Published: August 13, 2026 | Reading Time: ~9 minutes | Channel: business
A company that ended 2024 pulling in roughly $1 billion a year is now being positioned to list at $2 trillion. Not a typo. Not a meme. That is the actual number investors fed to the Financial Times this week: Anthropic will stage its IPO in October at a valuation of $2 trillion or more, the largest public offering ever — larger than SpaceX, larger than anything Wall Street has ever underwritten.¹
And the same day that story broke, Bloomberg reported Anthropic is quietly trying to buy an Israeli startup called Decart for about $6 billion — a company most people outside the AI bubble have never heard of — to shave the cost of training its models.²
Two headlines. One company. The same underlying question: is this the greatest wealth-creation event of the century, or the single most expensive act of collective self-delusion in market history? Let's do the math, because the math is the only thing that doesn't care what you want to be true.
Here's the revenue walk. Read it slowly, because each jump is more absurd than the last:
| Period | Annualized Revenue | Source |
|---|---|---|
| End of 2024 | ~$1 billion | TheNextWeb / Axios |
| End of 2025 | ~$9 billion | TheNextWeb |
| February 2026 | ~$14 billion | TheNextWeb |
| March 2026 | ~$19–20 billion | TheNextWeb |
| April 2026 | ~$30 billion | TheNextWeb / Bloomberg |
| Early May 2026 | ~$44 billion | CNBC |
| Mid-May 2026 | ~$47 billion | TechCrunch / CNBC |
| Late July 2026 (est.) | ~$69–74 billion | NextBigFuture (3rd-party tracker) |
Axios put it bluntly and it deserves repeating: no company in American history has ever grown like this. Roughly 1,400% year-over-year. Salesforce took two decades to reach a valuation Anthropic hit in four years. Microsoft needed three.³
But here's where it gets uncomfortable. The Fortune report says investors believe Anthropic's run rate will be $100–120 billion by the end of the year.¹ Even if we take the lower bound — $100 billion — and price the company at $2 trillion, that's a 20x forward revenue multiple. If revenue misses and lands at $69 billion (the tracker's realistic late-July figure), you're at roughly 29x.
Twenty-nine times revenue. For a company that has not yet delivered a single full year of operating profit.
Anthropic's cap table reads like a fever dream. Here's the valuation history:
| Date | Round | Valuation |
|---|---|---|
| March 2025 | Series (various) | $61.5 billion |
| September 2025 | Series F | $183 billion |
| February 2026 | $30B round | $380 billion |
| April 2026 | Investor offers | ~$800 billion |
| May 2026 | Series H ($65B) | $965 billion |
| July 2026 | Valuation change | ~$1.2 trillion |
| October 2026 (target) | IPO | $2 trillion+ |
Total raised: $162.255 billion across 35 rounds, per CB Insights.⁴ That's more than the GDP of most countries. The latest $65 billion Series H — co-led by Altimeter, Dragoneer, Greenoaks, Sequoia, Capital Group, Coatue, and D1 — was so oversubscribed that one institutional investor reportedly pledged $5 billion just to get a meeting with the CFO.⁵
And the existing backers are already sitting on paper fortunes. Google owns roughly 14% of Anthropic and has booked $10.7 billion in net gains on the stake. Amazon invested ~$8 billion and reported a $9.5 billion pretax gain. Both multiples of their original checks.³
This is the part the bull case points to and says "see, smart money is already in." The bear case points to the same numbers and says "this is what a top looks like."
Here's the contrarian take, and I want you to sit with it: every single number that makes $2 trillion look sane is a forecast, not a fact.
The revenue walk is real — I verified it across four independent sources. But the valuation is built on an assumption that the current growth rate continues, essentially uninterrupted, for another two-plus years. The FutureSearch model — the most detailed public forecast I could find — puts Anthropic's median annualized run rate at $115 billion by May 2027, with a 90th percentile of $190 billion and a 10th percentile of $75 billion.⁶
Read that again. The pessimistic case still has Anthropic at $75 billion in ARR. The optimistic case has it near $200 billion. That spread is enormous — and it's exactly why the IPO is being priced on hope, not on a discounted cash flow.
Two things everyone is glossing over:
1. The coding moat is real but not unassailable. Claude Code went from $500 million ARR (Sept 2025) to $8 billion (May 2026). That's the engine. But Cursor launched its own model in November 2025 and is already near $4 billion ARR with a 60% enterprise mix.⁶ The disintermediation risk isn't theoretical — the exact tools Anthropic monetizes are being rebuilt by competitors every quarter.
2. Concentration cuts both ways. Anthropic has 1,000+ customers spending over $1 million annually (up from 500 in February), and it owns 54% of coding-specific LLM spend.⁶ That's a moat and a liability. When 54% of a category is one vendor, the switching cost argument works both ways — if a cheaper, good-enough model shows up, the churn would be violent.
And the elephant in the room: OpenAI. It raised $122 billion in March at an $852 billion valuation and has its own IPO clock running. The entire AI valuation race is two companies sprinting toward the same liquidity event, each hoping to price before the other pops the bubble.
I'm not telling you to buy or fade the Anthropic IPO. I'm telling you to stop treating a 20x-revenue listing as a "sure thing." Here's the discipline:
1. Separate the company from the stock. Anthropic is a genuinely extraordinary business. That does not mean the IPO price is a good entry. A great company can be a terrible stock at the wrong multiple. These are different questions.
2. Size for the tail, not the headline. If you must participate, the position should be sized so that a 50% drawdown in the first 90 days doesn't change your life. That's not hedging language — that's the median outcome for parabolic AI listings that price on forward fantasy. Ask anyone who bought at the peak of the last "greatest company ever."
3. Watch the Decart deal as a tell. The $6 billion acquisition isn't just talent — it's a signal that Anthropic's compute costs are so high they're willing to buy a startup whose core tech reduces training cost.² That's a company quietly admitting its own cost curve is a problem. When a business this profitable-looking is spending billions to shave infrastructure costs, that's the margin story nobody's putting in the pitch deck.
4. Mark the date, not the price. The real signal is what happens 90 days after the IPO — the lockup expiry. Late-stage holders (Google, Amazon, and 160+ investors) will be looking to trim. If you're bullish, your entry is more likely to be better in January 2027 than in October 2026.
The export-control landmine. Anthropic's Mythos-class models were pulled offline for three days in June by a Commerce Department export-control order, and it had an earlier run-in with the Pentagon that got its models blacklisted.⁶ The company is one regulatory decision away from a headline that guts the multiple. This is a political risk priced at zero in a $2 trillion valuation.
The concentration risk is hiding in plain sight. Fifty-four percent of coding LLM spend is one vendor. One. If a Chinese lab ships a good-enough open-weight model at a fraction of the price — and that's been the pattern all year — the "moat" narrative reverses overnight.
The multiple assumes perfection. At $2 trillion and a ~$70 billion realistic run rate, you're paying ~29x revenue. That means the market is pricing in years of uninterrupted compounding. Any deceleration — corporate AI fatigue is already showing up in Reuters commentary⁷ — and the air comes out fast.
Lockup overhang. 160+ investors, $162 billion raised, all of it marked up on paper. When the lockup expires, the sell pressure is structural, not psychological. You're not buying a fresh float; you're buying into a queue of holders who've been waiting years for liquidity.
Anthropic is the most impressive revenue story I've seen in two decades of watching markets — and simultaneously the clearest example of a valuation that has fully decoupled from any reasonable discounting of cash flows. The two statements aren't contradictory; they're the same company. Buy the narrative and you're paying 29x revenue for a growth rate that has to be a miracle to justify itself. That's not analysis, it's faith — and faith is a terrible position size.
The one thing to remember: extraordinary companies and extraordinary entry prices rarely show up at the same moment. The company is real. The $2 trillion is a bet. Treat them accordingly.
All claims verified against Gold-tier (Bloomberg, Reuters) and Silver-tier (CNBC, Fortune, TechCrunch, Financial Times) sources. Each source URL was scraped and confirmed accessible. Last verified: August 13, 2026.
The only thing crazier than a $2 trillion price tag is how fast it starts to feel normal. 🎯
