This episode of Exponential View explores the financial health and future risks of Anthropic following its S1 filing, alongside broader questions about the AI economy. Azeem Azhar and Noah Smith discuss whether traditional companies will adopt AI, the competitive threats from Google, OpenAI, and Chinese models, and the possibility of a single AI company dominating the economy. They also delve into the concept of 'super-science', where AI can predict and control complex systems beyond human understanding.
Summarized by Podsumo
Anthropic's S1 shows a healthy business with 80% of revenue from models less than six months old, but risks include reliance on continuous innovation and slow enterprise adoption.
The biggest competitive threats to Anthropic are Google's potential recursive self-improvement (RSI) advantage and OpenAI's aggressive spending, not just Chinese models.
A single AI company could theoretically capture 30% of every market by using its smartest AI to outcompete incumbents, raising unprecedented antitrust and power concerns.
AI is creating a 'super-science' where models can predict and control phenomena (like human language) that are too complex for humans to ever fully understand.
Traditional companies face a 'duration mismatch' - they may take five years to integrate AI, while bills come due sooner, potentially causing a crash.
"Anthropic's moat is just Claude being smart. If you have the smartest AI, it can just maybe tell you how to make a better insurance business than what other people are selling. So in a sense, it doesn't matter if the insurance companies buy your product, you can just kill them."
"We are now genie summoners. We will be able to use relationships that are too diffuse and complex for us to understand to do things. We will basically be along for the ride."
"Anthropic financially behaves like a mature company. OpenAI behaves like a startup. OpenAI could just borrow a shit ass ton of money."