Anthropic is preparing to file for what could be one of the largest initial public offerings in history, expecting the deal to match or exceed the record-setting SpaceX debut, according to a Bloomberg report on Friday, August 21, 2026. People familiar with the process said the Claude maker could file its registration documents as soon as the end of August.
Revenue tipping into public-market territory
Annualised revenue at Anthropic reportedly hit around $65 billion at the end of July, up sharply from late 2025 as enterprise Claude usage, coding tools and AI agents accelerated. Q2 alone topped $11.5 billion, as reported in the company's Q2 filing. Bankers view the numbers as sufficient to underwrite a listing at valuations that comfortably clear the previous frontier-AI benchmark.
Compute, capex and the case for public capital
An IPO would give Anthropic a durable source of capital to feed a compute build-out that already includes a $9.1B Riot data-centre lease, a $10B Volta compute contract and a growing in-house silicon team. Public equity would also add a currency for future M&A after this week's $6 billion Decart bid and let backers Google and Amazon partially crystallise their stakes.
Scrutiny once the S-1 lands
The filing will expose Anthropic's spending, margins and governance to public-market scrutiny for the first time. Investors will be watching gross margins on Claude API calls, cost per training run, revenue concentration among a small set of hyperscaler customers and the sustainability of a model where compute costs still scale close to revenue. A September listing target would put Anthropic on the market ahead of a similarly telegraphed OpenAI S-1.
Reporting based on coverage from Bloomberg and TechStartups.