OpenAI breaks free of Azure: Microsoft rewrites its contract with its key partner

Microsoft and OpenAI have rewritten the terms of their multiyear partnership, and did so with surprising calm. The main change: OpenAI can now sell its models and services through any cloud platform, not just Azure.

How the deal changed

The Microsoft–OpenAI partnership dates back to 2019. Over the years Microsoft has invested, by various estimates, more than $13 billion in the company, and in return received exclusive access to its technology. The key condition: OpenAI's flagship models were distributed to enterprise customers exclusively through Azure. If a company wanted to work with GPT-4, it had to come to Azure. Neither Google Cloud, nor AWS, nor any other provider could offer these models directly.

The updated agreement removes that restriction. The day after the announcement, OpenAI signed a deal with AWS, a pointed signal to the market that the rules of the game have changed.

What is known about the terms of the new agreement:

• OpenAI can sell its API and products through any provider, including AWS, Google Cloud, Oracle and others
• Microsoft keeps the right to use OpenAI technology in its own products: Copilot, Office 365, Windows
• Microsoft gets "priority partner" status, an advantage when all other conditions are equal
• Financial terms and investment stakes have been revised in favor of greater independence for OpenAI
• Infrastructure commitments have been reworked to account for the Stargate project

Why the split was peaceful

The relationship between Microsoft and OpenAI has been marked by serious crises. The most high-profile came in November 2023, when the board fired CEO Sam Altman and employees en masse threatened to follow him to Microsoft. The crisis exposed deep governance conflicts and disagreements with the largest investor. On top of that came disputes over the pace of data center construction, disagreements over control of monetization, and competition for computing capacity.

Many observers expected a painful break: lawsuits, public accusations, leaks. But the negotiations ended quietly. Both sides got what they needed. Over five years Microsoft has managed to embed AI technology deeply across its product portfolio, from Windows Copilot to Azure AI Foundry. Its dependence on the exclusive contract has dropped significantly. OpenAI, for its part, has finally gained the operational freedom it had long lacked.

"This partnership has evolved along with the industry,"

Microsoft said in a comment.

What OpenAI gains

For OpenAI, the new status is more than a commercial maneuver. The company is in the middle of a sweeping transformation: a shift from a nonprofit structure to a for-profit one, preparation for a possible IPO, and the Stargate project with SoftBank and Oracle worth $500 billion. All of this demands full operational independence and the freedom to choose infrastructure.

Tying itself to a single cloud provider created a real barrier in negotiations with enterprise customers. Companies that wanted to use GPT-4o or o1 in production were effectively obliged to accept Azure. Now OpenAI can compete for enterprise budgets directly, and the AWS deal already shows how that will work in practice.

What this means

Azure's monopoly on access to OpenAI's leading models is over. Competition among cloud providers for AI workloads is becoming truly open: Google Cloud and AWS have a real chance of joining the ranks of OpenAI's key partners. For businesses, the message is clear: the choice of AI infrastructure is no longer dictated by a single exclusive board-level contract.