Anthropic at $2 trillion would be pricier than Cisco at the top of the dot-com bubble

Short answer: on every yardstick used for a normal company, yes, $2 trillion looks overvalued. But Anthropic is not growing like a normal company, and that is the whole argument. Investors told the Financial Times they expect Anthropic’s IPO to value it at $2 trillion or more. That is 43 times its revenue pace: a higher price for each dollar of sales than Cisco had in March 2000, the peak of the dot-com bubble, and 4th highest among 32 big tech IPOs since 2011. Its revenue also grew about 14-fold in a year. No listed tech company comes close to that, so there is no track record to check the price against. Below are four comparisons and two statistical tests. Make up your own mind.

The price tag

At $2 trillion, Anthropic would cost 43× its revenue. Cisco, at the very top of the dot-com bubble, cost 32×.

Market value divided by the latest quarter's revenue × 4: $46B a year for Anthropic (Q2 2026), $17.4B for Cisco (Q2 of fiscal 2000)

$965 billion

Series H funding round, May 2026

21×revenue (price-to-sales)

Rank among 32 big tech IPOs

9th priciest

Costs about as much as

6 famous brands

Bigger than the GDP of

168 economies

$2 trillion

IPO price investors expect, per the FT

43×revenue (price-to-sales)

Rank among 32 big tech IPOs

4th priciest

Costs about as much as

11 famous brands

Bigger than the GDP of

176 economies

Anthropic's revenue grew 1,361% in a year (Q2 2026 vs Q2 2025). No listed tech company in this study grew faster than 167%.

1. The same price buys eleven famous companies

Sources: market values at the close on Oct 2, 2026 (Yahoo Finance); revenue from each company's latest 10-K (SEC EDGAR); Anthropic per CNBC and the FT (links in Methodology).

For $2 trillion you could buy all of Coca-Cola, PepsiCo, McDonald’s, Starbucks, Nike, Disney, Netflix, Boeing, Ford, GM and Costco at their share prices on October 2, 2026, and have change left: together they are worth $1,996 billion. The first six alone are worth more than Anthropic’s last funding round valued it at ($965 billion, in May).

Those eleven companies sold $1,129 billion of goods and services in their latest fiscal year, 25 times Anthropic’s revenue pace of $46 billion a year, and earned $63 billion in net profit between them. Anthropic reported its first profitable quarter in Q2 2026: $559 million of adjusted operating profit, by its own (unaudited) measure.

The counter-argument: a market price is a bet on future profits, not a receipt for past ones. The question is whether Anthropic’s next ten years can earn what these eleven companies earn today.

2. Nobody got this big this fast

Sources: quarterly revenue from SEC EDGAR filings; incorporation dates from each company's IPO prospectus; Anthropic per CNBC (links in Methodology).

This is the strongest case for the price. Anthropic was founded in January 2021. By the second quarter of 2026 its revenue reached $11.5 billion, a pace of $46 billion a year. It took 5.5 years. Google needed 13.8 years from incorporation to the same pace, Facebook 13.5 and Amazon 16.5. Measured the same way, Anthropic got there about two and a half times faster than the fastest of them.

And it has not slowed down yet: in the first quarter of 2026 revenue was $4.73 billion, so it more than doubled in three months, and Anthropic said its revenue pace reached $65 billion in July. On that newer figure, $2 trillion is 31 times revenue rather than 43.

3. The Cisco lesson

Sources: Cisco 10-K filings and annual reports (SEC EDGAR); share prices from Yahoo Finance, split-adjusted, without dividends.

On March 27, 2000, at the peak of the bubble, Cisco was worth $555 billion, or 32 times its latest quarter’s revenue × 4 (37 times its past year’s revenue). Like Anthropic, it sold the picks and shovels of a technology everyone agreed would change the world, and it was right: the internet did change the world. Cisco’s revenue was still growing 53% a year.

Its business kept growing. Its share price did not. By October 2002 it had lost 89%. It took until December 2025, more than 25 years, for Cisco to close above its bubble-peak price again. Over the same period, revenue went from $19 billion to $63 billion a year. Counting dividends, someone who bought at the peak made 119% by October 2, 2026; the same money in the Nasdaq-100 made 656%.

Being right about the technology is not the same as being right about the price. At $2 trillion, Anthropic would start at a higher multiple of revenue (43×) than Cisco did at its peak (32×), measured the same way.

4. Bigger than 176 of the world’s 190 economies

Sources: IMF World Economic Outlook (GDP, current prices); Anthropic valuations per Anthropic and the FT. A valuation and a GDP measure different things: see the note below.

At $2 trillion, Anthropic would be worth more than the whole yearly output of South Korea ($1,931 billion in 2026, per the IMF) and just below Spain’s. Only 14 economies would be larger. Even at $965 billion it is roughly Taiwan’s GDP.

Yes, this is apples to oranges. GDP is what a country produces in one year; a company’s value is what investors pay today for all of its future profits. A fairer reading: investors are pricing Anthropic as if it will one day earn, every year, a sizeable slice of what a mid-sized rich country produces. That is the size of the bet.

5. Do expensive IPOs do worse? Not in our data

Sources: offer prices and share counts from each final prospectus (SEC Form 424B4); revenue from SEC filings; prices from Yahoo Finance (with dividends). Rivian, which had almost no revenue at its IPO, is off the scale.

If a high price at the IPO were a warning sign, the most expensive IPOs should have done the worst. We checked 32 big US tech IPOs since 2011, from Facebook to CoreWeave. There is no link between how many times revenue a company was priced at and how its shares did over the next one or three years, compared with the Nasdaq-100. The statistical tests could not tell the pattern apart from chance (details in Methodology).

What the chart does show is that buying a hot tech IPO has usually been a bad deal, cheap or expensive. Bought at the first day’s close, only 6 of the 30 that are at least three years old beat the Nasdaq-100 over three years. The typical (median) one ended up 61% behind it. Of the eight IPOs priced above 21× revenue (Anthropic at $965 billion) with three years of history, only Datadog beat the index. All four priced above 43× (Anthropic at $2 trillion), Snowflake, SentinelOne, GitLab and Rivian, trailed it by 53% to 92%.

So the price alone does not predict the outcome, and neither does a low price save you. Nor does it help that Anthropic would be 12 times larger than the biggest IPO in this sample (Alibaba, $168 billion at its offer price).

6. Does the growth justify the price?

Sources: revenue from the last four quarters vs the four before (SEC EDGAR); market values at the close on Oct 2, 2026 (Yahoo Finance).

Among 40 listed US tech companies, from Nvidia to Reddit, faster growers do trade at somewhat higher prices, but the link is loose: a typical company trades at about 8 times revenue with no growth, and about 14 times when revenue doubles in a year. The fastest grower, Micron (+167%), trades at 13×. Palantir, Cloudflare, CrowdStrike and Astera Labs cost more than 43× revenue, but none of them is worth even $500 billion. Of the ten companies worth more than $1 trillion, the priciest, AMD, trades at 25×.

Anthropic’s growth (+1,361% year on year) is about eight times faster than anything on this chart. That is why both camps can claim the data. If you read the line, $2 trillion asks for “only” about 287% growth, so Anthropic looks cheap. But that is far beyond any company the line was drawn from, and nothing grows 14-fold a year for long. The real question is not today’s growth, but how quickly it slows down.

So, overvalued?

• The case against the price: 43× revenue is more than Cisco paid at the bubble’s peak; no company worth over $1 trillion trades at more than 25×; the typical big tech IPO trailed the market by 61% over three years; the profit so far is one quarter, on Anthropic’s own measure.
• The case for it: no company in this study has grown this fast at this size; on the July revenue pace, the price is 31× rather than 43×; and in our 15 years of IPOs, a high price on its own did not predict worse returns.

Which side are you on? The answer depends on one number: how fast Anthropic is growing two years from now.

What the numbers leave out

• Anthropic’s figures are not audited. Revenue comes from preliminary shareholder reports described by CNBC. Anthropic has not published a prospectus (Form S-1) yet; when it does, the numbers may change, and this page will be updated.
• $2 trillion is an expectation, not a price. It is what six investors told the FT they expect; the company has not set a price range.
• Growth is measured differently. Anthropic’s +1,361% compares one quarter with the same quarter a year earlier; listed companies’ growth compares the last four quarters with the four before. A year-on-year quarter flatters a company that is speeding up.
• One quarter × 4 is generous to seasonal businesses. Amazon and Facebook crossed $46 billion on a holiday quarter. It is also unflattering to Anthropic, whose revenue rose every month.
• Survivors only. IPOs of companies that were later bought or taken private (LinkedIn, Zynga, Twitter, Qualtrics, Confluent) have no price history and are missing. Several were bought at a premium, so leaving them out may make the IPO returns look worse than they were.
• Timing. Many IPOs in the sample came in 2019–2021 and then lived through the 2022 crash. Over three years, only 10% of the 2019–21 group beat the Nasdaq-100, vs 43% of those from 2011–18.
• Share counts. Valuations use the shares outstanding after the IPO, not counting options and restricted stock, so the price-to-sales of past IPOs is, if anything, a bit low. Anthropic’s valuations are reported as totals, which usually count every share and option.

Data

Returns are from the first day's close, with dividends, compared with the Nasdaq-100 (QQQ) over the same days. "–" means the IPO is less than 3 years old.

Download the datasets: ipos.csv (32 IPOs, with offer prices and returns), peers.csv (40 listed companies), basket.csv, race.csv, cisco.csv and gdp.csv.

Methodology

• Anthropic: Q2 2026 revenue of $11.5 billion, Q2 2025 of $787 million, Q1 2026 of $4.73 billion and $559 million of adjusted operating profit from CNBC, August 15, 2026 (preliminary figures from shareholder reports). The $65 billion revenue pace in July: CNBC, August 17, 2026. $965 billion valuation: Anthropic’s Series H announcement. $2 trillion: Fortune, August 13, 2026, citing the Financial Times. Founded in January 2021: Wikipedia.
• Price-to-sales (P/S) is market value divided by yearly revenue. For Anthropic, Cisco at its peak and every IPO, yearly revenue is the latest reported quarter × 4, because young companies grow too fast for a past-year figure to be fair. For the 40 listed companies it is the sum of the last four quarters.
• IPOs (section 5): 33 US-listed tech IPOs since 2011 with a priced offering (no direct listings or SPACs). Offer price and shares outstanding after the offering come from each final prospectus (Form 424B4 on SEC EDGAR); revenue for the last quarter before the IPO from the prospectus’s XBRL data or the first 10-Q. Returns come from Yahoo Finance prices adjusted for splits and dividends, measured from the first day’s close over 1 and 3 years, and compared with the Nasdaq-100 ETF (QQQ) over the same days (“vs QQQ” is the share’s value relative to the same money in QQQ). Rivian (P/S over 1,000× because it had almost no revenue) is left out of the tests, leaving 32. Tests: correlation of log P/S with the relative return, Pearson and Spearman, with p-values from 20,000 random shuffles. Every p-value is between 0.33 and 0.96 (0.05 or less would count as a real link).
• Listed companies (section 6): market value is the close on October 2, 2026 times shares outstanding of all classes (Yahoo Finance); revenue from SEC EDGAR XBRL filings. The line is a least-squares fit of log P/S on growth. The link is weak: Spearman rank correlation 0.38 (p = 0.016), but growth explains only 4% of the differences in P/S.
• Same price (section 1): market values at the close on October 2, 2026; revenue and net income from each company’s latest 10-K on SEC EDGAR (Ford’s net income: loss attributable to common shareholders).
• Speed (section 2): quarterly revenue from SEC EDGAR XBRL filings (Google’s before 2015 under Google Inc.). Incorporation month from each IPO prospectus: Amazon (1997), Google (2004) and Facebook’s 2012 prospectus. Counting Anthropic from January 2021 is the least flattering choice for it.
• Cisco (section 3): peak close of $80.06 on March 27, 2000 (Yahoo Finance, split-adjusted) times 6.94 billion shares (split-adjusted) from its 10-Q for the quarter to January 2000; that quarter’s revenue was $4.35 billion, vs $2.85 billion a year earlier. Yearly revenue from the 2000 10-K, the 2004 and 2009 annual reports, and XBRL filings after that.
• GDP (section 4): IMF World Economic Outlook, GDP in current US dollars, 2026 projections for 190 economies.
• Code: plain Python scripts (collect.py, analyze.py). They save every filing locally, and each offer price, share count and incorporation date is checked against the exact sentence in the filing.
• Trademarks: company names are trademarks of their respective owners. This site is independent and not affiliated with or sponsored by Anthropic or any other company mentioned. This study was produced with Claude, a model made by Anthropic.