Who Pays for Failure? Venture Capital and the Uneven Economics of Experimentation

Venture capital attracts criticism for familiar reasons: too much money chasing fashionable ideas, companies scaling before they understand their customers, founders and investors protecting themselves while employees discover that their options are worthless. Look only at the visible failures and the verdict seems straightforward.

Yet venture capital also finances a kind of company that ordinary debt handles badly. A young firm may have little collateral, uncertain revenue, a prototype rather than a product and a serious chance of producing nothing valuable at all. Requiring regular loan payments while those questions remain unresolved can end the experiment before it produces an answer.

The usual defence of venture capital nevertheless states the case too neatly. Venture investors do not absorb failure on behalf of society. They construct portfolios in which losses are expected because a small number of exceptional companies may repay them many times over. The financial write-off lands first with investors, but the consequences spread through founders, employees, suppliers, customers, communities and sometimes public institutions that carried the earlier scientific risk.

Venture capital can make uncertainty investable. It does not make failure evenly shared.

What holds it up is not always visible. Editorial image generated by the author.

What Venture Capital Is Built to Carry

A bank lends against a reasonably predictable future. It looks for income, assets, collateral and evidence that principal and interest can be repaid. Venture capital replaces that fixed claim with ownership. If the company fails, the invested capital may disappear; if it becomes sufficiently valuable, the investor participates in the upside.

This structure suits a particular combination of uncertainty and opportunity: limited present revenue, the possibility of rapid growth, a market large enough to support an exceptional valuation and an eventual acquisition or public listing through which the fund can realise its return. The last conditions matter as much as the first. Venture capital is not patient money for every worthwhile experiment. It seeks uncertainty whose successful resolution can produce a large, privately capturable financial event.

The portfolio is designed around uneven outcomes. Research on venture returns describes payoffs as infrequent and highly skewed; a small number of extraordinary successes can matter disproportionately to the result. This changes the logic applied to an individual company. A project does not need to have been likely to succeed if the possible payoff was large enough relative to the capital placed at risk.

Failure can therefore be rational from inside the fund without becoming the purpose of the system. An unsuccessful company may reveal that a technology is not ready, that customers will not pay the proposed price or that a market cannot support the business. Other failures teach little. They copy crowded models, purchase growth before establishing demand or survive because neither founders nor investors want to acknowledge that the original thesis has collapsed.

Waste and experimentation are not synonyms. A useful experimental system changes course when evidence changes. Capital that only delays the write-down is not purchasing discovery; it is purchasing time before recognition.

Innovation, but Within a Narrow Band

The positive case for venture capital is substantial. A classic study of twenty US manufacturing industries from 1965 to 1992 found that venture-capital activity was strongly associated with patenting and estimated that venture funding contributed disproportionately to industrial innovation relative to its share of research spending. The estimate belongs to a particular historical sample and patents are not a complete measure of innovation, but the broader point remains plausible: venture-backed firms have helped commercialise technologies that might otherwise have remained small, specialised or inaccessible.

Investors can also contribute more than capital. They may recruit executives, arrange later financing, introduce customers and manufacturing partners, strengthen governance or force a technical team to confront assumptions it would rather postpone. A discovery that never becomes reliable, manufacturable, distributable or affordable has limited social reach. Commercial exploration and scale are real parts of innovation.

They are not the entire process. The science underneath a startup may have been developed in universities, public laboratories, corporate research departments or military programmes. Standards, infrastructure and public procurement may have created the market before a venture-backed company appeared. As the Journal’s essay on opening a technological field rather than merely harvesting it argues, venture capital is often strongest after a frontier has become legible enough to support ownership, a company and a plausible route to scale.

This is not merely opportunistic harvesting. Product design, manufacturing, distribution and business-model discovery can involve difficult technical work. The limitation is narrower: institutional venture capital selects for technologies capable of fitting its financial architecture. A major review of the evidence identifies the narrow band of innovations compatible with venture requirements as one of the model’s central concerns.

Projects with very long timelines, unclear ownership, modest eventual markets or benefits that spill widely beyond the company can be socially valuable without offering venture-scale returns. The system does not simply select the best ideas. It selects ideas that can become large financial events within the lifetime and structure of a fund.

Different Institutions Tolerate Different Failures

Public institutions also fund uncertain work. Accountability does not require avoiding failure; it requires a defensible reason for accepting the risk and a process for learning from the outcome. The current NSF SBIR and STTR programmes support small firms developing high-risk technologies and take no equity. NIH’s small-business programmes provide non-dilutive funding for early-stage biomedical research and development. ARPA-E is explicitly designed to pursue high-risk energy technologies in areas where markets cannot or will not yet go.

These programmes do not form a morally cleaner mirror of venture capital. Public funding can become bureaucratic, politically directed or slow to terminate weak projects. Venture funds can move quickly, impose discipline and connect technical work to markets. The meaningful distinction is not that one institution tolerates failure while another fears it. They tolerate different failures for different reasons.

The phrase “startups explore; institutions scale” is useful only as a first approximation. Startups can become scaling machines early. Large firms can conduct basic and applied research without a near-term product. Universities preserve questions whose applications remain obscure, while public procurement can create the first durable market for a new capability.

A healthier distinction separates the uncertainties being carried:

• Scientific uncertainty: whether the underlying phenomenon can be understood or produced.
• Technical uncertainty: whether it can be made to work reliably.
• Market uncertainty: whether anyone will buy or adopt it.
• Scaling uncertainty: whether it can be manufactured, delivered or operated economically.
• Financial uncertainty: whether success will generate a return large enough for the capital provider.

These uncertainties overlap, but they are not interchangeable. A technology can be scientifically promising and technically immature, commercially useful but too small for a venture fund, or socially valuable while offering no practical way for one investor to capture most of the benefit. An innovation system becomes brittle when it asks one mechanism to carry all five.

The Invoice Extends Beyond the Fund

Within a venture portfolio, the immediate answer to the title’s question appears simple: the investors lose the money committed to the failed company. Even that answer contains more actors than the word “investor” suggests. Venture funds frequently invest money supplied by pension funds, university endowments, foundations and other limited partners. The risk has already been pooled before it reaches the startup.

The losses outside the fund are less diversified. Founders may lose years of work, personal savings and much of their ownership through successive financing rounds. Employees lose jobs, deferred compensation and options that never acquire value. Suppliers may remain unpaid, while customers may have built operations around a service that abruptly disappears. Communities can subsidise offices, laboratories or factories that never reach the promised scale.

The same company failure is therefore not the same risk for everyone involved. Professional investors can diversify across many ventures, demand information, negotiate voting and liquidation rights and reserve capital for later rounds. A founder or employee usually has one company rather than a portfolio. Research on entrepreneurship consequently describes the founder’s stake as a large, non-diversifiable risk and finds that the expected financial reward often compares poorly with salaried work for people without substantial wealth. The ability to survive the loss matters as much as the willingness to take it.

This inequality is not limited to Silicon Valley employment contracts. As the Journal’s essay on why risk-taking depends on institutions that make failure survivable argues, people may have ambition and ideas while lacking credit, reliable infrastructure, secure rights or the ability to recover from one unsuccessful attempt. Venture capital can make a corporation’s experiment financially survivable. It does not automatically make personal failure survivable for everyone who works inside it.

Legal form and bargaining power decide where the loss stops. Limited liability protects shareholders from many claims beyond their investment, while contracts determine priority over remaining assets. Those rules are essential to making risky enterprise possible, but they also distribute the residue of failure. The invoice does not disappear because the company disappears.

Neither Countercyclical nor Neutral

Venture capital is sometimes described as a social shock absorber: a mechanism willing to finance uncertainty that the rest of the economy cannot bear. If that were fully true, we might expect it to become especially experimental when conventional finance retreats. The evidence is less reassuring.

A study of four decades of US patenting found that venture-backed innovation is strongly procyclical. During downturns, the quality and economic importance of VC-backed patents deteriorated as investors changed the kinds of startups they financed. Preserving cash for existing portfolio companies and favouring ventures nearer to revenue can be rational from the fund’s perspective. It also means that venture capital may become more cautious when patient experimentation is hardest to finance elsewhere.

The opposite environment creates a different distortion. When capital is abundant, several firms can be financed to pursue nearly identical markets. Spending can postpone the moment when weak demand becomes undeniable. Teams, geography and infrastructure expand before the underlying model has earned that scale.

Portfolio incentives sharpen the pressure. A stable, modestly profitable company may be an excellent outcome for founders and employees but too small to influence the returns of a large venture fund. Investors therefore have reason to preserve the possibility of an enormous outcome, even when a narrower strategy would give the company a better chance of durable survival.

This does not mean that every venture investor demands irrational growth or that every funded company must become a monopoly. It means the definition of success is shaped by fund size, ownership and return requirements. Venture capital selects not only for ideas worth attempting, but for ideas that can become sufficiently large financial events.

Progress Still Requires a Settlement

Venture capital performs a real and valuable function. It can finance companies that debt would destroy, tolerate repeated portfolio losses, support rapid iteration and help successful technologies reach a scale their inventors could not achieve alone. It is neither a universal innovation engine nor a stable public insurance system.

A mixed system remains necessary. Public research can investigate questions before ownership and markets are clear. Corporate laboratories can combine specialised infrastructure with longer horizons. Banks can finance predictable expansion. Venture funds can back companies whose uncertainty is too great for debt and whose potential commercial scale is unusually large. Every mechanism can waste money, and every mechanism can also become too cautious.

The difficult question appears when an experiment leaves the portfolio and enters ordinary life. Aggregate gains may be real while particular workers, suppliers and places bear losses that the aggregate never compensates. The history of containerisation, discussed in the Journal’s review of technological progress and the social settlement around it, offers a useful parallel: efficiency can increase while the transition distributes costs according to bargaining power rather than merit.

Innovation does not become socially legitimate merely because eventual benefits exceed total costs. Someone pays before those benefits arrive, and someone else may capture most of the upside after they do. The task is not to prevent every failure but to see its structure clearly enough to decide which risks deserve support and who should be protected from ruin.

That is why “who pays?” is more useful than asking whether venture capital is good or bad. Which uncertainty is the fund actually carrying? Which losses remain with investors, and which are transferred to people with less power? Which valuable experiments are never attempted because success would benefit too many people to make any one investor rich?

A system that celebrates risk without tracing its losses is not funding failure. It is hiding the invoice.