Amazon Is Gutting Its AI Division After Sustained Failure

333 points · 262 comments on HN · read original →

Points and comments are a snapshot, not live.

Amazon is gutting its AI division after its Nova models failed.

Amazon is scaling back most of its Nova AI models and reorganizing AI teams after its large language models failed. The company has put text, video, and image-generating Nova models into a "keep the lights on" state with minimal resources. Resources are being diverted to a "frontier-model effort" led by Pieter Abbeel, focused on AI for robotics. This comes after Amazon closed an 80-person AI office in San Francisco.

What commenters are saying

Commenters note Amazon's culture is ill-suited for frontier AI innovation, citing high turnover, data access restrictions, and a risk-averse approach that contrasts with competitors. Several describe Amazon's compensation structure incentivizing departure after two years. One commenter contrasts AWS's strict data security and compliance culture with Meta's disregard for privacy, suggesting this makes Amazon struggle in AI. Another notes Amazon's reputation as a place "nobody with a choice will work."