The steepest revenue curve in software history is about to hit the public market. The question isn’t whether it grows — it’s what $965B already prices in.
Anthropic — maker of Claude — is targeting an October 2026 NASDAQ listing after closing a $65B Series H at a $965B post-money valuation, making it the most valuable private company on earth. Run-rate revenue has gone from ~$1B to ~$45B in eighteen months. But the headline ARR is an annualized snapshot, profitability isn’t expected until 2028, and you can’t actually buy the stock yet. Below: the business, the real revenue picture, the cap table, what the valuation implies, the bull and bear cases, and the only ways to express this trade today — which run through Amazon and Alphabet, not Anthropic.
The most valuable private company on earth is heading for the exit. October 2026, NASDAQ, after a $965B round.
Anthropic closed a $65B Series H on May 28, 2026 at a $965B post-money valuation — surpassing OpenAI's prior $852B mark to become the most valuable private company in the world. The round, led by Altimeter, Dragoneer, Greenoaks, and Sequoia, sets up a targeted October 2026 IPO. Unlike OpenAI (still untangling a nonprofit-to-for-profit conversion), Anthropic is a straight venture-backed PBC with audited financials and a clean cap table — a faster path to market.
| Item | Detail |
|---|---|
| Company | Anthropic PBC — maker of Claude |
| Target IPO | ~October 2026 (per Bloomberg, Mar 27 2026) |
| Venue | NASDAQ (expected) |
| Last Round | $65B Series H · $965B post-money · closed May 28 2026 |
| Lead Investors | Altimeter · Dragoneer · Greenoaks · Sequoia |
| Legal Counsel | Wilson Sonsini |
| Founded | 2021 · Dario & Daniela Amodei + ex-OpenAI team |
| Structure | Public Benefit Corp + Long-Term Benefit Trust |
An enterprise API company wearing a chatbot's clothes. ~80% of revenue is business, not consumer.
Anthropic did the opposite of OpenAI. No 900M-weekly-user viral consumer app — instead enterprise API contracts, developer adoption, and one breakout product: Claude Code. Roughly 80% of sales come from business customers. The company reports 300,000+ business customers, up from fewer than 1,000 two years ago, and enterprise accounts spending $1M+ annually have grown nearly 7x in a year.
The Revenue Engine
| Segment | What It Is | Scale |
|---|---|---|
| Claude API | Developers + enterprises calling Claude models programmatically | Core of the business · majority of revenue |
| Claude Code | Agentic coding tool launched mid-2025 | $1B ARR in 6 months · $2.5B+ by Feb 2026 · fastest-growing product in company history |
| Claude (consumer/Pro) | Chat apps, subscriptions | Smaller slice · ~287M monthly web visits (Feb 2026) |
| Cloud distribution | Claude on AWS Bedrock + Google Vertex AI | Dual-cloud reach into enterprise procurement |
Marquee Customers
Netflix · Spotify · KPMG · L'Oréal · Salesforce — enterprise use is over half of Claude Code revenue. The newest enterprise-facing release, Claude Mythos Preview (via "Project Glasswing"), is a cyber-defense model being tested by Microsoft, NVIDIA, Amazon, CrowdStrike, and Broadcom for defensive use.
~$1B to ~$45B in eighteen months. Read the curve — then read the asterisk.
Anthropic's run-rate revenue trajectory is, by several measures, the fastest ramp ever recorded in software. Salesforce took ~20 years to reach $30B annual revenue. AWS took ~13 years to reach $35B. Anthropic went from an $87M run-rate in January 2024 to a $30B run-rate by April 2026, and reportedly crossed $45-47B in May 2026.
| Date | Run-Rate Revenue (annualized) | Note |
|---|---|---|
| Jan 2024 | $87M | Starting point |
| Dec 2024 | ~$1B | ~11x in a year |
| Aug 2025 | ~$5B | |
| Dec 2025 | ~$9B | |
| Feb 2026 | $14B | Series G announcement |
| Mar 2026 | $19B | |
| Apr 2026 | $30B | CEO: 80x annualized growth, 8x ahead of own forecast |
| May 2026 | ~$45-47B | Series H announcement |
Two cloud giants own a quarter of it. Neither gets a vote.
The most important structural fact for a public-market buyer: Anthropic is controlled by its founders and an independent trust, not its largest investors. Google holds ~14% (capped at 15%), Amazon ~7.8% — and neither has voting rights or board seats. A Long-Term Benefit Trust (LTBT) holds special voting power over the board to protect the AI-safety mission.
| Holder | Stake | Investment | Rights |
|---|---|---|---|
| ~14% | >$3B in, up to ~$40-43B committed | No vote · no board seat · capped at 15% | |
| Amazon | ~7.8% | $8B in (now worth >$70B), up to ~$33B committed | No control · primary cloud/training partner |
| Founders | Undisclosed, significant | Amodei siblings + co-founders | Retain operational control |
| VC / Strategic | Remainder | Sequoia, Altimeter, Dragoneer, Greenoaks, Salesforce Ventures, Spark | Standard preferred |
| LTBT | Governance, not economic | Independent trustees | Special voting rights over board |
The Circular-Markup Wrinkle
In Q1 2026, more than half of both Amazon's and Alphabet's pre-tax profit came from marking up their Anthropic stakes (Amazon disclosed $16.8B of pre-tax gains from its Anthropic investment). The markup is triggered when a new round re-prices the equity — and Amazon and Google are themselves investors and compute vendors helping drive that valuation higher. The accounting is uncontroversial; the reflexivity (vendors funding a customer whose rising valuation then inflates the vendors' own earnings) is something the bear case leans on hard. See § 07.
The growth is real. So is the multiple.
At $965B post-money against a ~$45B run-rate, Anthropic is valued at roughly 21x run-rate revenue — and meaningfully higher against trailing GAAP revenue, which is the number that matters. The valuation has more than doubled since February 2026 ($380B → $965B). For an unprofitable company targeting break-even in 2028, the IPO price will rest on continued hyper-growth and market appetite holding through Q4 2026.
Anthropic vs OpenAI — The Only Real Comp
| Metric | Anthropic | OpenAI |
|---|---|---|
| Run-rate revenue | ~$45B | ~$24-33B |
| Last valuation | $965B | ~$852B (prior) |
| Break-even target | 2028 | 2030 |
| Revenue mix | ~80% enterprise/API | Consumer-heavy (ChatGPT) |
| Structure | PBC + LTBT (clean) | Nonprofit→for-profit conversion (messy) |
| IPO readiness | Audited financials, clean cap table | Structurally more complex |
Trade ideas like this, before they hit the timeline.
Join Discord →Why this could be the listing of the decade.
- Fastest revenue ramp in software history — ~$1B to ~$45B run-rate in 18 months · doubling roughly every 6 weeks at points in 2026.
- Enterprise-led, not hype-led — ~80% business revenue · 300,000+ customers · $1M+ accounts up ~7x YoY · stickier than consumer.
- Claude Code is a genuine category winner — $2.5B+ run-rate, fastest-growing product in company history, real enterprise adoption (Netflix, Spotify, Salesforce).
- Break-even targeted 2028 — two years ahead of OpenAI · a distinction public investors may pay a premium for.
- Cleanest path to market among AI leaders — PBC with audited financials and a clear cap table vs OpenAI's structural overhang.
- Compute secured — deals across Amazon, Google/Broadcom (TPUs), NVIDIA, Microsoft, plus SpaceX and xAI capacity to relieve near-term strain.
- Strategic backing — up to ~$76B committed from Amazon + Google combined; they're incentivized to support the franchise.
Why a near-trillion-dollar price tag is a high bar.
- The ARR is annualized, not trailing — "$45B run-rate" is latest-month × 12. Actual Q1 2026 revenue was ~$4.8B. The realized trailing number is a fraction of the headline. Wait for the S-1.
- Not profitable — break-even isn't expected until 2028. The IPO prices in years of continued hyper-growth before the company earns a dollar.
- Reflexive valuation — Amazon and Google are simultaneously investors, compute vendors, and beneficiaries of Anthropic's rising mark; over half of their Q1 profit came from marking up the stake. Critics call the loop circular.
- Single-product concentration — a disproportionate share of growth rides on Claude Code · competitive coding tools (OpenAI, Google, Cursor, GitHub Copilot) are coming hard.
- Compute dependency & cost — enormous capex · "inevitable strain on our infrastructure" already hit reliability during peak hours · margins gated by cloud partners who are also rivals.
- Governance blunts shareholders — PBC + LTBT means public buyers get economics without control; the mission can legally outweigh returns.
- Competition — OpenAI (also IPO-bound), Google Gemini, Meta, and others are extremely well-capitalized · model leadership is not guaranteed to persist.
- AI-bubble / macro risk — a Q4 2026 listing depends on AI-capex enthusiasm and equity conditions holding. Any sentiment crack compresses the multiple fast.
You can't buy Anthropic yet. So the trade runs through its owners.
The IPO is targeted for ~October 2026. Until shares price and list, there is no ANTH ticker to buy on a normal brokerage. That leaves a small set of real expressions — and the cleanest public proxies are the two cloud giants holding ~22% of the company between them.
The Tradeable Expressions Today
| Route | Mechanics | Read |
|---|---|---|
| $AMZN | ~7.8% stake · $8B in now worth >$70B · primary cloud partner | Cleanest large-cap proxy · but Anthropic is a rounding error on Amazon's total cap — diluted exposure |
| $GOOGL | ~14% stake · up to ~$40-43B committed · Vertex AI distribution | Largest single equity holder · same dilution caveat · also a direct Gemini competitor to its own holding |
| Pre-IPO secondary | Forge Global, EquityZen, etc. | Accredited investors only · illiquid · wide spreads · headline-valuation pricing |
| Compute / chip ecosystem | NVDA, AVGO (TPUs), and AI-infra names | Second-derivative — they sell into the buildout regardless of which model wins |
| Wait for the S-1 | File → roadshow → price → list | The S-1 reveals GAAP trailing revenue, burn, margins · the only honest basis for an entry |
Best company, hardest price. Wait for the S-1, watch the proxies.
Anthropic is the strongest fundamental story in AI — fastest revenue ramp ever, enterprise-led mix, a genuine category-winning product in Claude Code, break-even targeted ahead of OpenAI, and the cleanest path to market of any frontier lab. But $965B is a demanding entry, the headline ARR overstates realized revenue, profitability is two years out, and the governance structure hands public buyers economics without control. You also can't buy it directly today. Until the S-1 turns run-rate into GAAP, the honest expressions are diluted (AMZN, GOOGL) or illiquid (secondaries).