Anthropic CEO says the path ahead is unclear before record IPO

The risk of getting capacity wrong

At a meeting with JPMorgan Chase investors in New York on Friday, Amodei said it was unclear how rapid AI progress would translate into economic value. Yet decisions about buying capacity have to be made in advance.

Buying too little could leave Anthropic without the resources to serve customers and deploy new technology, hurting revenue. Buying too much, by contrast, could put excessive strain on the company’s finances. Amodei called purchasing computing capacity Anthropic’s “deepest business dilemma.”

According to Bloomberg, Anthropic’s commitments total nearly $901 billion and are mostly tied to computing capacity and chips. About $71 billion of that sum is due by September 2027. Bloomberg notes that this amount exceeds the market capitalization of roughly two-thirds of the companies in the S&P 500.

In February 2026, Amodei warned of a possible bankruptcy scenario: starting in late 2027, the company could buy $1 trillion worth of computing capacity but go bankrupt if revenue fell short. He said a mistake of just one year could lead to bankruptcy. Amodei doubted that some competitors had fully accounted for the risks or understood what they were taking on. Since then, the stakes have continued to rise.

Why Anthropic is going public

Amodei named two ways to reduce the risk: gross margin and cash on the balance sheet. He said Anthropic’s gross margin is fairly high compared with those of other market participants, and that the company already has a substantial cash reserve and plans to raise more.

Another way to reduce the risk is to buy a relatively small amount of capacity in advance and turn to the spot market if demand unexpectedly surges. Amodei compared this approach to insurance: it helps the company handle worse-than-expected results while leaving room to increase investment if things go well.

Uncertainty around capacity purchases was the main reason for going public, Amodei explained. The more money Anthropic has, the more room it will have if it ends up with too little computing capacity—or too much.

According to Bloomberg, Anthropic plans to raise up to $64.4 billion in its IPO and another $26 billion through a private placement. If the company raises those sums, the offering will be the largest in history. Media reports say demand is already about six times the amount on offer.

Anthropic’s annualized revenue run rate reached $81 billion at the end of October, up $10 billion from the end of September, according to company documents. Bloomberg links the growth to the popularity of AI products, including coding models.

Amodei spoke calmly about the IPO’s timing and said he was not fixated on a particular month or market conditions. Anthropic grew from nothing to its current scale in six years, he said. The most common piece of advice he received before going public—or the advice that stuck with him most—was not to let the company’s valuation exceed its actual worth. Amodei compared that principle with Anthropic’s internal approach: underpromise and overdeliver.

In the same conversation, Amodei said he hoped AI would help cure cancer and Alzheimer’s disease within five to fifteen years, but added that he did not know the precise timeline.

In Dataist’s view, the IPO should give Anthropic the financial room to make decisions about computing capacity well before revenue becomes clear. But even a large fundraising round will not resolve the central uncertainty: the company still does not know whether future revenue will justify its current commitments.