Public tech multiples, with SpaceX, OpenAI and Anthropic.
The Rule of X relates a company’s revenue multiple to revenue growth and EBITDA margin. In Dealroom’s current sample of 138 public B2B software companies, growth weighted 1.5 times explains 41% of the variation in forward revenue multiples after six ±2σ outliers are removed.
Each bubble is a public B2B software company, sized by enterprise value. Horizontal axis = 1.5 × revenue CAGR + EBITDA margin. Vertical axis = EV / NTM revenue. The worst-fitting ±2σ outliers (e.g. Palantir) are excluded.
Dealroom analysis · public multiples via multiples.vc (refreshed monthly). Outliers are the ±2σ worst-fitting names at X = 1.5; private companies are not shown here.
SpaceX, OpenAI and Anthropic sit outside this public-company sample. For the comparison below, Dealroom uses X = 3 across a wider basket of 160 public technology companies and does not remove outliers. The fit is weaker: R² = 0.30. The three giants are plotted but excluded from the regression.
Bessemer’s Rule of X analysis where X = 3. Horizontal axis = 3 × revenue CAGR + EBITDA margin, where CAGR is the 2-year forward growth rate, FY2026→FY2028 (consensus). Vertical axis = EV / NTM Revenue multiple.
Dealroom analysis · public multiples via multiples.vc · giants from latest funding rounds & reported revenue (Aug 2026); SpaceX at its listed market cap less net cash. The three giants are excluded from the regression and R².
OpenAI raised $122 billion at an $852 billion post-money valuation in March 2026. At the time, it said monthly revenue had reached $2 billion, or $24 billion annualised. Anthropic raised $65 billion at a $965 billion post-money valuation in May and said its revenue run-rate had crossed $47 billion.
More recent external tracking is higher. TickerTrends estimated July annualised revenue at $41.3 billion for OpenAI and $74.1 billion for Anthropic. The roughly $80 billion figure now circulating for Anthropic is an extrapolation of that curve, not a company disclosure.
The figures are not like-for-like. They annualise recent activity, and Sacra says Anthropic records cloud-reseller revenue gross, which inflates its top line relative to net-reporting peers. Anthropic is valued at 21x its disclosed $47 billion run-rate and 13x the tracker estimate. OpenAI is at 36x its March disclosure and 21x the tracker estimate.
Combination of analyst projections, company guidance, leaks and commentary. 2025 is actual; everything after it is an estimate.
Forecast range, $B — full-year booked revenue, not the year-end run-rate (at these growth rates the two differ by nearly 2×). Each box spans low → high per company; the line marks the best guess, with the company name labelled above each box. 2025 is an actual, so its box collapses to the marker. Shown on a like-for-like net basis: SpaceX & OpenAI report net; Anthropic is net equivalent (≈$110 / 153 / 204B in 2028E vs $210B gross best guess). Hover any box for exact figures. Sources as above.
Two distinctions matter when reading the chart. It shows full-year booked revenue, not the latest annualised run-rate. Anthropic is shown on a net-equivalent basis so it can be compared with SpaceX and OpenAI; its higher gross figures are retained in the table below. The forecast ranges are Dealroom estimates, not consensus.
Growth is the common feature. Dealroom’s midpoint estimates imply annual growth of 78–96% from 2026 to 2028. The evidence behind those estimates differs: SpaceX reports filed quarterly results, while OpenAI and Anthropic are based on company disclosures and external tracking. The table separates these measures.
Anthropic books the full cloud-partner bill, so where the figures differ its gross number leads and the net equivalent sits beneath it. A dotted underline marks a figure that is an estimate rather than a disclosure.
| SpaceXlisted · Nasdaq: SPCX | OpenAIprivate · filed 8 Jun 2026 | Anthropicprivate · filed Jun 2026 | |
|---|---|---|---|
| Value | |||
| Now | $1.75Tmarket cap, 7 Aug · EV $1.69T | $1Texpected at IPO, H2 ’26 – ’27 | $1.2Texpected at IPO, late ’26 – ’27 |
| Last priced round | $1.25TFeb 2026 · xAI merger | $852BMar 2026 · $122B raised | $965BMay 2026 · Series H |
| Revenue | |||
| Run-rate disclosed | $31BQ2 filed × 4 · TTM $23.0B | $24B$2B a month, Mar 2026 | $47B$34B netMay 2026 |
| Run-rate tracked | —files quarterly | $41.3BTickerTrends, Jul | $74.1B$54B netTickerTrends, Jul |
| Next 12 months estimate | $66B | $45B | $95B$69B net |
| 2026E | $38BH1 filed $12.5B | $30B | $57B$42B net |
| 2027E | $80B | $70B | $135B$99B net |
| 2028E | $120B | $115B | $210B$153B net |
| Rule of X | |||
| Revenue CAGR FY2026 → FY2028 | 78% | 96% | 92% |
| EBITDA margin forward estimate | +37%H1 2026 filed · 45% in Q2 | −45% | 0% |
| Rule of X 3 × CAGR + margin · the x-axis above | |||
Net equivalent restates Anthropic comparably with OpenAI: it books the full cloud-partner bill, so we strip about 27% (The Information / leaked OpenAI memo, Apr 2026) to approximate net revenue. Valuation, growth and Rule of X are unchanged — only the revenue levels differ. The Rule of X track runs 0–300%; the two ticks are the public basket's median and highest name. Sources: Dealroom.co, SpaceX Q2 2026 Form 10-Q and S-1; Reuters, Bloomberg, FT, The Information, CNBC.
SpaceX is the only company in the comparison with public financial statements. It reported $7.814 billion of revenue in Q2 2026, up 92% year on year. AI revenue was $2.561 billion, up 247%. Adjusted EBITDA was $3.538 billion, or 45.3% of revenue, while the company recorded a $541 million GAAP net loss.
For the first half, revenue was $12.508 billion and adjusted EBITDA was $4.665 billion, a 37.3% margin. SpaceX ended June with about $60.6 billion of net cash.
Dealroom’s current house view is $38 billion of revenue in 2026, $80 billion in 2027 and $120 billion in 2028. That gives 78% annual growth from 2026 to 2028. With the H1 adjusted EBITDA margin, SpaceX scores 271% on the Rule of X.
At a $1.69 trillion enterprise value, SpaceX trades at:
The public-company regression implies about 24.4x at SpaceX’s growth and margin. On Dealroom’s forward revenue estimate, SpaceX is therefore around 5% above the line. On filed trailing revenue, it is three times the public benchmark.
The conclusion depends on the revenue assumption. SpaceX looks broadly in line with fast-growing public technology companies if Dealroom’s forward case is right. It looks expensive on revenue already reported.
Annual revenue by reported segment, $B. 2020–2025 are history and do not change with the scenario.
SpaceX reports three segments and these are they. Connectivity = Starlink and Starshield: the national-security work sits inside this line per the prospectus revenue-recognition policy, so we do not break out a separate defense bar — no standalone Starshield revenue level is disclosed in any period, only year-on-year deltas, which reconcile against any assumed base. AI = xAI / Colossus compute (incl. rental to Anthropic, Google and Reflection) plus Grok and X. Space = launch, Dragon and other; Dragon is never broken out separately. H1 2026 filed: Q1 Connectivity $3,257M + AI $818M + Space $619M = $4,694M; Q2 Connectivity $4,291M + AI $2,561M + Space $962M = $7,814M. 2023–25 are the filed ASC 805-50 recast figures, which consolidate xAI and X back to 2023; earlier years are pre-merger estimates on a different basis and are not comparable. On Medium the totals reconcile to the table above ($18.7B in 2025 → $38B 2026E → $120B 2028E) and to the best-guess line of the range chart out to 2031; Low and Medium are a Dealroom house view, not consensus (FY2026 consensus is about $39B and Morgan Stanley models $45B; FY2027 consensus is ~$72B against our $80B). Elon’s plan is Musk’s own target path, not ours: $1T in 2030, of which the AI segment is about 90%. 2029–2031 are a taper, not a modelled year. No Cursor revenue is included — the $60B all-stock merger had not closed at the filing date.
The Rule of X treats SpaceX as one company. A sum-of-the-parts view separates connectivity, AI infrastructure, xAI and launch.
Morgan Stanley’s base case totals $3.95 trillion. It assigns $1.69 trillion to Starlink, $2.00 trillion to Enterprise AI, $0.16 trillion to xAI and Grok, and $0.11 trillion to the remaining space and launch business.
The illustrative Dealroom case totals $6.06 trillion. It uses $2.00 trillion for Starlink, $3.20 trillion for AI compute, $0.75 trillion for xAI, Grok and Cursor, and $0.11 trillion for other space and launch. The xAI, Grok and Cursor value remains below OpenAI and Anthropic, but reflects the competitive upside discussed by Friedberg and Gerstner. The main sensitivity is AI compute pricing.
Starlink’s current scale remains small relative to global telecom. A comparison shared by Christian Keil places its $11.4 billion annual revenue around Singtel and above Chunghwa Telecom and AIS. Direct-to-cell, and higher bandwidth demand from AI and robotics, expand the addressable market.
Rule of X = X × two-year forward revenue CAGR + EBITDA margin. The vertical axis is enterprise value divided by next-twelve-month revenue. Public-company data comes from multiples.vc. Dealroom calculated the regressions and prepared the private-company estimates. SpaceX uses its listed market value less net cash. OpenAI and Anthropic use their latest priced funding rounds unless the chart is switched to the expected-IPO scenario.
OpenAI and Anthropic forecasts, expected IPO values and forward margins are Dealroom assumptions, not company guidance. Adjusted EBITDA is a non-GAAP measure. Figures are current to 9 August 2026.