May 7, 2024
Most OpenAI customers only use OpenAI. But almost everyone who uses Anthropic also uses OpenAI
From Pilot’s data as of March 2024, 87% of OpenAI customers used them exclusively. By contrast, only 17% of Anthropic customers only use Anthropic. And same with Cohere, where only 24% of them use it exclusively.
Is there something super-sticky about OpenAI that lends itself to exclusive use? Or is it just that they got there first?
Should we expect increased use of multiple foundational models at companies?
At Pilot, we use both OpenAI and Anthropic internally, using each company’s models for the tasks they’re especially well-suited-for—and my inclination is that other folks will do the same.