Why Microsoft is telling enterprises to ditch OpenAI and Anthropic
microsoft's satya nadella is pivoting to pitch homegrown models and agents to enterprises to avoid vendor lock-in and data leaks, threatening its partners' revenue.
following a profitable quarter with $90b in revenue, microsoft is openly competing with its invested partners, openai and anthropic, by pitching its own ma-1 models and copilot agents as cheaper alternatives. nadella argues that enterprises sharing data with frontier labs face security risks and lock-in, citing the hugging face sandbox breach where a model refused to help analyze logs. he is urging it leaders to decouple the application harness from the model to allow for swappable ai infrastructure.