Venture Math · Fall 2026

Series B & Scaling Product-Market Fit

Session 04 · Sept 23 · Nnamdi Iregbulem (Lightspeed)
Data: Carta & whoisnnamdi.com · re-visualized for classroom use
The filter — who gets to a B, and what the round looks likeTEAL = THE FIELD · CLAY = SERIES B

Of 4,369 startups founded in 2018, 643 had raised a Series B by 2025

Carta’s 2018 incorporation cohort, stage by stage. Right column: each stage as a share of the one before it. Seed → A keeps 64% of companies; A → B keeps 40%.

VS. STAGE BEFORE Incorporated in 2018 4,369 · 100% of cohort Raised a seed 2,464 · 56.4% of cohort 56% Raised a Series A 1,588 · 36.4% of cohort 64% Raised a Series B 643 · 14.8% of cohort 40% Raised a Series C 194 · 4.4% of cohort 30% Raised a Series D+ 50 · 1.1% of cohort 26% STATUS ON JAN 1, 2025 CLOSED 2,707 61.9% STILL PRIVATE 1,408 32.2% ACQUIRED 239 5.5% IPO 15 0.34% PRIVATE UNICORN 57 1.3% Closed + still private + acquired + IPO = the full 4,369. Private unicorns are a subset of “still private.”

Carta · US companies incorporated in 2018 · status as of Jan 1, 2025

40%
Why the funnel narrows here: the investor’s job changes. Nnamdi’s argument — building on AngelList’s return data — is that seed investors should index: back every credible deal, because the cost of missing one winner is unbounded. Later-stage investors have track records to judge and far fewer companies to choose from, so they select. The Series B is where selection starts in earnest. whoisnnamdi.com · “Why Don’t VCs Index Invest?”

Series B rounds halved, then came back

Median Series B primary round size peaked at $34M in Q4 2021, fell to $16M by Q1 2023, then rebounded. Bridges at B mostly sat near $8M.

$0M $10M $20M $30M Q1 2020 Q1 2021 Q1 2022 Q1 2023 Q1 2024 $34M $16M $27M $16M $6M $8M Q1 2024: primary $27M vs. bridge $8M — a bridge at B buys a few quarters, not a plan.

Carta · median Series B primary vs. bridge round size · Q1 2020–Q1 2024

The wait for a B is now ~2.4 years

Median years between primary rounds. Both gaps have stretched by about a third since 2018 — and A → B is now the longer wait.

2018 2019 2020 2021 2022 2023 2024 Seed → A 1.6 1.5 1.5 1.6 1.7 1.8 2.1 A → B 1.8 1.7 1.8 1.7 1.9 2.1 2.4 A → B: +33% longer than 2018 · ~29 months

Carta · 11,413 US startups raising priced seed or Series A rounds · 2018–2024

23%
What the founders still own after the B. The median founding team holds 23% of the company once its Series B closes. A quarter hold less than 13%; only one team in ten still holds more than 43%. That is the stake equity refreshers and founder secondaries — the second half of tonight — get negotiated against. Carta · founding-team ownership by stage · rounds raised 2021–2025
The price — what a B costs, and what the valuation actually measuresTEAL = EXPLAINED · CLAY = SERIES B & THE TAIL

What a Series B looks like in 2025

The median B prices at 2.6× the median A ($152M vs. $58.6M post-money), raises $25M, and sells 14.5% of the company — less than the A, on a much larger base.

Post-money valuation (log scale) $25M $50M $100M $250M $500M $1B Series A $58.6M Series B $152M $87.6M $295.7M Series C $360.9M Cash raised $0M $20M $40M $60M $80M $10.9M $25M $11.4M $37M $39M Dilution (% sold) 0% 5% 10% 15% 20% 25% 19% 14.5% 9.5% 20% 9.9% Bar = 25th–75th percentile · dot = median. Dilution is computed round by round — it cannot be summed across the rows.

Carta · 2,369 rounds raised by US software startups · Jan–Sept 2025

Growth and profitability explain only about half of a software valuation

Spread in public software multiples (total) vs. the part growth and profitability explain (explained), 2020–22. Financials “never explain much more than half”; adding the macro economy lifts it to ~80% at the peak.

0.0 0.2 0.4 0.6 0.8 “Dark matter” what growth and profitability do not explain explained by growth + profitability Jul 2020 Jan 2021 Jul 2021 Jan 2022 Nov 2021 peak Y-axis: variance of log revenue multiples, as published. Monthly points traced from the published daily chart — approximate.

Nnamdi Iregbulem · whoisnnamdi.com · “The Dark Matter of Software Valuations” · Jul 2022

Same financials, very different prices

Most trade at unremarkable multiples; a few sit far out in the tail. Controlling for financials, the 90th percentile traded ~75% above prediction, the 10th ~50% below.

0× 10× 20× 30× 40× 50× most trade at 10–20× a long tail past 30× Revenue multiple → · shape traced from the published chart, approximate

whoisnnamdi.com · “Dark Matter” essay · Jul 2022

Customer revenue is fat-tailed too

Power laws fit to 13 public software companies’ customer revenue all had α below 2. At α = 1.25 the top 20% bring ~72% of revenue, the top 1% ~40%.

0% 25% 50% 75% 100% top 0.1% top 1% top 10% top 100% top 1% → 40% top 20% → 72% his fitted range, α ≈ 1.1–1.8 α = 2, for comparison Dark line: α = 1.25. Share of revenue from the top X% of a company’s customers, computed as X1−1/α — the Pareto formula, not a trace.

whoisnnamdi.com · “Monetization is Fat-Tailed” · Oct 2020

α<2
Land-and-expand is an index fund. Land in as many organizations as cheaply as possible; every so often one turns into a Google-sized account and carries the revenue — the same logic as the seed portfolio on the front. The wins get bigger the more broadly you cover the market. Discussion — Nnamdi Iregbulem, Lightspeed · His 2022 work said financials explain about half a valuation. Which half does he still believe?