OpenAI or Anthropic trades at a $5 trillion valuation
Will OpenAI or Anthropic close a trading day on a public exchange with a market capitalisation of $5 trillion or more?
Chance by Q4 2031
39%
Cumulative probability
- Latest forecast
| Year | Q1 | Q2 | Q3 | Q4 |
|---|---|---|---|---|
| 2026 | <1% | |||
| 2027 | 1% | 3% | 6% | 9% |
| 2028 | 12% | 15% | 18% | 20% |
| 2029 | 22% | 24% | 26% | 28% |
| 2030 | 29% | 31% | 32% | 34% |
| 2031 | 35% | 36% | 37% | 39% |
Model reasoning
Anthropic is expected to list in November 2026 at about $2 trillion, and OpenAI probably in 2027 at a similar or lower value. Either would need to grow about two-and-a-half-fold to close at $5 trillion, a level only Nvidia has reached. Given very fast revenue growth but high bubble, rate and regulatory risks, I put this near 1% this year, about 9% by end-2027 and about 38% by end-2031. The biggest swing factor is whether Anthropic's revenue keeps compounding after listing.
Where things stand (13 Oct 2026)
Neither company is listed yet.
- Anthropic filed a confidential draft S-1 on 1 June 2026. The filing is not yet public on EDGAR. Its investor day is on 14 October, and Bloomberg (1 Oct) reports marketing could start the week of 9 November, with trading before Thanksgiving. Prospective investors see fair value at about $1.8–2.0T (Bloomberg; TradingView summary).
- Revenue has grown extremely fast. The annualised rate was about $9B at end-2025, $65B in July 2026 and more than $100B by mid-September (Axios, 18 Sep). Q2 2026 revenue was $11.5B.
- The company's own projection is $190–200B of revenue in 2028 (Reuters via Parekh).
- The leaked S-1 shows a 2025 net loss of $42B and $518B of compute commitments (Reuters/Devdiscourse, 28 Sep).
- The share structure has several classes, including a Founder LLC that holds a Class F share (CNBC/Reuters, 29 Sep). Under the question's rules, unlisted classes with the same economic rights count toward market cap.
- OpenAI: Altman has ruled out an IPO in 2026 (Reuters, 12 Sep). It is seeking a private round at about $1.4T (Bloomberg, 29 Sep). Its last valuation was $852B, with about $50B of annualised revenue at end-September (CNBC, 8 Oct).
- Market context: only Nvidia is above $5T (about $5.46T); Apple is about $4.98T (companiesmarketcap). The Fed raised rates in September and the 10-year Treasury yield is about 5.3%, its highest since 2002. Worries about an AI bubble come and go.
- Closest comparable: SpaceX listed on 12 June at $1.77T and closed its first day 19% higher, around $2.1T. It fell below its IPO price in July and is about $2.15–2.2T now.
Prediction markets (Polymarket, read via its data feed on 13 Oct):
- Anthropic IPO by 31 Dec 2026: 72.5%; by end-2027: about 97%.
- Anthropic's market cap at the first-day close: median about $2.1T, with only 2.5% on $3T or more.
- OpenAI IPO by mid-2027: 36.5%; by end-2027: about 79%.
Reference class and model
Reaching $5T needs about a 2.4x rise from Anthropic's expected first-day value. OpenAI would need roughly 2.5–3.5x from its likely IPO value.
Historical base rates:
- Of companies that reached about $2T, only Nvidia went on to 2.4x within five years. Apple, Microsoft, Alphabet and Saudi Aramco did not.
- Among very large IPOs, the record is mixed: Facebook yes; Alibaba, Aramco and Uber no.
- Hot tech IPOs in general hit 2.4x within five years about 40–50% of the time.
My simulation:
- I simulated market-cap paths as a random walk, run day by day.
- IPO dates follow the Polymarket month-by-month odds. Anthropic's first-day value is centred on $2.1T. OpenAI starts at its $1.4T private value with a 15% first-day gain.
- Expected yearly return is 6–12%. Yearly price swings are about 45–60% in the first year, easing to 32–45%. The two companies' moves are 0.6–0.8 correlated.
- I counted the first close at $5T or more.
- Across the parameter sets, the chance by end-2031 ranged from 0.31 to 0.50, with an average of 0.39. By end-2027 it was about 9%, and by end-2028 about 20%.
- A scenario check gives about 0.40. If AI revenue keeps compounding (about 35% likely), a hit is around 85–90% likely. Under commoditisation (about 38%), it is around 12–18%. Under a bust or crackdown, it is about 3%.
Hazard shape
- Q4 2026 (about 0.6%): a $5T close would need roughly a 140% gain within weeks of listing. Markets put a first-day close at $3T or more at only 2.5%.
- 2027–28 (3–3.4% per quarter): this is the peak. Earnings reports, the end of the post-IPO lock-up, Claude 6 and an OpenAI listing all fall here.
- 2029–31 (declining to about 1.8% per quarter): the chance falls as the time left to cross shrinks.
Strongest counter-arguments
- Against YES: no company has ever gone from $2T to more than 2.4x quickly except Nvidia. Very hyped IPOs tend to underperform, and SpaceX has gone sideways. Rates are rising, and recent stock moves show the AI trade is fragile. Anthropic's governance may put safety first, US government orders have already taken its top models offline, and it carries $518B of fixed commitments.
- For YES: revenue grew more than tenfold in nine months and is already profitable on an adjusted basis. If the run rate merely slows to reach about $300B by 2028, $5T is likely. OpenAI gives a second chance, and inflation plus new share issuance push nominal market cap up.
These roughly offset, so I kept the model average.
Resolution criteria
Question. By the last day of each quarter, has OpenAI or Anthropic closed a trading day on a public exchange with a market capitalisation of at least US$5 trillion?
Definitions.
- OpenAI or Anthropic: either company's operating business, including a renamed or reorganised successor or a holding company established to list that business. A separately listed subsidiary counts only if it contains substantially all of that company's AI development and commercial operations. An unrelated acquirer's total market capitalisation does not count merely because it owns OpenAI or Anthropic.
- Public exchange: a regulated securities exchange on which the company's common equity, or depositary receipts representing it, is publicly listed and traded. An IPO, direct listing, or completed merger into a listed company can qualify.
- Close: the official closing price for the regular trading session on the security's primary listing exchange, including an officially shortened session. The first trading day qualifies. Intraday, pre-market and after-hours prices do not.
- Market capitalisation: total common-equity value, not free-float value, enterprise value or a fully diluted valuation. Calculate it using the official closing price multiplied by common shares outstanding, including restricted and insider-held shares but excluding treasury shares, unexercised options, warrants and unconverted securities. Use the most recently publicly disclosed outstanding share count applicable to that close, adjusted for publicly documented subsequent issuances, repurchases, conversions and splits effective by that close. Sum separately priced common-share classes; value unlisted classes with identical economic rights at the listed class's price, adjusted for their conversion ratio. Do not double-count cross-listings or depositary receipts. Where an unlisted class has different economic rights and no observable price or fixed conversion ratio, its speculative value cannot be used to reach the threshold.
- US dollars: nominal dollars. Convert non-dollar values using the Federal Reserve H.10 daily exchange rate for the trading date, or its latest preceding available rate; for currencies not covered, use the issuing central bank's published daily dollar reference rate on the same basis.
Resolves YES if a qualifying public record establishes a closing market capitalisation of at least $5,000,000,000,000 for either company on at least one trading day. Apply the resolution window and observation-period conventions: both the qualifying publication and the observed trading close must fall within the permitted period. If calculation requires several publications, the resolution date is the publication date of the last necessary input under A3. A rounded figure of '$5 trillion' alone is insufficient unless the source explicitly states that the threshold was reached or exceeded, or the underlying figures establish it.
Resolves NO / Exclusions. Resolve NO for a horizon if no qualifying closing valuation has been established by that horizon. Private funding rounds, tender offers, IPO offer prices before trading, proposed listings, acquisition prices, over-the-counter transactions, pre-IPO markets, synthetic tokens and derivatives implying a valuation do not count. Neither company's valuation may be added to the other's to meet the threshold.
Resolving source. Official closing-price records of the primary listing exchange (including Nasdaq's Official Closing Price or the NYSE official close), combined with share counts in issuer regulatory filings (SEC EDGAR Forms S-1, 424B4, 10-Q, 10-K, 8-K, or foreign equivalents) or official issuer capital announcements. Public closing-market-cap reports from Reuters, Bloomberg or the Financial Times also qualify if they identify the company, trading date and a value meeting the definition above; underlying exchange and issuer records take precedence in a conflict. Nasdaq's Market Activity methodology describes its closing price and shares-outstanding market-cap calculation; live quote-page market caps must be tied to the regular-session close to qualify. These specified financial records, rather than the A2 public-confirmation process, govern resolution. Apply A3 and the resolution-window and observation-period conventions.