Anthropic’s IPO may force public markets to invent a new financial language for AI companies.

Revenue growth alone no longer tells investors enough, because an AI lab can grow quickly while its token economics, compute intensity, customer retention, and capital requirements move in very different directions.

That is why, already many prospective Anthropic investors are asking for metrics such as revenue per million tokens, serving cost per token, revenue per gigawatt of compute, net revenue retention, and revenue mix across model generations.

And Anthropic is also doing a lot, for strategically positioning itself for an IPO, cleaning up ARR reporting, managing how investors perceive revenue quality, customer concentration, model performance, and growth immediately before going public, to strengthen the story for investors.

if Meta later stops spending billions with Anthropic, public-market investors won’t see a sudden ugly revenue drop because that revenue was already removed from the headline number.