OpenAI, Anthropic, Google, and xAI are no longer competing only on model quality. According to the source, the real battle is for compute capacity, in-house chips, cloud presence, API pricing, and the ecosystem for developers. In other words, the infrastructure around the model comes to the fore, not a single successful release. For teams and developers, this changes the selection criteria: they have to look not only at the model itself but also at the platform on which it runs. The cost of API access and the terms of working with the cloud are becoming just as important arguments as the quality of responses. It seems the race for models is gradually turning into a race for the platform.
Source: habr.com · post in Telegram