What does OpenAI inside Google's cloud really change?
Let's rewind a few months, because this is a story about a door that used to be locked.
For years OpenAI's most powerful models were sold to businesses through one cloud, Microsoft's, because the contract said so. Then on 27 April that exclusivity ended. Microsoft kept a licence to OpenAI's technology through 2032 and stayed its main cloud partner, but OpenAI was free to sell elsewhere. The very next day its models and its coding agent appeared on Amazon's platform. And on 21 October, OpenAI announced its models are coming to Google Cloud too.
So the same family of models is now on the shelves of all three giant clouds.
Wait. Why would Google sell a competitor's model right next to its own Gemini?
Because Google's cloud customers asked for it, and because a cloud earns its money on the computing either way. If a company wants to build on OpenAI, Google would much rather host that than lose the customer altogether. The model is the product the customer wants. The cloud is the shop.
What actually moves
When the same model is available everywhere, the model alone stops being the reason to choose a cloud. The fight shifts to everything around it: the tools, the price, how well it plugs into the rest of your systems, how much you trust the provider with your data.
For an ordinary company that is good news. You can pick the cloud you already use and still reach the model you want. For the cloud giants it is a quieter, more brutal competition. They are not selling you a brain any more. They are competing to be the best place to rent one.