What Anthropic's 25 Gigawatts Can Earn in 2030
On Wednesday I showed what OpenAI has to believe to break even in 2030 and said I would run the same exercise for Anthropic. However, when I started working on Anthropic, I realized the relevant questions are a bit different for the company. Indeed, it was reported just yesterday that OpenAI actually increased its cash burn estimates from $180 Billion to $278 Billion by the end of 2030. If OpenAI indeed becomes public next year, I continue to believe public market may be less forgiving to such persistent and gargantuan cash burn which may necessitate OpenAI to show investors a more less chaotic path to profitability. It will be particularly challenging if Anthropic, their primary competitor at the frontier model race, shows much better economics which can lower the appetite from investors to fund such cash burn. Anyways, given media reports suggest Anthropic already posted operating profit (ex SBC) in Q2 this year and expects to be profitable in Q3 as well, the more relevant and useful question for Anthropic is what GAAP operating margin the fleet Anthropic is building can support on assumptions one can defend, and how the margin behaves if pricing at the frontier goes the wrong way. So, that’s the key focus on my Anthropic exercise.
Speaking of becoming public, Anthropic is now apparently going to IPO a month later than expected. All I want is the IPO to actually happen so that we can follow at least one frontier lab with audited financials. It will also be useful to stop the charade of crossover investors enjoying an unusual informational advantage from frontier labs ARR trajectory which remains a key driver to much of the broader AI trade. Given that IPO is hopefully a couple of months away, this exercise is timely to wrap my head around Anthropic’s economics.
Just as I did with OpenAI, I will show you the math, and share the downloadable spreadsheet (which you can change to fit your narrative and point of view) behind the paywall.
Subscribers get the daily journal and five+ years of Deep Dives, i.e. full-length analyses with financial models on 70 companies. The daily is just how I think out loud between the Deep Dives!