SAN FRANCISCO: On Sunday, April 13, OpenAI’s chief revenue officer Denise Dresser sent an internal memo to staff acknowledging something the company had never said publicly: its partnership with Microsoft, OpenAI’s biggest financial backer, had spent years limiting who OpenAI’s own sales team was allowed to pursue. The memo, obtained by CNBC, reframes what the company’s new $50 billion Amazon deal actually fixes.
How Microsoft’s Cloud Exclusivity Capped OpenAI’s Enterprise Sales
Dresser wrote that the Microsoft partnership “has been foundational to our success. But it has also limited our ability to meet enterprises where they are,” adding that for many of those enterprises, “that’s Bedrock.” Amazon Bedrock is AWS’s managed AI platform that lets enterprise clients run large language models, including OpenAI’s GPT-4o and o3, without routing through Azure. Under OpenAI’s prior arrangement with Microsoft, its sales team was effectively blocked from pursuing companies that run on AWS infrastructure directly.
Microsoft has invested more than $13 billion in OpenAI since 2019. For most of that period, Azure served as OpenAI’s exclusive cloud infrastructure partner. That exclusivity helped OpenAI scale, but it created a ceiling: roughly 30% of the enterprise cloud market runs on AWS, not Azure, and OpenAI’s commercial team couldn’t go after it. Amazon’s investment, announced in late February, removed that restriction. Dresser told staff that inbound customer demand since the announcement “has been frankly staggering.”
The relationship between Microsoft and OpenAI had already been showing strain before Sunday’s memo. In its fiscal year 2024 annual report, Microsoft listed OpenAI among its competitors as the two companies increasingly build overlapping products across productivity software and AI assistants. Dresser joined OpenAI in December 2025, having previously served as Slack’s CEO and in senior roles at Salesforce. She arrived at a company where enterprise clients account for 40% of total revenue and the mandate is to grow that fast.
The memo adds another chapter to what has been a turbulent stretch for OpenAI’s leadership and strategy. Earlier this year, a New Yorker investigation prompted questions about Sam Altman’s management style and company culture, covered by GJ in What the Sam Altman New Yorker Investigation Means for Founders. CFO Sarah Friar separately said the company is not ready for a 2026 IPO. The Amazon deal, and now the public acknowledgment that Microsoft constrained its sales reach, suggests OpenAI is working to close those gaps rather than paper over them.
What Does OpenAI’s Microsoft Admission Mean for Founders on AWS?
For founders building on AWS, this is the most direct commercial signal of the year. OpenAI’s sales team can now pursue your business for the first time. If you assumed Azure-native infrastructure carried purchasing advantages when buying GPT-4o or o3 access, that assumption no longer holds.
The stakes explain why Dresser went to staff with the announcement. OpenAI’s enterprise segment is at 40% of total revenue, and the company expects it to match consumer revenue, which includes ChatGPT subscriptions, by the end of 2026. That growth target requires reaching companies in AWS’s portion of the market. Previously, OpenAI could reach those companies only through its self-serve API. Now it can send a sales team.
In the same memo, Dresser went after Anthropic directly. She wrote that Anthropic’s strategy “is built on fear, restriction, and the idea that a small group of elites should control AI,” and claimed Anthropic made “a strategic misstep to not acquire enough compute,” according to reporting on the memo. That’s not a casual jab. Anthropic, valued at $380 billion in a February 2026 funding round, has built its enterprise business largely on a safety-first, compliance-friendly pitch. Dresser is now contesting that positioning by name.
What to Watch as OpenAI Pushes Into Multi-Cloud Territory
The memo does not frame the Microsoft relationship as broken. Dresser called it “foundational” even while acknowledging it constrained sales for years. Microsoft has not publicly commented. Whether it will renegotiate the exclusivity terms of its arrangement with OpenAI, and on what timeline, is a question neither company has addressed publicly.
On the product side, a separate internal OpenAI memo obtained by The Decoder referenced a model the company calls “Spud” that Dresser told staff will make its products “significantly better.” If that model ships to AWS Bedrock customers at the same time as Azure customers, it confirms that multi-cloud parity is OpenAI’s standing commercial policy. That matters for founders making infrastructure bets now, because the timing of model access could influence which cloud they build on.
OpenAI’s valuation crossed $850 billion in a March 2026 funding round, part of the broader record $297 billion in Q1 2026 venture capital that flowed largely into AI companies. The company now has commercial partnerships with both the world’s largest cloud provider and its second-largest. For founders choosing where to build their AI stack, the practical effect is that the model layer is decoupling from cloud vendor relationships. Picking AWS over Azure no longer means being outside OpenAI’s enterprise sales reach. That’s a real change from how things worked a quarter ago.
