Amodei says AI centralizes by nature; open models just shift power to chip owners

The dispute between Amodei and his critics reveals a fundamental tension in AI governance that has split even technologists and business leaders. On one side stands Amodei's argument that concentrating AI among regulated companies with strong institutional oversight can prevent the accumulation of power in individual hands. On the other side stand open-AI advocates who believe that wide distribution of AI systems—mimicking the model of the printing press or the internet—creates a checks-and-balances system where diverse actors with different values prevent any single point of control. Amodei's critics argue that his regulatory proposals, combined with Anthropic's hiring of former government officials, suggest the company is not trying to solve the concentration problem but rather to secure a privileged regulatory position for itself. Sacks pointed out that Amodei did not actually dispute Baker's original claim that Amodei had told people internally that Anthropic could one day be the only private firm left in the world—a silence that, in Sacks's reading, speaks volumes. The fact that almost every major company except Anthropic has backed Jensen Huang's letter in favor of open models suggests that the broader industry may be rejecting Amodei's framing, even if his argument about the structural concentration of AI power through computing resources has a technical grain of truth.