I recently caught wind of some interesting comments from Anthropic's CEO, Dario Amodei, about the current state of the AI industry. He was speaking at The New York Times DealBook Summit, and while he didn't outright say we're in an AI bubble, he definitely raised some eyebrows about how some companies – and let's be honest, he was probably talking about OpenAI – are managing risk.

Amodei mentioned that while he’s optimistic about AI's potential, the timing of economic payoffs is uncertain. This creates a **dilemma** for companies. They need to invest in data centers and compute power to compete and serve customers. If they don't invest enough, they risk falling behind. If they invest too much, they could face financial ruin.

He even used the term "YOLO-ing," which, let's be real, is a pretty blunt way of saying some players are taking unnecessary risks. It seems that the CEO is concerned that some AI companies are not managing the risk well, taking what he considers unwise risks.

The core issue, as he sees it, is the uncertainty surrounding how quickly AI's economic value will grow compared to the long lead times required to build more data centers. He pointed out that AI chips depreciate in value as newer, faster, and cheaper options come out. This obsolescence can impact the industry's economics, especially if GPUs lose value faster than expected.

Conservative Planning vs. Risky Bets

Anthropic, according to Amodei, is taking a more conservative approach, planning for a more uncertain future. He highlighted Anthropic's impressive revenue growth, but also acknowledged that expecting that growth to continue indefinitely would be foolish.

Remember when OpenAI's CFO suggested the government should "backstop" their infrastructure loans? That didn't go over too well. Amodei seems to be suggesting that such moves are a sign of overextension and excessive risk-taking.

In the end, Amodei expressed confidence in Anthropic's ability to navigate the AI landscape, regardless of how things play out. However, he refrained from commenting on the situation of other companies in the AI industry.