Lead Analysis
Anthropic crosses $100B revenue pace as its ~$2T IPO moves to November
The largest public offering in history is now on a dated calendar — with a possible pre-IPO model launch in the window Indian enterprises plan Q4 procurement against
Monday, September 21, 2026: The most significant AI development of the day is Anthropic storming past a $100 billion annualised revenue pace — up 50% from July — and shifting its IPO to November at a valuation investors are discussing near $2 trillion.
The New York Times, via Axios and Yahoo Finance, reports Anthropic is now pacing to generate over $100 billion in annualised revenue this year, up 50% from the $65 billion disclosed in July and more than ten times end-2025 levels, driven by Claude Code and Cowork enterprise adoption. The company has pushed its planned IPO from October to November to fold in Q3 financials, with investors discussing a raise of up to $100 billion at a roughly $2 trillion valuation — a combination that would make it the largest public offering in history. The Wall Street Journal carried the November timing on September 18; Nvidia and Microsoft are cited as key backers.
The revenue milestone explains the competitive tension reported over the weekend: Reuters said Anthropic is considering releasing a new model ahead of the IPO to counter OpenAI’s momentum since GPT-6 Astra — now about 13% of enterprise AI spend tracked by Ramp versus roughly 8% for Claude’s Fable tier. A lab publicly calling for frontier deceleration while privately preparing a record listing and a counter-model is the structural split Indian AI planners must price into every vendor decision this quarter.
The pricing landscape is simultaneously deflating underneath the giants. StepFun opened its 600B-parameter Step 5 Preview API on September 20 at $1 per million input and $2.70 per million output tokens with a 95% cache discount — an Artificial Analysis Intelligence Index of 44 matching Kimi K3 Max at roughly a seventh of GPT-5.6 Sol’s price, with open weights promised for October 15. TypeSafe’s decision model Jev became Vercel’s fastest-adopted model ever, reaching nearly 13% of paid teams in 24 hours, double the GPT-5.6 family at the same mark, on claims of up to 194x faster and 445x cheaper workflow evals.
For Indian enterprises, the sequence is a calendar, not a rumour: a record Anthropic listing in November, a possible counter-model before it, continued deflation at the commodity tier, and a US-China negotiating channel — Treasury Secretary Bessent and Vice Premier He Lifeng opened New York talks on AI, tariffs and rare earths on September 20 ahead of the September 24 Trump-Xi summit. Export-control outcomes from that session will shape which compute and models Indian firms can procure at what price for the rest of the year.
Domestic context is set by the NSE IPO closing today at 1.16x subscription — a ₹22,561 crore offer for sale at Rs 1,700-1,785 per share, India’s largest exchange listing — alongside markets reopening after a sixth straight weekly loss. Nifty futures near 23,380 and GIFT Nifty ~23,300 on Sunday night imply a cautious start, with Brent above $100 and the rupee hovering near 96. The distance between India’s exchange listing and Anthropic’s $2 trillion ask is a useful measure of how much of the AI value stack still prices abroad.
