# Scaling versus profitability: evidence audit

**Research date:** 2026-09-05  
**Article audited:** Aswath Damodaran, [“The Scaling and Profitability Trade off: Venture Capital's Weakest Link!”](https://aswathdamodaran.blogspot.com/2026/09/the-scaling-and-profitability-trade-off.html)  
**Scope:** analytical and empirical synthesis, not investment advice.

## Executive assessment

Publication edition: the cash-flow formula and aggregation descriptions have been clarified following an additional source and methods review. The original research snapshots remain preserved locally.

Damodaran’s central business proposition is sound: revenue growth creates enterprise value only when the present value of incremental contribution and strategic option value exceeds acquisition, capacity, working-capital, and risk-adjusted financing costs. “Scale first” is therefore a conditional strategy, not a universal law. His emphasis on market size, contribution economics, capital intensity, competition, and economies of scale is directionally useful.

The article’s empirical case is much weaker than its operating logic. Several descriptive trends are real—private capital has expanded, later-stage companies can remain private longer, VC outcomes are highly concentrated, many recent IPOs are unprofitable, and VC fund performance is heterogeneous. But the article repeatedly changes the unit of analysis or metric and then makes causal claims the evidence cannot identify:

- The “top 1% produced 80% of returns” chart measures **gross company exit value**, not VC investment returns, fund returns, or net LP returns. It covers a partial 2023–2026 window and is dominated by SpaceX; excluding SpaceX, the source’s share falls from 80% to 45%. [CF Private Equity](https://www.commonfund.org/cf-private-equity/ai-is-steepening-the-power-law-in-venture-capital?hs_amp=true)
- Cambridge Associates’ figures are capitalization-weighted, pooled horizon IRRs net of fees and carry, compared with cash-flow-matched mPMEs. They do not describe a “median venture capitalist.” [Cambridge Associates methodology and CY2025 results](https://www.cambridgeassociates.com/insight/us-pe-vc-benchmark-commentary-calendar-year-2025/)
- Skewness alone cannot imply that “about a quarter” of VCs beat the average. That percentage requires a defined population, year, weighting, and actual distribution.
- The market chart is Kenneth French’s **short-term reversal** factor, formed on the prior month’s return. It is neither the long-term reversal factor nor the value factor, and therefore cannot establish that prices have stopped reverting to fundamentals. [French short-term reversal definition](https://mba.tuck.dartmouth.edu/pages/faculty/ken.french/Data_Library/det_st_rev_factor.html)
- Ritter’s IPO evidence supports older age, larger real sales, and lower profitability compared with the 1980s, but “median age rose **by** about 11 years in the last 15 years” confuses a level with a change. The annual all-IPO median was 11 in 2011 and 12 in 2025. [Ritter IPO data hub](https://site.warrington.ufl.edu/ritter/ipo-data/)

The most defensible conclusion is narrower than Damodaran’s: abundant private capital changes the feasible growth path and governance bargain, while portfolio payoffs and staged financing can create incentives to pursue upside. Whether those incentives cause value-destructive scaling is unresolved without company-level counterfactuals, contract terms, ownership, cash flows, and comparable non-funded firms.

## 1. The decision is incremental economic value, not growth versus GAAP profit

For a proposed customer cohort, geography, or capacity block, a useful starting test is:

$$
\Delta FCFF_t=\Delta EBIT_t-\Delta CashTax_t+\Delta D\&A_t-\Delta Capex_t-\Delta NWC_t.
$$

Compare expected incremental after-tax cash flows with the specified alternative, including the initial outlay. Capex includes maintenance and growth; working capital is operating working capital. Discount at a cost of capital consistent with the cash-flow risks. Count acquisition costs once and do not double-count execution risk through both a risk-adjusted rate and an extra blanket deduction. Any terminal residual must include remaining reinvestment needs or liquidation recoveries. Do not add unpriced strategic option value. Financing survival is a separate constraint; dilution affects the distribution of equity value, not mechanically operating enterprise value.

Scaling creates value when the expected incremental NPV is positive after allowing consistently for failure, competitive response, and the opportunity cost of capital; assess financing and shareholder dilution separately. Growth can rationally lower current accounting income when customer acquisition, R&D, or platform costs are expensed before the related benefits arrive. It is still destructive if retention, price, or cost assumptions cannot support positive cohort NPV.

Five measures answer different questions:

| Measure | What it tests | Main trap |
|---|---|---|
| Unit contribution | Revenue less costs that genuinely vary with serving the unit/cohort | “Contribution” definitions often omit support, implementation, refunds, cloud usage, or local operations. |
| Operating profit | Accrual revenue less operating expenses | R&D and acquisition expense timing can depress profit before uncertain benefits; adjusted profit can also omit real recurring costs. |
| Operating cash flow | Cash generated/used by operations, including working-capital timing | Upfront billings can flatter cash flow; stock compensation is added back under the indirect method but remains an economic dilution cost. |
| Free cash flow | Operating cash flow less investment required to sustain/grow the business | Definitions vary; excluding capitalized software, finance leases, acquisitions, or stock compensation may overstate economics. |
| Economic profit | After-tax operating profit less a charge for all invested capital | Requires judgment about capitalizing R&D/CAC and estimating risk-adjusted capital. |

The SEC’s financial-statement guide distinguishes operating, investing, and financing cash flows; issuing shares or debt is financing, not evidence that operations are self-sustaining. [SEC guide](https://www.sec.gov/about/reports-publications/beginners-guide-financial-statements) U.S. GAAP generally expenses R&D as incurred, so an analyst may capitalize selected R&D economically—but only with a failure-adjusted useful life, not by relabeling all spending as an asset. [FASB Statement No. 2 summary](https://fasb.org/page/PageContent?bcpath+=+tff&pageId=%2Freference-library%2Fsuperseded-standards%2Fsummary-of-statement-no-2.html)

## 2. When delaying accounting profitability is rational

Delay can be rational when all of the following are observable:

1. Mature cohorts have positive fully loaded contribution, acceptable CAC payback, and durable retention or repeat purchase behavior.
2. New cohorts resemble validated cohorts after controlling for channel, geography, customer size, and price—not just on blended averages.
3. Fixed platform or capacity costs are genuinely shared, so incremental margins improve rather than every market recreating the cost base.
4. Expansion protects or builds a scarce advantage—network density, data, distribution, standard setting, or switching costs—and the advantage survives subsidy withdrawal.
5. The company can finance to the next falsifiable milestone with a stress buffer; the plan does not require continuously rising valuation multiples.
6. Expected incremental return on invested capital exceeds the risk-adjusted hurdle rate.

Amazon is strong counterevidence to a simple “profits now” rule. Its 1997 letter paired rapid scale with repeat-order evidence, a large market, explicit return testing, a commitment to drop programs that failed, and a stated balance between growth, long-term profitability, and capital management. Sales rose from $15.7 million to $147.8 million and repeat-customer order share rose from above 46% to above 58%. [Amazon 1997 shareholder letter in SEC filing](https://www.sec.gov/Archives/edgar/data/1018724/000119312517120198/d373368dex991.htm) This was disciplined, testable reinvestment—not proof that losses are harmless.

Conversely, scaling is presumptively destructive when contribution remains negative after correcting the denominator, CAC payback exceeds plausible customer life, gross retention is weak, expansion markets deteriorate, support or local fixed costs grow with revenue, capex and working capital consume cash faster at larger scale, or the plan requires a financing event rather than an operating milestone.

## 3. What the VC evidence actually says

### Returns and concentration

VC performance must keep four levels separate:

- **company exit value**: headline equity value at IPO or acquisition;
- **investment return**: cash or fair value received relative to that investor’s entry basis and ownership after dilution;
- **fund return**: portfolio cash flows, residual NAV, fees, expenses, and carry, reported as IRR, TVPI, DPI, or PME;
- **LP return**: the LP’s actual net cash flows, sometimes altered by secondary transactions, credit lines, and timing.

The CF/PitchBook statistic uses U.S. VC-backed exits of at least $10,000 between 2005-01-01 and 2026-06-18. The top 1% account for 80% of **exit value** since 2023, or 45% without SpaceX. It is compelling evidence of transaction-value concentration, but it ignores entry price, ownership, dilution, liquidation preferences, fees, unrealized positions, and which funds held the winners. The microdata are proprietary, so the exact statistic cannot be independently reproduced. [Source and footnote](https://www.commonfund.org/cf-private-equity/ai-is-steepening-the-power-law-in-venture-capital?hs_amp=true)

Cambridge Associates reports that its U.S. VC index returned 21.1% in 2025, but also says results were concentrated and that the index’s comparison with Nasdaq has been inconsistent across horizons. Its CY2025 database comprised 2,816 funds formed from 1981–2024 with $764 billion of value. Pooled horizon IRRs are capitalization weighted and net of fees, expenses, and carry; public comparisons use mPME cash-flow matching. [Cambridge Associates](https://www.cambridgeassociates.com/insight/us-pe-vc-benchmark-commentary-calendar-year-2025/) These facts support “large heterogeneity and horizon sensitivity,” not “the median VC earns the displayed pooled return.”

Academic results reinforce measurement caution. Kaplan and Schoar found average net fund returns roughly matched the S&P 500 in an older Venture Economics sample and strong manager persistence, with substantial heterogeneity and poorer funds formed in boom periods. [Kaplan & Schoar (2005)](https://doi.org/10.1111/j.1540-6261.2005.00780.x) Korteweg and Nagel show that infrequent, skewed, cross-dependent VC cash flows make risk adjustment difficult; start-up investments can show positive abnormal returns while fund-level abnormal returns are near zero or negative under an option-based stochastic discount factor. [Korteweg & Nagel (2016)](https://doi.org/10.1111/jofi.12390) Later Burgiss research finds VC persistence survives when using information available at fundraising, although the result is not transferable to buyouts and does not mean every successful manager has repeatable company-selection skill. [Harris et al. (2023)](https://doi.org/10.1016/j.jcorpfin.2022.102326)

Nanda, Samila, and Sorenson provide an important alternative mechanism: one extra IPO among a VC’s first ten investments predicted up to an 8% higher later IPO rate, but early success arose substantially from being in the right place and time; it improved access to later rounds and syndicates. Persistence can therefore reflect reputation and deal access, not just forecasting skill. [Study](https://doi.org/10.1016/j.jfineco.2020.01.004)

### VC conduct and governance

The claim that VCs merely chase prices is too categorical. A survey of 885 institutional VCs at 681 firms found they rated deal selection most important, but also reported strategic guidance, customer and investor connections, operational guidance, and recruiting. It is self-reported and not causal evidence of value added, yet it directly contradicts a universal “VCs do not build businesses” claim. [Gompers et al.](https://www.nber.org/papers/w22587)

There is nevertheless credible concern. Lerner and Nanda identify a narrow band of innovations suited to institutional VC, concentration among decision-makers, and a recent relaxation of historically intensive governance. [Lerner & Nanda (2020)](https://doi.org/10.1257/jep.34.3.237) Classic contracting evidence shows VCs condition cash-flow, control, liquidation, and employment rights on perceived internal and external risks. [Kaplan & Strömberg](https://www.nber.org/papers/w8764) The net implication is contingent: staged capital and control can discipline experiments, while competition for scarce deals and founder bargaining power can weaken oversight.

## 4. Crossover capital and delayed IPOs

The article correctly identifies a blurred public/private boundary. Kwon, Lowry, and **Yiming Qian** document a sharp rise in mutual-fund investments in private VC-backed companies. Their 16-family sample grew from two mutual funds holding three private companies in 1995 to 92 funds holding 93 in 2015; mutual-fund participation supplied incremental primary capital and, in their instrumental-variable estimates, enabled firms to stay private about 2.8 years longer. [Authors’ research summary](https://corpgov.law.harvard.edu/2017/04/18/mutual-fund-investments-in-private-firms/) The published study evaluates delayed listing and fund return/IPO-allocation motives; it does not establish that FOMO is the dominant cause. [Kwon, Lowry & Qian (2020)](https://doi.org/10.1016/j.jfineco.2019.10.003)

Mutual funds are not equivalent to VCs. Contract-level evidence on unicorn financing finds mutual funds have weaker cash-flow rights, less board involvement, and stronger redemption rights, consistent with their own liquidity needs. [Chernenko et al. (2021)](https://doi.org/10.1093/rfs/hhaa100) This may change governance, but does not by itself imply poor investment quality.

A stronger causal design comes from Ewens and Farre-Mensa. They use the 1996 NSMIA deregulation to show expanded late-stage private capital helped firms grow larger and remain private longer, alongside increased founder bargaining power. They explicitly reject a simple IPO-market-failure story and caution that NSMIA is not the sole driver. [Ewens & Farre-Mensa (2020)](https://doi.org/10.1093/rfs/hhaa053) This supports the financing-channel claim, not the downstream claim that delayed IPOs cause bad business models.

## 5. Public markets: the passive-investing mechanism is unproven

French’s momentum signal uses prior months 2–12, while short-term reversal uses the prior month and long-term reversal uses prior months 13–60. Value (HML) sorts on book-to-market. These are different empirical objects. [French data library](https://mba.tuck.dartmouth.edu/pages/faculty/ken.french/data_Library.html) The article’s chart is explicitly short-term reversal. A weaker recent one-month bounce cannot be treated as the disappearance of fundamental valuation discipline.

The article offers passive ownership as one possible reason. Evidence is mixed and mechanism-specific. Index inclusion can reduce some price-informativeness measures, which supports a possible price-demand channel, but does not show that passive flows caused the article’s reversal pattern or private firms’ scaling decisions. [Coles, Heath & Ringgenberg working paper](https://www.nber.org/papers/w27593) In the other direction, a Russell-index discontinuity design finds passive mutual funds used their voting blocs to increase board independence, remove takeover defenses, and equalize voting rights, with improvements in longer-run performance. [Appel, Gormley & Keim (2016)](https://doi.org/10.1016/j.jfineco.2016.03.003) “Passive investment” therefore does not mean passive corporate ownership, and the article supplies no causal bridge from index flows to VC incentives.

## 6. IPO claims

Ritter’s public series covers U.S. operating-company IPOs and excludes, among other categories, SPACs, ADRs, REITs, banks, unit offers, and very small best-efforts offerings. Sample definitions matter. [Ritter data hub](https://site.warrington.ufl.edu/ritter/ipo-data/)

- **Age:** the median for all IPOs was 11 in 2011 and 12 in 2025. The long-period median was 11 for 2001–2025. The article’s wording “rose by about 11 years in the last 15 years” is unsupported; “rose to about 11” is plausible.
- **Real sales:** recent cohorts generally list with much greater real trailing sales than 1980s cohorts, but annual medians are cyclical (for example, the public table gives $1 million in 2022, $7 million in 2023, $57 million in 2024, and $180 million in 2025 in constant dollars). Averaging annual medians is not the same as the pooled median company.
- **Profitability:** the broad decline is supported. Older Ritter research reports that only 19% of 1980s IPOs had negative earnings, while recent annual data often show a minority profitable. Exact “less than one quarter over the last decade” depends on whether years or firms are weighted and on the period boundary. [Ritter review](https://site.warrington.ufl.edu/ritter/files/2016/01/A-Review-of-IPO-Activity-Pricing-and-Allocations-2002-08.pdf)
- **Tech versus all IPOs:** the article’s market-cap/float chart is titled “Tech IPOs,” but its period counts appear to correspond to the all-operating-IPO sample; Ritter’s public 1980–2025 tech series contains 3,365 IPOs plus 13 direct listings. The article also extends a “2020–2026” bin beyond Ritter’s public 2025 cutoff. Exact reconstruction is therefore not possible from the cited public tables.

SpaceX did complete a June 2026 IPO. It sold 638.9 million Class A shares at $135 including the overallotment and received $85.675 billion net. [SpaceX June 2026 10-Q](https://www.sec.gov/Archives/edgar/data/1181412/000162828026052535/spcx-20260630.htm) Applying $135 to the pro forma 7.380 billion Class A and 5.696 billion Class B shares gives about $1.765 trillion, so the article’s $1.8 trillion is a reasonable rounded **offer-price implied equity value**. [SEC offering disclosure](https://www.sec.gov/Archives/edgar/data/1181412/000162828026040874/spacexukfwp.htm) It should not be confused with the roughly $2.1 trillion first-day closing market capitalization. [Nasdaq](https://www.nasdaq.com/newsroom/spacex-makes-history-raising-85-billion-through-nasdaq-listing) OpenAI and Anthropic values mentioned in the article are prospective offering expectations, not completed IPO evidence.

## 7. Company examples: useful mechanisms, not representative proof

- **Google and Facebook:** Google reported 2003 revenue of $961.9 million and net income of $105.6 million before its 2004 IPO. [Google S-1](https://www.sec.gov/Archives/edgar/data/1288776/000119312504073639/ds1.htm) Facebook reported 2011 revenue of $3.711 billion and operating income of $1.756 billion before its 2012 IPO. [Facebook filing](https://www.sec.gov/Archives/edgar/data/1326801/000132680113000003/fb-12312012x10k.htm) They prove simultaneous scale and profit is possible with powerful economics, not that it is common.
- **WeWork:** the model had a genuine duration mismatch: long, largely non-cancelable property leases financed shorter member commitments. By 2022 it disclosed $27.9 billion of future undiscounted lease payments versus $2.5 billion of committed sales contracts, with average lease terms of about 15 years and average membership terms of 19 months. [WeWork filing](https://www.sec.gov/Archives/edgar/data/1813756/000181375623000016/we-20221231.htm) This is a strong warning about financing structure and governance, but a single failure cannot estimate the average effect of scaling.
- **Ferrari:** 2025 shipments were 13,640, revenue €7.146 billion, EBIT margin 29.5%, and industrial free cash flow €1.538 billion; management explicitly described “carefully-managed volume” and value over quantity. [Ferrari results filed with the SEC](https://www.sec.gov/Archives/edgar/data/1648416/000164841626000018/fnvfy2025results.htm) It supports the viability of constrained premium scale. A changing market-cap comparison with mass automakers would not establish that restraint caused value.

These cases should be used to generate hypotheses—retention, fixed-cost leverage, duration matching, scarcity pricing—not as winners-only proof.

## 8. Strongest counterarguments to the article

1. **Option value and learning:** early losses can buy information and strategic position. Under uncertainty, staged experiments with explicit stop rules may be more valuable than premature optimization of reported profit.
2. **Network and standard-setting races:** when value depends on liquidity, density, complementors, or developer adoption, moving slowly can destroy the option to compete. The relevant comparison is not “profit now versus losses,” but positive expected NPV of accelerated adoption versus the counterfactual of losing the market.
3. **Accounting mismatch:** R&D and CAC are generally expensed while their benefits may last. Current losses can coexist with improving economic profit—provided cohort evidence justifies capitalization.
4. **VCs can add governance and resources:** contracting and survey evidence show active monitoring, control rights, recruiting, and customer/investor introductions. The industry is heterogeneous.
5. **Private capital can be an efficient institutional response:** later private financing may reduce premature listing costs and let founders preserve long-horizon projects. Ewens and Farre-Mensa explicitly find the new equilibrium need not be an IPO-market failure.
6. **Public-market demand is not devoid of fundamentals:** passive investors vote and engage; active arbitrage, short sellers, creditors, acquirers, employees, and product markets remain information channels.

## 9. What remains unanswered

- What share of company-level exit value accrues to VC investors after entry price, dilution, ownership, preferences, and fees—especially with and without SpaceX?
- Within matched firms, does faster externally funded growth improve or reduce eventual DPI, survival, and economic profit? What instrument creates plausibly exogenous scale speed rather than capital availability alone?
- Do crossover investors weaken monitoring, or do they select later-stage firms with already validated economics? How do redemption and governance rights mediate outcomes?
- How persistent are current VC returns once stale NAVs, subscription lines, sector beta, downside covariance, and vintage timing are treated consistently?
- Has IPO profitability fallen within comparable industry/age/size cells, or mainly because listing composition shifted toward biotech and intangible-intensive firms?
- Does passive ownership causally change companies’ reinvestment or founders’ private-market strategies, rather than only prices or public governance?
- Which operating metrics predicted the successful transition from losses to self-funded growth ex ante, without conditioning on survivors?

## Bottom line

Use Damodaran’s operating checklist, but not his causal narrative as established fact. The business decision should be made at the incremental cohort/capacity level, with financing and governance treated as constraints. The evidence supports high return concentration, expanded private capital, delayed listings, and changing IPO composition. It does **not** show that 80% of VC investor returns came from the top 1%, that exactly a quarter of VCs beat the mean, that passive investing weakened fundamental anchoring, or that later IPOs caused value-destructive business models.

The accompanying claim audit records each verdict, the founder framework turns the economics into gates and thresholds, and the reproducibility folder preserves inputs and calculations.