Anthropic Files for IPO as “ANT” — $16.3B Revenue, an $80.7B Loss, and Founders Who Keep 50.1% of the Vote

Introduction: The S-1 Is Public

Yesterday morning at 9:00 a.m. Eastern, the SEC's EDGAR system accepted a registration statement from Anthropic, PBC — the company behind Claude, Claude Code, and the MCP agent ecosystem. After months of leaked drafts, the real S-1 is finally public: Anthropic intends to list its Class A common stock on the Nasdaq Global Select Market under the ticker ANT, with Goldman Sachs and Morgan Stanley leading a 22-bank underwriting syndicate.

The filing is the most consequential AI IPO document since the AI boom began — not because of the ticker, but because of what it reveals: a business growing fourteen-fold in a year, an eye-watering net loss that is mostly an accounting artifact, half a trillion dollars in compute commitments, and a governance structure that Anthropic itself admits has “no direct public company precedent.” Here's what the paperwork actually says, stripped of the hype.

1. The Numbers: 14x Growth, a Paper Loss, Real Cash

The revenue line is the one everyone will quote. For the six months ended June 30, 2026, Anthropic booked $16.29 billion in revenue, up from $1.16 billion in the same period a year earlier — roughly fourteen-fold growth. The operating loss actually shrank, from $2.1 billion to $693 million, which is nearly unheard of for a company scaling that fast.

Then there's the number that will lead every headline: a net loss of $80.75 billion for the half-year. But about $72.4 billion of that is a non-cash fair-value charge — convertible notes and similar instruments being revalued upward as the company's private valuation climbed. It's the accounting cost of being a rocketing success, not money burned. Anthropic holds roughly $131.7 billion in cash and short-term investments against it.

Two enterprise metrics stand out for anyone building a business on AI: more than 11,900 customers now spend over $100,000 a year with Anthropic, and reported net revenue retention is extraordinarily high — existing customers keep multiplying their spend as agent workloads grow. The filing still carries the standard warning that Anthropic “may not achieve or sustain profitability,” and with the compute bill below, you can see why.

2. You Can Buy ANT, But You Won't Control It

The most unusual part of the S-1 isn't financial — it's the share structure. Public buyers get Class A shares with one vote each. But a single Class F share, held by a “Founder LLC” made up of Dario Amodei, Daniela Amodei, and the five other co-founders, carries 50.1% of the total vote. On top of that, Anthropic's Long-Term Benefit Trust holds Class T stock that lets it elect a majority of the board, and employees get Class E shares with limited voting rights.

Anthropic doesn't pretend this is normal. The filing states plainly that the structure “has no direct public company precedent.” The practical translation: even if you bought every share on the open market, the founders and the trust would still run the company. For safety-minded observers, that's the point — the mission lock survives the listing. For activist investors, it's a wall. Either way, ANT is a bet on management's judgment, full stop.

3. $417 Billion of Compute IOUs

Buried in the commitments section is the number that explains the entire AI economy right now: Anthropic signed roughly $417 billion in new compute obligations this year, with an estimated $125–150 billion more for capacity arriving between 2028 and 2031. That's over half a trillion dollars promised to data centers — money that must be earned back by selling Claude subscriptions, API tokens, and enterprise agent deployments for the next decade.

This is the real story of the IPO. Anthropic isn't raising money because the business is failing; it's raising money because the business is growing so fast that even $131.7 billion in cash gets absorbed by GPU contracts. Every AI lab going public in this cycle — and several are queued up behind Anthropic — is essentially converting investor capital into compute capacity, then racing to grow revenue faster than the bill comes due.

4. The Deal: What Listing Could Look Like

The S-1 leaves the share count and price range blank, as expected at this stage. But the Wall Street Journal, citing people familiar with the plan, reports that pricing details could arrive within about two weeks of investor feedback, the roadshow is being prepared, and the target listing window is the week of November 16. The offering could raise $100 billion or more at a valuation between $1.5 and $2 trillion — which would make it one of the largest IPOs in market history, and put Anthropic ahead of where OpenAI's private rounds have been trading.

The underwriter roster signals how big the banks think this is: Goldman Sachs and Morgan Stanley lead, followed by J.P. Morgan, Citigroup, Barclays, Wells Fargo, BofA, Deutsche Bank, RBC, UBS, and twelve more.

5. What an IPO Means for Claude Users

If you're one of the millions of people using Claude, Claude Code, or the growing fleet of tools built on Anthropic's API, a public Anthropic changes three things:

Quarterly transparency

Revenue, spend, rate limits' underlying capacity, and compute commitments will be reported every quarter. Expect the community to dissect each release — like it already does for every model card.

Pressure to monetize the ecosystem

Public markets reward growth, and agent workloads are the growth engine. Expect more usage-based pricing tiers, more enterprise features, and more aggressive bundling of Claude Code-style products rather than price cuts.

Mission lock, for better or worse

The Founder LLC and Long-Term Benefit Trust mean safety priorities can't easily be voted away. If you choose AI tools partly on the stewardship of the lab behind them, the S-1 just made that stewardship structural — and visible.

None of this changes which model is best for your work today. It changes who's accountable for it — and gives you quarterly receipts.

Frequently Asked Questions