OpenAI is sprinting toward Amazon Web Services with a velocity that has caught even its own executives off guard — and in the process, it is laying bare just how constraining its foundational partnership with Microsoft has become.
In a conversation with GeekWire’s Todd Bishop, OpenAI Chief Financial Officer Sarah Friar described demand for the company’s models on AWS as “staggering,” a word that carries particular weight given the financial magnitude involved. The partnership with Amazon, first announced in late 2025, is already generating what Friar characterized as a $100 billion pipeline of enterprise demand. That figure, if it holds, would dwarf the revenue expectations most analysts had modeled for OpenAI’s near-term cloud distribution strategy.
The number alone is remarkable. But what Friar said next was more revealing.
“We’ve been somewhat held back by the exclusivity that we had with Microsoft,” Friar told GeekWire, speaking at the company’s San Francisco headquarters. It was a rare, unvarnished admission from a C-suite executive about a partnership that has defined OpenAI’s commercial identity since Microsoft first invested $1 billion in 2019. The relationship has since expanded to roughly $13 billion in total Microsoft investment, making it one of the largest corporate bets in technology history. And yet here was OpenAI’s finance chief essentially saying the arrangement had become a ceiling, not just a floor.
Friar’s comments landed during a period of extraordinary financial acceleration for OpenAI. The company has said it expects to reach $12.7 billion in annualized revenue, a figure that has roughly tripled in the span of a year. Its consumer products — ChatGPT chief among them — continue to grow. But the enterprise cloud market is where the truly enormous dollars live, and OpenAI’s ability to capture them has, until recently, been structurally limited.
Under the original terms of its deal with Microsoft, OpenAI’s models were available exclusively through Microsoft Azure for enterprise cloud customers. That arrangement gave Microsoft a powerful competitive weapon against AWS and Google Cloud Platform, both of which have their own AI model offerings. For OpenAI, the deal provided critical infrastructure and capital at a time when the company desperately needed both. The trade-off was reach. AWS commands roughly 31% of the global cloud infrastructure market, according to Synergy Research Group estimates. Azure holds about 25%. By locking itself to Azure, OpenAI was voluntarily forgoing access to the largest slice of enterprise cloud spending on the planet.
The renegotiated partnership, finalized earlier this year, changed the calculus. OpenAI secured the ability to sell its models through AWS and other cloud providers, while Microsoft retained certain preferential terms and continued as OpenAI’s primary infrastructure partner. It was a deal that reflected a new reality: OpenAI had grown too large and too ambitious to remain a single-cloud company.
Amazon moved quickly. AWS began offering OpenAI’s models through its Bedrock platform, the managed AI service that enterprise customers use to deploy foundation models. The response from AWS customers was immediate and, by Friar’s account, overwhelming. Enterprise buyers who had been running their operations on AWS for years — and had no intention of migrating to Azure — could finally access GPT-4o, o3, and other OpenAI models without switching providers. The friction vanished, and demand surged.
“The pipeline is about $100 billion,” Friar said, referring specifically to prospective enterprise contracts flowing through the AWS channel. She acknowledged the figure represents potential rather than booked revenue, but emphasized that conversion rates have been strong. “We’re seeing enterprises that were waiting for this exact moment.”
That waiting game is itself an indictment of how the exclusivity arrangement functioned in practice. Large enterprises are notoriously reluctant to adopt multi-cloud strategies purely to access a single vendor’s AI models. The switching costs are enormous — not just in dollars but in engineering time, security certification, and operational complexity. Many CIOs simply refused to move workloads to Azure for the sole purpose of using OpenAI. They waited. Now they don’t have to.
Microsoft, for its part, has publicly maintained that its relationship with OpenAI remains strong and mutually beneficial. Satya Nadella has repeatedly framed the partnership as the cornerstone of Microsoft’s AI strategy, and Azure’s own AI revenue growth has been staggering — the company reported that its AI business has surpassed a $13 billion annual revenue run rate. But the subtext of Friar’s comments is hard to ignore. If OpenAI’s CFO is willing to say on the record that the Microsoft exclusivity “held back” the company, the internal frustration must run deeper than the diplomatic language suggests.
The timing matters. OpenAI is preparing for what would be one of the largest private-to-public transitions in Silicon Valley history. The company recently closed a funding round that valued it at $300 billion, and it is expected to convert from its unusual capped-profit structure to a more traditional corporate form. Every dollar of revenue growth, every new enterprise contract, every expanded distribution channel feeds directly into the valuation narrative that will underpin an eventual IPO. In that context, a $100 billion AWS pipeline isn’t just a business milestone. It’s a valuation catalyst.
Friar, who joined OpenAI in 2024 after serving as CEO of Nextdoor and CFO of Square, has been instrumental in professionalizing the company’s financial operations. Her willingness to speak candidly about the Microsoft relationship reflects a broader strategic confidence — a sense that OpenAI no longer needs to defer to its largest investor on questions of distribution. The power dynamics have shifted. Microsoft still provides critical compute infrastructure, and its Azure credits remain a significant economic subsidy. But OpenAI now has alternatives, and it’s using them aggressively.
Amazon’s motivations are equally transparent. AWS has been engaged in a fierce competition with Microsoft and Google to become the dominant platform for enterprise AI deployment. Its homegrown models — developed by Anthropic, in which Amazon has invested up to $4 billion, and its own internal teams — have gained traction, but none carry the brand recognition or developer mindshare of OpenAI’s GPT series. By bringing OpenAI onto Bedrock, AWS instantly neutralized one of Azure’s most potent differentiators.
The competitive implications ripple outward. Google Cloud, which offers its own Gemini models and hosts third-party models through Vertex AI, now faces a market where both of its major rivals carry OpenAI. Anthropic, which had positioned itself as the premium alternative to OpenAI for enterprise customers — particularly those on AWS — must now compete with OpenAI on its home turf. And Microsoft, which spent over a decade and billions of dollars cultivating its OpenAI advantage, must reckon with the erosion of that exclusivity.
None of this means the Microsoft-OpenAI relationship is falling apart. The financial entanglements are too deep, the infrastructure dependencies too significant, the contractual obligations too binding for a clean break. Microsoft’s investment gives it a substantial share of OpenAI’s profits (or the equivalent under the company’s evolving corporate structure), and Azure remains the primary compute backbone for training OpenAI’s largest models. But the relationship is clearly evolving from a quasi-exclusive partnership into something more transactional. More arm’s length.
For enterprise CIOs, the practical implications are significant. The availability of OpenAI models on AWS means that organizations can now build AI applications using the industry’s most widely adopted models on the industry’s most widely adopted cloud — without compromise. Integration with existing AWS services like SageMaker, Lambda, and S3 becomes straightforward. Security and compliance teams don’t need to certify a new cloud provider. Procurement doesn’t need to negotiate a new master services agreement. The barriers that kept OpenAI out of AWS-centric enterprises have simply been removed.
Friar told GeekWire that OpenAI is also investing heavily in its own direct enterprise sales operation, separate from its cloud partnerships. The company has hired aggressively in go-to-market functions over the past year and now employs a dedicated enterprise sales force targeting Fortune 500 companies. The multi-channel approach — direct sales plus Azure plus AWS plus potentially other cloud partners — gives OpenAI a distribution surface that no other AI model provider can currently match.
The revenue math tells the story. If even a fraction of that $100 billion AWS pipeline converts to actual contracts over the next two to three years, the contribution to OpenAI’s top line would be transformative. The company’s current revenue, while growing rapidly, is still heavily concentrated in consumer subscriptions and API usage through Azure. Diversifying that revenue base across multiple cloud channels reduces concentration risk, increases pricing leverage, and strengthens the narrative for public market investors.
So where does this leave Microsoft? In an uncomfortable but not untenable position. Azure still has deep integration with OpenAI models, and Microsoft’s Copilot products — built on top of those models — continue to be a major growth driver. The company retains economic exposure to OpenAI’s success regardless of which cloud delivers the models. And Microsoft’s own AI infrastructure investments, including custom silicon and massive data center buildouts, give it a durable role in the value chain even as OpenAI distributes more broadly.
But the psychological blow is real. For three years, Microsoft could credibly claim that the most important AI company in the world was effectively a captive partner. That claim no longer holds. And in the cloud wars, where perception shapes enterprise buying decisions as much as technology does, the loss of that narrative advantage matters.
OpenAI’s Friar was careful to praise Microsoft during her conversation with GeekWire, noting the critical role the company played in OpenAI’s growth and emphasizing that the partnership remains “incredibly important.” The diplomatic language was intact. But the strategic message was unmistakable: OpenAI has outgrown its original arrangement, and it intends to sell its models to anyone who will buy them, on whatever cloud they prefer.
The $100 billion pipeline on AWS is the proof point. Whether it fully materializes or not, the signal it sends is clear — OpenAI is no longer Microsoft’s AI. It’s its own.
WebProNews is an iEntry Publication