As a layman external observer, this seems a bit rushed to me? I know there's a race amongst the frontier AI labs, but I don't quite understand the rush (apparently there is still a lot of money to go around), so as some comments imply, it does look like "our financials look good right now let's strike while the iron is hot!"

I do think Anthropic's business has very good long-term prospects, but the current run rates are not sustainable and they know it which is why they, more than OpenAI, are under higher pressure to IPO. Some things to consider:

1. This was surprising to me, but enterprises Claude Code (and Codex) plans are billed on token usage at API rates. I was expecting lower rates for volume subscriptions. This explains their huge spike in ARR, but I expect competitive pressures will soon come into play, especially as companies start to get more budget-conscious. Specifically...

2. Tokenmaxxing is finally encountering the inevitable pushback. My theory is it was an effort to incentivize devs to experiment and figure out ways to get productive with AI by throwing money at the problem, which was always going to be a short-term dynamic. Companies are going to be much more intentional about token budgets (especially as Anthropic is apparently now asking for volume commitments for enterprise plans.) Smaller, open-weight models may start looking much more attractive.

3. I've said before but I think Anthropic severely underestimated their own popularity and corresponding demand for compute and has to enter costly deals to acquire capacity to keep Claude's 9's above GitHub's, even as they alienated customers with short term tweaks to optimize usage. These deals will eat into their margins and Claude's problems likely pushed customers to competitors, the effects of which could take time to be more evident.

So maybe whatever looks good on their financials right now is time-limited, and the current boost may start petering out at some point, which may influence when OpenAI files their own IPO.