Microsoft feared OpenAI reliance, Musk-Altman trial testimony reveals

Elon Musk’s courtroom battle with Sam Altman over the past few weeks has primarily revolved around the evolution of OpenAI. But it also shed light on the challenges posed by the rapid growth of artificial intelligence OpenAI’s closest partner: Microsoft.

The discovery in the high-profile case showed that Microsoft CEO Satya Nadella was anxious about OpenAI replacing his company in the tech hierarchy by April 2022, seven months before the launch of ChatGPT, which launched the prolific AI boom and turned Altman into a household name.

“I don’t want to be IBM’s and OpenAI will be Microsoft,” Nadella wrote in an email to executives in April, nearly three years after Microsoft wrote its first $1 billion check to OpenAI.

Nadella was talking about an earlier era of technology when Microsoft became more important than IBM, the dominant computer maker at the time. In 1980, IBM agreed to distribute Microsoft’s operating system on its computers, but found that the software maker eventually took the lion’s share of the market value.

To avoid the same fate, Microsoft needed to make sure it was in a position to not only provide Azure cloud services to OpenAI, but also capitalize on its early intellectual property agreement with the lab at the forefront of major language model development.

Nadella held the Musk v. debate in Oakland, California, on Monday. “It was becoming even more fundamental and important that we had real representation at every layer of the stack,” Altman testified in his trial.

In the more than four years since IBM’s joke, history has vindicated Nadella’s initial fears. OpenAI’s models and services are so ubiquitous that the company has reached an $850 billion valuation and has established partnerships with Microsoft’s cloud rivals Google, Seer and as of February Amazon.

“Essentially, we were giving up the opportunity to develop it on our own, so it was very important for us to have the intellectual property rights,” Nadella said Monday, referring to the initial deal with OpenAI.

Microsoft and OpenAI had to change the terms of their agreements multiple times. Most recently, in April, they renewed the agreement so that revenue-sharing payments from OpenAI to Microsoft were limited and that OpenAI could now offer “all of its products” to customers at any provider, including Amazon and Google.

The partnership remains critical to Microsoft’s cloud business as demand for the computing resources needed to run large models continues to grow. About 45% of Microsoft’s remaining business performance obligations were tied to OpenAI at the end of 2025, and the software giant is building data centers to meet demand.

“It’s better to be an investor and not even take all this practice risk!” Nadella told executives in an email that surfaced in July 2022.

Microsoft will spend more than $100 billion on OpenAI by June 2026, including investment commitments, infrastructure and hosting costs, Microsoft corporate development executive Michael Wetter said in court Wednesday.

Microsoft’s chief technology officer, Kevin Scott, said the first supercomputer his company built with OpenAI years ago took about six months to assemble 10,000 graphics processing units that took up a large portion of a data center.

Scott said he is “very proud of what we’ve enabled OpenAI to do, both in terms of research and the products they send out into the world.”

Nadella said at the hearing that adopting OpenAI taught Microsoft how to build supercomputers suitable for AI work and that “one benefit of that is the technical knowledge we’ll gain from doing this.”

But while Microsoft has established itself as a major provider of AI infrastructure, it is no longer uniquely positioned to deliver OpenAI technology to customers. And Microsoft has struggled to sell its own AI story, which explains why While the stock is down 16% this year, its cloud peers are all trading higher.

Nadella as of January 2024 saying that models have become “more of a commodity.” Two months later, the company hired Mustafa Süleyman, co-founder of the DeepMind AI laboratory, now owned by Google. Nadella made Suleiman CEO of a new artificial intelligence unit and put him in charge of Bing and other products.

With Suleiman at the helm of AI, Microsoft has focused on building its own models that can compete with OpenAI as well as other leaders in the space such as Anthropic and Google. In August last year, the company said it had begun testing a self-developed artificial intelligence model that could lead to improvements to its Copilot assistant for consumers.

While Microsoft is trying to find its place in the world of models, the market is moving at an increasing pace. In March, Microsoft restructured the Copilot leadership team and named a former recent hire. explode executive Jacob Andreou to run the consumer and commercial Copilot experience. As part of the changes, Suleiman was left to develop models.

Since 2024, Microsoft has openly acknowledged that OpenAI is a competitor and is allying with other companies developing AI models, including Musk’s xAI, and making those tools available to developers through Azure. Last year, Microsoft announced it was leveraging Anthropic’s AI models for certain use cases, and soon after agreed to invest $5 billion in its OpenAI rival.

Microsoft has AI to thank for much of the continued growth in Azure. Its balance sheet includes highly profitable equity investments. However, unlike ChatGPT, none of the AI-powered software products have been a big hit.

This left the company with a scattered strategy that involved engaging many model developers while also building its own stack parts, including models and chips.

“Some of the rights that we had in the early days, we were flexible and kind of released them so that OpenAI could continue to scale and evolve,” Nadella said Monday. he said. “To me, our core values ​​as a company are to be a good partner and a good platform company, so we keep that front and center no matter what happens on the side.”

WRISTWATCH: Microsoft CEO Satya Nadella testified in OpenAI case