Atlassian says it uses no Chinese AI models in production
Atlassian co-founder and CEO Mike Cannon-Brookes said on the 20VC podcast that the company runs no Chinese AI models in its products in production, instead using Nvidia's Nemotron models through what he called a deep partnership, along with models from Mistral and others, picking the best model per task as cheaply as possible. He said there is customer pushback in some places to "Chinese models" but not to open models, and argued that refusing inspectable, cheaper open technology will be hard for enterprises to sustain. He was also sceptical of companies claiming to have built their own AI models, saying most have fine-tuned a Chinese open-weight model for one job, and said AI running costs will fall further than he once thought.
Customer preferences about where models come from are becoming a real constraint on open-weight models in the enterprise market.