear that AI’s future will demand enormous computing power to match the ambitions of developers and hyperscalers building with it.
“Theoretically, at some points, you can see that a significant fraction of the power on Earth should be spent running AI compute,” Altman said during an AMD-hosted conference this past summer. “And maybe we’re going to get there.”
It appears OpenAI is definitely trying to get there.
Last month, the company purchased 6 gigawatts of AMD’s (AMD) Instinct GPUs as part of a multi-year agreement.
In September, OpenAI also reached a deal with Nvidia (NVDA) for 10 gigawatts of Nvidia systems, which also saw the chip giant agree to invest up to $100 billion in the company.
A couple of weeks before inking the deal with Nvidia, OpenAI signed a contract to purchase $300 billion of computing power from Oracle (ORCL).
And on Monday, the company entered into a $38 billion agreement with Amazon’s (AMZN) AWS that gives OpenAI access to AWS compute over the next seven years, featuring hundreds of thousands of Nvidia GPUs.
The deal also gives OpenAI an opportunity to expand to tens of millions of CPUs.
As part of the agreement, OpenAI will immediately begin deploying AWS compute, with a plan to reach full capacity by the end of 2026. It can also expand further into 2027 and beyond.
“Scaling frontier AI requires massive, reliable compute," Altman said in a statement. “Our partnership with AWS strengthens the broad compute ecosystem that will power this next era and bring advanced AI to everyone.”
Amazon’s shares rose 4% on Monday.
Wedbush analysts, led by Dan Ives, maintained their Outperform rating on Amazon, while raising their price target to $340 from $330.
Ives said that the deal with OpenAI reflects “AI-driven AWS growth and cloud infrastructure leverage,” and sees these “AI partnerships as key catalysts for long-term AWS leadership.”
OpenAI is planning its blockbuster IPO
The AWS infrastructure for OpenAI will include both Nvidia GB200 and GB300 chips through its Amazon EC2 UltraServers.
The compute will support multiple workloads for OpenAI, including serving inference for ChatGPT and for training its next generation models.
“As OpenAI continues to push the boundaries of what's possible, AWS's best-in-class infrastructure will serve as a backbone for their AI ambitions,” AWS Matt Garman said in a statement.
“The breadth and immediate availability of optimized compute demonstrates why AWS is uniquely positioned to support OpenAI's vast AI workloads."
The big question on Wall Street is how much longer OpenAI will be able to stay a private company.
While it has secured substantial investment from Nvidia, Altman has said that he wants to have OpenAI eventually get to the point where it’s spending over $1 trillion a year on infrastructure.
The company announced a reworked deal with Microsoft (MSFT) last week that lifts a cap on how much money OpenAI can raise, giving it significantly more flexibility than it previously had.
But as Altman’s ambitions for the company continue to grow, there is a sense that it’s only going to be able to achieve them by becoming a public company, which would give it much greater access to capital.
Altman himself has admitted that in order to reach his goal of spending over $1 trillion a year on infrastructure, the “most likely path” for the company is to pursue an IPO.
And in fact, Reuters reported last week that OpenAI is currently “laying the groundwork” for its public offering, which might happen as soon as the second half of 2026.
The company could be valued at $1 trillion, which would almost certainly create one of the biggest blockbuster IPOs that Wall Street has ever seen.