Anthropic IPO: Priced For 2031, Or 2036?

AMZNYTD+11.7%SPYYTD+8.2%XLYYTD-1.3%
Analyze AMZN →

Anthropic is expected to price its IPO tonight, Wednesday Nov 18. Its shares are expected to start trading Thursday under the ticker ANT. Most people ask whether AI is real. For Anthropic, the numbers already answer that.

Anthropic took in $17.0 billion of sales last quarter, up from $1.4 billion a year earlier. Its sales pace jumped $10 billion in October alone, to about $81 billion a year. That pace is what it would take in over a full year at its current rate.

The harder question is when. A stock price is a bet on when the money arrives, not just whether it does.

Image from Pixabay

What Does $74 A Share Buy?

The filing sets a price range of $70 to $78 a share. There is no final price yet. At the $74 midpoint, Anthropic would be worth about $968 billion. That is 12 times its yearly sales pace.

This uses the basic share count, the 13.1 billion shares after the offering. On that count, the range runs from $917 billion to $1.02 trillion. Stock options and Amazon‘s convertible notes, loans that can turn into shares, would add more shares. Counted that way, called fully diluted, the value is higher.

If you plan to wait for the first trades before you buy, your cash does not have to sit idle. The right price can take months. Meanwhile, the Trefis Income Solution pays a dividend yield about 3x the S&P 500’s. It holds 50 stocks with at least 10 straight years of dividends. See if it fits.

Which Year Is The Price Counting On?

Analysts expect Nvidia to sell $1.2 trillion a year by 2031. In our analysis, that forecast rests on one assumption. By 2031, AI would do about 1 in 7 hours of the world’s knowledge work. That is the work people do at desks and computers.

Getting there by 2031 means work changing at the fastest pace America has seen. Anthropic’s price leans on that pace too. America has matched it only in bursts. Farm jobs, for example, fell 32% from January 1948 to January 1960.

The average pace is slower. US farm jobs fell 1.6% a year from 1948 to 2025, using annual averages. At that pace, AI gets to 1 in 7 hours in about 10 years, around 2036. That is five years later. US factory jobs fell 0.9% a year from their 1979 peak. At that pace, it takes about 18 years, to around 2044.

So that is the question in our title: 2031 or 2036? The full case is in Who Really Pays For Nvidia’s $1.2 Trillion?

What Is Five Years Worth?

More than a third of the price. Money that comes later is worth less today, because money in hand can earn a return while you wait.

Say you want 10% a year for owning a young company that still loses money. Then money that arrives five years late is worth 38% less today. At the factory pace, 13 years late, it is worth 71% less. Even at 8% a year, a five-year delay costs 32%.

None of this needs AI to fail. It only needs the payoff to come later.

The payoff arrives Years late Worth today
2031, on the burst pace 0 100%
2036, farm average 5 62%
2044, factory pace 13 29%

Source: Trefis. Value today of a dollar due on the 2031 path, at 10% a year. Farm and factory job data from FRED.

Has This Happened Before?

Yes, to the internet’s winners. The internet came. The buyers at the peak still waited.

Amazon’s stock is about 48 times its December 1999 peak price today. But its peak buyers first saw a 94% fall. They needed 9.8 years to get back to even. Microsoft took 14.5 years, the Nasdaq 100 fund 14.9 and Cisco 21.4. All of these buyers were right about the technology. eBay, the exception, was back in 3.7 years.

Stock, 1999-2000 peak Worst fall Years to get back
Amazon, Dec 1999 -94% 9.8
Oracle, Sep 2000 -84% 14.3
Qualcomm, Jan 2000 -87% 14.1
Microsoft, Dec 1999 -69% 14.5
Nasdaq 100 (QQQ), Mar 2000 -83% 14.9
Intel, Aug 2000 -82% 17.5
Cisco, Mar 2000 -89% 21.4
S&P 500 (SPY), Mar 2000 -48% 6.6

Source: Trefis. Daily closes, dividend-adjusted, Jan 1998 to Nov 17, 2026. The worst fall is measured before the stock got back above its peak.

The price in 2000 had already counted the future, too early. Being right about where a technology ends up does not protect you from paying for it too soon.

What Is Different This Time?

Anthropic is already selling, not just promising. Amazon in 1999 lost money on growing sales. Anthropic’s sales in the September quarter were up 1,127% from a year earlier, while its computing costs rose 382%. Its operating loss was just 3 cents for every dollar of sales last quarter. That is the loss from running the business, before interest and taxes.

More than 11,900 customers now pay Anthropic $100,000 or more a year, up from more than 950 a year earlier.

But the bills are already booked. Anthropic has committed $900.9 billion to computing. Of that, $70.7 billion falls due in the 12 months after September. If the payoff date slips, those bills do not.

So, Should You Buy On Day One?

Ask what the price assumes about the date, not about AI. In our view, at 12 times its sales pace, the IPO price leans on 2031. Our record of giant IPOs says the first months are rarely the cheapest time to buy. The details are in Anthropic IPO: Buy Day One Or Wait?

If you own an S&P 500 fund, you already own the companies that Anthropic’s computing bills go to.

AI can be right, and the IPO price can still be early. In 2000, being early cost some of the internet’s believers up to 21 years.

Nobody can tell you today whether the payoff comes in 2031 or 2036. A plan that does not depend on guessing that date is easier to live with.

Stop Guessing. Start Positioning.

No one can predict surprises like this one. The Trefis High Quality (HQ) Portfolio doesn’t try: it holds 30 quality stocks in equal weight, so no single surprise decides your result. Come see how it works.

See How HQ Works

For education, not a recommendation to buy or sell.

For questions about this article or underlying research, email help@trefis.com.