Microsoft feared OpenAI jump to AWS

Cloud contracts are usually boring. This one isn’t. Internal Microsoft emails now in evidence at Elon Musk’s OpenAI trial show the company wasn’t just sizing up a research lab in 2018 — it was worrying that OpenAI could run to Amazon, validate AWS as the best place to build frontier AI, and leave Azure looking weak. That matters because Satya Nadella was expected to testify on Monday, May 11, about the same relationship that later became one of the most important alliances in tech. ### What actually came out? The new material is a batch of Microsoft emails and internal documents aired in Musk’s lawsuit against OpenAI, Sam Altman, Greg Brockman, and Microsoft. The case is in federal court in Northern California, and the surviving claims focus on whether OpenAI abandoned the nonprofit mission Musk says he helped fund. Microsoft already knocked out some claims earlier this year, but the court let parts of the case keep moving. (techxplore.com) ### Why was Microsoft nervous about AWS? Because this was never only about idealism. Azure was chasing AWS, and OpenAI’s researchers were already seen as unusually strong. If that team built its future on Amazon’s cloud, Microsoft risked losing twice — once on revenue, and again on reputation. GeekWire’s earlier review of the court record shows Nadella was sensitive to AWS from the start, even back in 2015 when OpenAI launched with Amazon listed as a donor. (caselaw.findlaw.com) ### What did the emails say? The bluntest line came from Microsoft CTO Kevin Scott, who worried OpenAI could “storm off to Amazon in a huff.” Other reporting on the same exhibits says Microsoft also feared OpenAI could badmouth Azure on the way out. Nadella’s own messages were skeptical in a different way — he said he couldn’t really tell what research OpenAI was doing or how it would help Microsoft get ahead, even while noting that Musk was talking up possible AGI breakthroughs. (geekwire.com) ### So was Microsoft a believer or a skeptic? Both. That’s the interesting part. The emails suggest Microsoft was not fully sold on near-term AGI claims in 2018, but it still saw OpenAI as strategically dangerous to ignore. Basically, Microsoft didn’t need certainty that OpenAI would win. It only needed to believe that losing OpenAI to Amazon would be worse than subsidizing the relationship. (techxplore.com) ### Why does Nadella’s testimony matter? Because it pulls the story up to CEO level. Nadella is expected to explain why Microsoft gave OpenAI Azure discounts and how the company thought about funding the lab before the 2019 investment. His appearance comes near the end of a closely watched trial, with Sam Altman expected after him and an advisory jury opinion anticipated the week of May 18. (techxplore.com) ### What is Musk trying to prove with this? Musk’s side wants the jury and judge to see Microsoft’s money not as neutral infrastructure support, but as part of OpenAI’s turn from nonprofit lab to commercial powerhouse. OpenAI’s side says Musk understood years ago that a for-profit structure was necessary and only turned hostile after losing influence and launching rival xAI. That fight is why these old cloud emails suddenly matter so much. (techxplore.com) ### Why should anyone outside the lawsuit care? Because the documents make a hidden dependency visible. Frontier AI labs talk about models, safety, and governance, but underneath all that sits compute. If your product depends on one model provider and one cloud partner, your leverage can disappear fast when the relationship sours. Microsoft’s old fear about OpenAI jumping to AWS is really a reminder that infrastructure “partnerships” are often contested power arrangements. (techxplore.com) ### Bottom line? The revealing part isn’t that Microsoft loved OpenAI from day one. Turns out it didn’t. The revealing part is that Microsoft feared being left out more than it feared being wrong — and that may explain the entire modern AI stack. (techxplore.com) (geekwire.com)