Anthropic's IPO rests on a $200B 2028 revenue forecast

Your model vendor's valuation is a forecast about how much you'll spend. Reuters reported on August 14 that bankers pricing an Anthropic IPO are anchoring on a 2028 revenue forecast of $190–200 billion — roughly four times the $47 billion run rate the company publicized in May. That gap is not trivia. It's the number someone has to hit, and the fastest path there runs through your invoice.

What actually happened

Reuters attributes the 2028 figure to two people familiar with Anthropic's financials, and the broader reporting to four sources. Anthropic did not respond to a request for comment before publication. Treat this as credible reporting on private numbers, not a company disclosure.

The trajectory the reporting describes: roughly $9 billion in run-rate revenue at the end of 2025, $47 billion by May 2026, and a $190–200 billion target for 2028. Bankers are reportedly using enterprise value-to-revenue multiples on 2028 forecasts rather than current results — the comparables cited include Palantir at about 53x revenue, and SpaceX and Cloudflare at about 41.6x expected 2026 revenue.

Valuing a company two years forward is unusual. It means the price depends less on what Anthropic sells today than on an assumption about 2028 demand holding.

Why the Anthropic IPO forecast matters for your business

A growth target is a pricing signal. Nobody quadruples revenue in two years by holding rates flat and waiting. It comes from more seats, more per-token spend, more usage-metered features, and fewer generous free tiers. We've watched this play out all year — Claude Opus 5 landed at half the price of Fable 5 while DeepSeek raised rates across the board. Prices move both directions. Plan for the direction that hurts.

Public markets change vendor behavior, not just ownership. A listed company answers to quarterly guidance. Deprecations get scheduled around margin, not around your migration window. Model retirements, tier changes, and rate limits become financial instruments.

None of this is a reason to leave. Claude is excellent and we build on it. It's a reason to keep the exit cheap: your prompts and evals in your repo, the model behind one interface, per-task cost tracked, and a second provider that already passes your test suite. If switching is a week of work, vendor risk is a line item you manage. If it's a quarter, it's a hostage situation.

Sources: Reuters.