Some players not managing risk in AI infrastructure race: Anthropic CEO Dario Amodei

Anthropic CEO Dario Amodei warned that as the competition to secure artificial intelligence (AI) infrastructure intensifies, “some players are not managing that risk well; they’re taking unwise risks". He did not name any companies.

Speaking at the New York Times Dealbook Summit, Amodei, who founded Anthropic in 2023, said, “Whenever there’s uncertainty, there is a risk of overextension and some (companies) are turning the dial too far.”

Amodei said both investors and AI companies are in a dilemma due to the speed of the technology. “There’s real uncertainty in how quickly the economic value will grow, and a lag in building the data centres needed to support it,” he added. “It’s a genuine dilemma that we at Anthropic try to manage responsibly, but some players are YOLO-ing (taking on too much risk), and that concerns me.”


Also Read: Anthropic plans an IPO as early as 2026

Amodei said that he and some of his cofounders were among the first to document the scaling laws of AI. “When you increase compute and data with small modifications, models improve dramatically,” he said. “They get better at coding, science, biomedicine, law, finance, materials, and manufacturing, essentially all the sources of value in the economy.”

That has translated into high business growth for the company. “Our revenue has grown tenfold each year for the last three years: from zero to $100 million in 2023, from $100 million to $1 billion in 2024, and it’s on track to land somewhere between $8 billion and $10 billion by the end of this year,” Amodei said.

In a fireside chat, Amodei described a “cone of uncertainty” around future growth, where revenue projections for the next year could range from $20 billion to $50 billion. “We plan conservatively for the lower side,” he said.