Interesting attention comes from the recent OpenAI funding deal completed at the end of February. A total of 110 billion dollars was invested with a valuation of 730 billion dollars, but more important than those numbers is how Microsoft and Amazon each position themselves.

Amazon invested 50 billion dollars, NVIDIA 30 billion, and SoftBank 30 billion. But if you pay attention to Sam Altman's announcement, he mentioned Amazon first before Microsoft — even though Microsoft was not part of this funding round. That is no coincidence. AI blogger Aakash Gupta points out details often overlooked: two technical terms Sam Altman mentioned, namely Stateless API and Stateful Runtime Environment, each representing the domain of Microsoft and Amazon.

Stateless API is the current model — one query, one answer, done. Once the request cycle ends, the system does not store context. This is used by the finance, retail, and manufacturing industries to plug AI into their existing systems. Easy to integrate, minimal disruption. But the problem is, as model capabilities converge and computing costs decrease, Stateless API charged per token will become a commodity with margins that continue to erode.

Stateful Runtime Environment is different — it is a continuous execution environment. Agents have memory, can last long, work across tasks, and perform long-term jobs. Its scale is still small now, but this is not just feature optimization. This is a paradigm shift. Not just answering questions, but functioning as a concrete digital workforce executing tasks. This means the budget impact expands from API call costs to automation, process management, and even some labor costs. Market expectations for this are much higher than the current scale.

Microsoft secures Stateless API with a 250 billion dollar deal and exclusive rights. Every OpenAI Stateless API call will be hosted on Azure — regardless of the customer, all traffic flows back to Azure. Cash flow is certain, but there is a risk of margin squeeze ahead. Amazon, on the other hand, uses a 50 billion dollar investment plus a 100 billion dollar expansion agreement to secure its position in the AI Agent era. When Agents become the main productivity engine, resources consumed — compute power, storage, scheduling, workflow orchestration — will accumulate in the AWS environment.

One controls cash flow now. The other speculates on future productivity structures. What’s interesting is OpenAI’s own strategy — by balancing preferences between Microsoft and Amazon, OpenAI is not tied to a single cloud provider. This is a classic distributed bet strategy. Previously, OpenAI was highly dependent on Microsoft infrastructure with a 27% shareholding. Microsoft is not only an investor but also an infrastructure controller. Now, with Amazon entering strongly, both parties will compete directly for future service rights. For OpenAI, this means bargaining power returns to their hands. No one can walk away from the table now. The clear boundary and strict separation of interests in the deal structure fundamentally change the power dynamics. The more both parties cannot detach from OpenAI, the more OpenAI can negotiate better terms in the future.