A Stanford Lecture Explains Why AI Value Gets Trapped In Chips

The traditional software playbook promised that more users meant better margins. Generative AI breaks that rule entirely; every new user requires burning expensive GPU compute, which means the application layer is effectively subsidizing a semiconductor monopoly that collects the lion's share of AI's explosive revenue growth.