OpenAI from Microsoft to Amazon turns aggressive after subtlety

OpenAI chief revenue officer Denise Dresser said the AI ​​company’s deal on Tuesday to make its models available Amazon it was unrelated to the announcement the previous day that the startup had renewed its relationship with it Microsoft the second time in six months.

“The two are not related in any way,” Dresser told CNBC in an interview after the OpenAI and Amazon announcement.

Analysts aren’t so sure.

A lot has happened since late October, when OpenAI completed a refinancing, giving Microsoft a 27% stake in the for-profit intelligence company. As part of that arrangement, OpenAI agreed to buy Azure services for an estimated $250 billion. And the funding agreement will last until OpenAI is certified by an independent panel to have achieved artificial general intelligence, or AGI.

Another big development since then is that OpenAI has been teaming up with Amazon, Microsoft’s biggest competitor in cloud infrastructure.

In November, OpenAI disclosed a $38 billion commitment with Amazon Web Services. And in late February, Amazon said it would invest $50 billion in OpenAI, which will use 2 gigawatts of AWS’s Trainium custom chips to train AI models.

Amazon and OpenAI also agreed at the time to jointly develop “customized models” for Amazon’s engineering teams to power their consumer products, and OpenAI’s commitment to spend on AWS was expanded by $100 billion.

“That was a big thing that was happening,” RBC Capital Markets analyst Rishi Jaluria, who recommends buying Microsoft shares, said in an interview.

This week’s one-two punch is the most striking sign yet of a major shift in the decade-long relationship between Microsoft and OpenAI.

It started in 2016, when OpenAI started doing its big tests on Azure. Three years later, Microsoft invested its first $1 billion in OpenAI, a number that would grow to $13 billion in several follow-up rounds.

But in 2024, Microsoft began calling OpenAI a competitor in its financial disclosures, and early last year the software giant lost its name as OpenAI’s exclusive cloud provider. In an internal memo earlier this month, Dresser wrote that OpenAI’s partnership with Microsoft “has been fundamental to our success,” but “has limited our ability to meet businesses where they are.”

Against that background, the latest deal between the two companies, “looks unchanged and for all we know it could change again in six months,” UBS analysts wrote in a note on Monday.

Other aspects of the agreement include the end of Microsoft’s exclusive license to OpenAI’s intellectual property and Microsoft’s royalty payments to OpenAI. Microsoft will also no longer be the only cloud provider of API products developed by third parties.

“While some changes seem inevitable, Microsoft appears to have made more compromises than benefits,” wrote analysts at UBS, which has a buy rating on Microsoft.

Amazon CEO Andy Jassy called Monday’s announcement “very interesting” in a post on X, adding that more details would be shared on Tuesday.

Hours later, his company jumped to announce a service for building AI agents with OpenAI models.

For years, developers interested in those models needed to go through Microsoft’s Azure cloud or work with OpenAI directly. Now companies with large AWS investments will be able to easily adopt models, while implementing volume deployment plans.

Dresser, speaking at the Amazon event, said that the restructuring of OpenAI and Microsoft is not encouraged by the growing cooperation with Amazon.

“Microsoft is our first partner,” he said. “They’re an amazing partner for us. They’re going to be a great partner as we go forward. Our focus is making sure, as we meet our customers where they are, that they have access to the places they work. And we want to make sure we’re bringing the best models to the best places for customers to succeed.”

The Financial Times reported that Microsoft was considering legal action over OpenAI’s plans with Amazon, and Microsoft told the newspaper that it “hoped that OpenAI understands and respects the importance of living up to its standards.” [its] legal obligation.” Microsoft did not provide comment beyond Monday’s announcement.

Microsoft is similarly making moves to distance itself from OpenAI.

In September, Microsoft said it was starting to draw on an AI model from Anthropic to answer specific questions in the 365 Copilot assistant for commercial customers. Two months later, Microsoft agreed to invest up to $5 billion in Anthropic, which committed to buy $30 billion of Azure computing.

Capitalizing on the growing popularity of Anthropic’s Claude Code, Microsoft announced in March an offering called Copilot Cowork in partnership with Anthropic.

Another aspect of Claude’s growing demand is that loyalty has become difficult. The company has reported minor or major outages during 37 of the past 90 days. Amazon, Anthropic’s first partner and investor, has taken notice.

Anthony Liguori, a vice president at AWS, said his team, which created the Bedrock service for working with AI models, switched to OpenAI’s Codex as its main development platform after relying on Claude Code and Amazon’s Kiro tool.

The reality of all the major parties involved is that they need each other.

The power is so lagging that OpenAI and Anthropic need to work with all major cloud vendors to secure as much computing as possible. And Microsoft and Amazon need easy access to all major models to serve their big customers.

So while Microsoft and OpenAI may have drifted apart, Jaluria was quick to note, “Microsoft still needs OpenAI, and OpenAI still needs Microsoft.”

WATCH: Private investors don’t believe OpenAI is worth what it’s pretending to be, says CFR’s Sebastian Mallaby.