Amazon's andy jassy sees AI cost reduction as key to enterprise adoption

Amazon CEO Andy Jassy emphasized that AI inference costs remain too high to enable widespread adoption in enterprise software, stating they need to drop to about one-tenth of current levels. In an interview with a16z, Jassy noted that AI usage is still limited by cost, with demand far outpacing supply. Amazon plans to invest $220 billion in capital expenditures this year but still lacks sufficient infrastructure to meet AI demands. Jassy highlighted that 2026’s main compute capacity is fully booked, with most of 2027 and 2028 also reserved. He compared AI demand to a 'barbell,' with one end being AI labs and the other being enterprise clients using AI for specific tasks. Jassy argued that the middle market—enterprise software—has the potential for the largest AI growth but remains underdeveloped. He cited Amazon’s past development of the Graviton processor as a model for overcoming reliance on dominant chipmakers like Intel. Amazon’s Trainium chip, which offers 30-40% better cost efficiency than competitors, has secured over $225 billion in future revenue commitments, including deals with Anthropic and OpenAI.