The listing venue was settled over the weekend, and yesterday the numbers followed. The Financial Times reports that Anthropic told a small group of shareholders it will post an adjusted operating profit for this quarter. That would make two profitable quarters in a row.
The FT also cites a gross margin above 80 percent. Bloomberg picked the story up and set it against everything else in motion: Anthropic is heading for an IPO meant to match or beat SpaceX’s record, which was $86.3 billion earlier this year. The Nasdaq is locked in as the venue, and a $15 billion revolving credit facility is due to close this month. Annualized revenue is running above $65 billion, more than seven times the pace at the end of last year.
The two words
Adjusted means certain exceptional and one-off costs have been taken out. Which ones is not stated. That is normal for a company this close to an IPO, and it is also the point where you should stop nodding along.
The margin is more interesting. Those 80-plus percent apply, per the FT, before revenue-sharing payments to partners like Amazon and before the cost of training the models. Training is not a line item for a lab like Anthropic. It is the reason the company needs a $15 billion credit line in the first place. A gross margin that carves out exactly that block tells you something about the serving business. About the business as a whole, not much.
Why this one stuck with me
Two profitable quarters back to back is still a real number. In May the story was that Anthropic was “on track” for its first. Now the first is done and the second is announced, both ahead of the IPO. Anyone preparing an S-1 wants precisely that order of events.
What I notice is the sequence. On Saturday Dario Amodei called on the industry to slow down. Over the same weekend Sam Altman and Elon Musk lined up behind him. Then came the venue, and yesterday the profit. Altman said in the same week that going public would be ill-advised for OpenAI right now, explicitly because of safety. Anthropic says much the same thing about the technology and keeps walking toward the exchange.
That is not a contradiction if you read Amodei’s essay carefully: slowing down there means letting independent evaluators into the building, not stopping. It just makes for an unusual pitch to future investors. Buy us, we are profitable, and we consider our own product dangerous enough that someone from outside should be watching.
Sources: