I found it very interesting how OpenAI closed this mega-round of $110 billion and no one is really talking about what matters. Everyone gets stuck on the astronomical number, but the crucial detail lies in two technical terms that the crowd overlooked: Stateless API and Stateful Runtime Environment.

To provide context: Amazon invested $50 billion ( with more $100 billion in future expansion), Microsoft didn't participate in this round but continues with its $250 billion agreement, NVIDIA put in $30 billion, and SoftBank added more $30 billion. The valuation reached $730 billion pre-investment.

But here’s the point that really makes sense: Microsoft basically locked in the current infrastructure with the Stateless API. Every time someone calls OpenAI’s stateless API, Azure bills in the background. It’s a predictable cash flow, but with an obvious problem — the margin on these calls tends to decrease over time as competition increases.

Meanwhile, Amazon made a different bet. It secured the right to build the Stateful Runtime Environment — basically an environment where AI agents are not just consulted once, but exist persistently, with memory, lasting context, and the ability to perform complex tasks over long periods. This isn’t just a functional optimization; it’s a paradigm shift. A stateful agent isn’t just an assistant that answers questions; it’s a digital workforce that can automate entire processes.

The difference is fundamental. Stateless API is the present — you ask a question, the AI responds, end. Stateful is the future — the agent keeps running, collaborating across tools, maintaining context. When this model becomes mainstream — and all signs point to 2026 or 2027 — real consumption won’t be just simple API calls but continuous computing, storage, workflow orchestration. It’s a much bigger cake.

What caught my attention is that Microsoft is controlling the current traffic engine, while Amazon is betting on the infrastructure that will dominate when agents become the central unit of enterprise productivity. One guarantees predictable revenue but potentially shrinking, the other bets on a market still forming.

For OpenAI, this is genius. It was somewhat tied to Microsoft before — 27% of shares, infrastructure controlled, unbalanced bargaining power. Now two giants are competing for its services. When neither can get off the table, power shifts back into the hands of those offering the product. OpenAI’s diversification strategy is paying dividends.