• How We Got Here: Microsoft’s Wall Comes Down
• Tuesday: AWS Moves In — Fast
• Behind the Curtain: The Money and the Megawatts
• The New Pact: OpenAI on AWS, On AWS’s Terms
• Agents, Coders, and the Battle for the Enterprise Desk
• Competing Narratives: Liberation, Leverage, or Lifeline?
• What Changes Next

Amazon Web Services to Offer OpenAI Models Following End of Microsoft Exclusivity AI AI sources depict the AWS–OpenAI partnership as a deepened technical integration that brings frontier models, Codex, and managed agents to AWS customers within existing security and workflow systems. They focus on how access to GPT‑5.5 and agentic tooling on AWS accelerates enterprise AI adoption and simplifies scaling from pilots to production. @OpenAI

Human Human sources frame AWS’s new OpenAI offerings as the end of Microsoft’s resale exclusivity and a reconfiguration of cloud AI competition, with Amazon quickly rolling out new models and agent services. They also stress OpenAI’s missed revenue and growth targets and its massive infrastructure commitments, casting the deal as both an opportunity and a potential financial strain. @TNW @TC Amazon’s AI cold war just went officially multipolar: after years of watching Microsoft guard OpenAI like a crown jewel, Amazon Web Services is now selling those same models to its own cloud customers, blowing up the most important exclusivity deal of the AI boom and raising sharp questions about OpenAI’s finances and future.

§How We Got Here: Microsoft’s Wall Comes Down

For three years, Microsoft’s Azure had something everyone else wanted: exclusive reselling rights to OpenAI’s models. That wall came down this week.

On Monday, Microsoft and OpenAI rewrote their partnership. The once-exclusive license to OpenAI’s intellectual property became non‑exclusive, but was extended through 2032, dismantling the arrangement that had “defined the first phase of the AI boom.”¹

The change wasn’t just legal fine print. It set off a chain reaction across the cloud world.

§Tuesday: AWS Moves In — Fast

Within 24 hours, Amazon pounced. On Tuesday, Amazon Web Services announced it would begin selling OpenAI models directly to its cloud customers — ending Azure’s status as the only hyperscale gateway to OpenAI’s technology.¹

“Amazon Web Services will begin selling OpenAI’s models to its cloud customers, the company announced on Tuesday, one day after Microsoft agreed to end the exclusive reselling arrangement that had given Azure sole access to OpenAI’s technology for the first three years of the generative AI era.”¹

AWS CEO Matt Garman made clear this wasn’t a science project, but pent‑up demand finally being met. “It’s something that our customers have asked for, for a really long time,” he told Bloomberg Television, underscoring just how much pressure AWS had been under to match Azure’s OpenAI offering.¹

Some of OpenAI’s latest models are available in preview on AWS starting this week, with the “most powerful GPT models” promised within weeks — a rapid rollout that shows both how ready AWS was and how eager OpenAI is to monetize across clouds.¹

TechCrunch captured the moment bluntly: “Amazon is already offering new OpenAI products on AWS,” framing this as the start of a new, less Microsoft‑centric chapter for enterprise AI.²

§Behind the Curtain: The Money and the Megawatts

This isn’t just a product launch; it’s the visible tip of a massive financial and infrastructure restructuring.

According to reporting summarized by The Next Web, the new AWS‑OpenAI distribution deal “completes a restructuring that began in February when Amazon committed up to $50 billion as part of OpenAI’s $110 billion funding round, a deal that valued the ChatGPT maker at $852 billion and gave Amazon its largest‑ever investment in any company.”¹

In exchange, OpenAI made an equally colossal promise: it “committed to spending $100 billion on AWS computing power and Trainium chips over eight years, consuming two gigawatts of capacity.”¹ Two gigawatts is the sort of number you normally see in national energy planning, not in a single software contract.

Layered on top are additional long‑term infrastructure commitments not just to AWS, but also to Microsoft’s Azure and Oracle — “hundreds of billions” in total that “assume growth OpenAI has not yet demonstrated.”¹

That’s where the tension bites. The Wall Street Journal, via The Next Web’s summary, reports that OpenAI has missed both revenue and user targets, even as it stares down an expected $25 billion in cash burn against $30 billion in revenue.¹ The math works only if usage surges, everywhere, and quickly.

“The question the deal answers is not whether OpenAI’s models are good enough to sell on rival clouds,” The Next Web notes. “The question is whether OpenAI can sell enough of them, anywhere, to justify what it has promised to spend.”¹

§The New Pact: OpenAI on AWS, On AWS’s Terms

From OpenAI’s side, the messaging is less about financial fire drills and more about enterprise pragmatism. In an announcement titled “OpenAI models, Codex, and Managed Agents come to AWS,” the company described a deepened “strategic partnership to help enterprises build using OpenAI capabilities in their AWS environments.”³

“Today, OpenAI and AWS are expanding our strategic partnership to help enterprises build using OpenAI capabilities in their AWS environments,” the company wrote. “We’re excited to give AWS customers access to the best frontier models, agents, and tools, which will operate within the systems, security protocols, compliance requirements, and workflows they already use.”³

OpenAI divides the expansion into three pillars, “all launching today in limited preview”:³

1. OpenAI models on AWS
2. Codex on AWS
3. Amazon Bedrock Managed Agents, powered by OpenAI

The marquee move: “OpenAI models, including our best frontier model GPT‑5.5, [are launching] on Amazon Bedrock,” giving customers the ability to “build with OpenAI models in AWS, alongside the services, security controls, identity systems, and procurement processes they already rely on.”³

For developers, OpenAI pitches this as more “flexibility in how they build,” from greenfield AI apps to “agentic workflows that can reason, take action, and support more complex business processes.” For enterprises, it promises a “clear single path from experimentation to production” inside their existing AWS setups — a not‑so‑subtle dig at the friction of stitching together tools from rival clouds.³

§Agents, Coders, and the Battle for the Enterprise Desk

If Microsoft’s first‑phase advantage was mostly about who could access GPT‑4 in the cloud, the second phase is about what sits on top: agents, coding copilots, and workflow automation.

OpenAI emphasizes that “more than 4 million people now use Codex every week,” leveraging it “across the software development lifecycle — to write code, explain systems, refactor applications, generate tests, modernize legacy codebases, and accelerate a broader set of professional workflows that extend beyond coding.”³

Increasingly, Codex is also being used to “accelerate research, analysis, and document‑based work” by tying into everyday apps: “from summarizing source materials to creating briefs, slide decks, and spreadsheets.”³

Now all of that comes dressed in AWS colors, via Bedrock and “Managed Agents, powered by OpenAI.”³ Strategically, this is where Amazon is trying to catch — and perhaps surpass — Microsoft: by offering not just raw models, but tightly integrated, AWS‑native agents that sit inside the governance, identity, and billing systems enterprise CIOs already understand.

In other words, Microsoft’s exclusive on OpenAI is gone, but the real contest has moved up the stack.

§Competing Narratives: Liberation, Leverage, or Lifeline?

Different players are reading this week’s events through very different lenses.

• AWS’s story is one of long‑overdue liberation. Customers have “asked for [OpenAI models] for a really long time,” Garman says, and now they can finally get them without switching clouds.¹ The TechCrunch framing — that “Amazon is already offering new OpenAI products on AWS” — underscores the sense of Amazon catching up and moving quickly.²

• OpenAI’s public story centers on convenience and compliance. Its post focuses on “systems, security protocols, compliance requirements, and workflows [enterprises] already use,” and on giving “organizations more ways to use OpenAI across application development, software engineering, and agentic workflows.”³ The message: this isn’t desperation, it’s distribution.

• The financial narrative, however, is harsher. The Next Web’s reporting, leaning on the Wall Street Journal, stresses the gap between OpenAI’s gigantic infrastructure obligations and its lagging revenue and user metrics. With “$25 billion in expected cash burn against $30 billion revenue” and multi‑cloud commitments in the “hundreds of billions,” OpenAI suddenly needs every cloud, every customer, and every workload it can get.¹

From that angle, ending Microsoft’s exclusivity looks less like a philosophical embrace of openness and more like a forced march toward monetization.

§What Changes Next

In the near term, enterprise buyers win. They can now run OpenAI’s latest GPT models — including GPT‑5.5 — on AWS, within their existing security and procurement frameworks, without re‑architecting around Azure.³ Multi‑cloud AI strategies that were once theoretical just became operational.

For Microsoft, the trade‑off is subtler. It loses the bragging rights of sole distributor but gains a partner better positioned to scale revenue fast enough to pay for the gargantuan GPUs and data centers it is also building out. A healthier OpenAI — one that can “sell enough” of its models “anywhere” — is arguably in Microsoft’s interest too.¹

For Amazon, this is both offensive and defensive. Offensive, because it can now plug OpenAI into Bedrock alongside its own and third‑party models, strengthening AWS as a neutral, “bring your own model” platform. Defensive, because its $50 billion bet and $100 billion compute contract with OpenAI only pencil out if OpenAI actually runs — and sells — a lot of workloads on AWS.¹

And for OpenAI, the next phase is clear: the era of a single cloud patron is over. Survival now depends on turning exclusivity into ubiquity fast enough to keep those two‑gigawatt data centers humming.

§1. AWS to sell OpenAI models after Microsoft drops exclusivity, as OpenAI misses revenue targets and faces $100B infrastructure commitments — Details on Microsoft ending exclusivity, AWS starting to sell OpenAI models, and OpenAI’s massive funding, spending, and infrastructure commitments.

§2. Amazon is already offering new OpenAI products on AWS — TechCrunch headline framing AWS’s rapid rollout of OpenAI products.

§3. OpenAI models, Codex, and Managed Agents come to AWS — OpenAI’s announcement of its expanded strategic partnership with AWS, launching models (including GPT‑5.5), Codex, and Managed Agents on Amazon Bedrock.

Story coverage