Anthropic throws open doors at $2T valuation as shares pop 34% in initial trading

Welcome to the world of public AI frontier labs: Anthropic is now a publicly-held company with its Thursday IPO, and investors are clamoring for a piece of the pie, for now anyway.

That $83 IPO price didn’t last long, with shares rocketing to $111.50 by the time the public could get their hands on them at the end of the initial auction. That means that the company was valued at around $2 trillion as of the public availability of the 980 million shares it put up for sale, a 34 percent increase over the IPO price. 

At just under six years old, Anthropic’s meteoric rise to that valuation has sent investors into a frenzy despite uncertainties surrounding the company’s future profitability. As we’ve noted, Anthropic is on the hook for nearly $1 trillion in compute obligations that could drag its profitability prospects down if any one of a number of scenarios happens to materialize. 

Amodei was on hand to address some of those questions with CNBC this morning following the opening of markets on his company’s big day. 

When asked about the risk Anthropic’s massive compute bills to its bottom line, Amodei admitted the company’s needs are difficult to predict, and that - get this - they try to err on the low side. 

That’s right: The $900 billion in infrastructure and compute commitments that Anthropic revealed in a recent regulatory filing, which could end up surpassing $1T in short order, might just be an underestimate of what the company will need in coming months and years. 

Not to worry, though, the Anthropic CEO insisted: It’s worth it. 

“We’re willing to put into place these very large compute spends because, historically, we’ve seen that it pays off,” Amodei told CNBC’s Ross Sorkin, adding that the returns on the cognitive capabilities of its new models are “so high” as to make all that spending worth it - eventually, one would assume based on his comments, Anthropic is sure to come out on top. 

But what of the threat of open models that could undermine the constant push at Anthropic and other frontier labs to create the most power-hungry, resource-intensive, and powerful AI? Bah, that’s not a problem either. 

Frontier model makers like Anthropic, Amodei said, have already proven that they can do the heavy lifting at the bleeding edge, so you can trust it to make reliable models that are cheaper and less capable, too. 

“Anthropic can compete both at and away from the frontier,” The Anthropic CEO said. 

Whether that’ll mean long-term success for investors is another thing altogether. Mozilla president Mark Surman doesn’t think valuation translates into long-term success, for one. He told this morning that the one thing he consistently hears from developers is that they want AI they can control, not rent from a frontier lab. As Amodei suggested in his comments, Anthropic is aware that those with less intense AI needs might not choose its models, and that could harm the company. 

“Pushing AI toward open source matters so much right now,” Surman said in an email. “Mozilla’s own research this year found that open source models account for a third of all tokens, while the gap with the best closed models is down to about 3.3 per cent.”

In other words, keep an eye on that Anthropic stock - it might be hot now, but reality could still catch up to the company with the biggest-ever IPO.

“The market is already moving,” Surman said. “No valuation changes that.” ®

Original source Anthropic throws open doors at $2T valuation as shares pop 34% in initial trading

Back to home