US Venture Capital Report 2009 — The Vintage Nobody Wanted

2009 presents a puzzle if you look only at the flows. Fundraising fell hard, investment activity fell, and the industry contracted. On every measure of activity it was a bad year.

It was also, on the evidence of the following decade, one of the strongest vintages in the asset class's history. Both statements are true, and the relationship between them is causal rather than coincidental.

The structural reason crisis vintages perform well:

• Less capital is competing, so entry valuations are lower for the same company.
• Fewer competitors are funded in any given market, so the companies that do get funded face less well-capitalised competition.
• Talent is available and cheaper, because larger employers are cutting.
• And the companies that seek funding in a hostile market are differently selected — the marginal, opportunistic company does not start in a year when starting looks foolish.

Each of these improves returns, and they all arrive together. The US Venture Capital Report 2008 describes the mechanism that produces them: the denominator effect forces institutions to cut commitments exactly when these conditions obtain.

The more important development in 2009 had nothing to do with the crisis. The cost of building and launching a software company had been falling for several years and by 2009 had fallen far enough to change the asset class structurally.

Three things had happened: infrastructure became rentable by the hour rather than purchased upfront; the software stack became substantially open source and therefore free; and distribution moved online, so reaching customers no longer required a sales force.

The consequence is arithmetic. A company that once needed several million dollars to reach a first product now needed a few hundred thousand. A venture industry organised around writing several-million-dollar cheques had a structural problem — and the response to that problem is the origin of the seed ecosystem that dominates the following decade.

Why the bad years are the good vintages

The claim is widely repeated and rarely mechanised. It deserves the mechanism, because the mechanism tells you which parts are reliable.

Four separate channels operate, and they are independent:

Entry price. With less capital chasing deals, the same company is funded at a lower valuation. This is the most direct channel and the easiest to over-weight — it matters, but venture returns are driven by outcome size more than entry price, so a 30% valuation improvement is not the main event.

Competitive intensity. This is the underrated one. In a capital-rich year, a promising market attracts five funded competitors, and they compete away each other's margins by bidding up customer acquisition and salaries. In a capital-poor year the same market gets one or two. The winner's eventual economics are substantially better, and this compounds over the company's life.

Input costs. Engineering talent, office space, and customer acquisition are all cheaper when larger employers are contracting, as the US Venture Capital Report 2008 notes. A dollar of investment buys more company.

Selection. In a hostile funding environment, the opportunistic founder does something else. The population of companies started is drawn from a different distribution — smaller, and more heavily weighted toward people who would have started regardless.

A good vintage is not a year when investors were smarter. It is a year when there was less competition for the same opportunities, at every level of the stack simultaneously.

Which parts are reliable? Competitive intensity and input costs are the durable channels, because they affect the company's whole life. Entry price is real but smaller than it feels, and selection is the hardest to evidence — it is a reasonable inference rather than a measured effect.

And the timing problem is the reason this knowledge is not acted upon. The vintage's returns arrive eight to twelve years later. The commitment decision must be made in the year when everything looks worst and no evidence is yet available, which is precisely when the institutional machinery described in the 2008 report is pushing the other way.

The cost of a company collapsed

The structural change of the period had nothing to do with the crisis and everything to do with three technology shifts arriving together.

Infrastructure became a rental. Serving a web application had required buying servers, contracting rack space and provisioning for a peak that might never come — a large capital expense before the first customer. Cloud infrastructure converted this to an hourly operating cost that scaled with usage, meaning a company could serve ten users for almost nothing and a million users by paying for a million users' worth of capacity.

The software stack became free. Operating systems, databases, web servers, languages and frameworks had substantial licence costs. By 2009 a complete, production-grade stack was available as open source at zero licence cost, and it was in many cases better than the paid alternative.

Distribution moved online. Reaching customers had required a sales force, a channel or advertising. Search, and increasingly social platforms, allowed a product to be found and adopted without any of them — which removed both a large fixed cost and a long lead time.

The combined effect on the capital required to reach a first product was roughly an order of magnitude.

Why this mattered more than the crisis:

• The crisis was cyclical. Capital returned within a few years.
• The cost collapse was structural. It never reversed, and it compounded.

The consequences reshaped the asset class:

• Companies could test an idea before raising institutional capital, which moved the point of first investment much earlier and made it much more uncertain.
• More companies could be started, since the capital barrier fell — which increased volume and lowered the average quality of the pool.
• And the information available at the first cheque changed. An investor could now see a working product and early usage rather than a plan, which is a genuinely different underwriting problem.

The Asia-Pacific Investment Report 2015 and India Venture Capital Report 2017 describe the same cost collapse arriving in markets where it mattered even more, because local capital was scarcer.

Why a large fund cannot write a small cheque

The cost collapse created a structural problem for the existing industry, and the arithmetic is worth setting out because it explains the entire shape of what followed.

A venture fund's economics are constrained by three things:

• Fund size determines the total capital to deploy.
• Partner capacity is limited — a partner can serve on a manageable number of boards and give real attention to a limited number of companies. This is a hard human constraint, not a policy.
• Ownership targets matter because returns depend on owning enough of the winners to move a large fund.

Now apply the cost collapse. A company that needs $500,000 rather than $5 million presents a large fund with a problem:

• Deploying a $300 million fund in $500,000 cheques requires 600 investments. At any plausible partner capacity this is impossible.
• Writing a $5 million cheque into a company that needs $500,000 means either buying an unreasonable share of the company or leaving the capital idle on its balance sheet, which is bad for both sides.
• And a $500,000 investment cannot move a $300 million fund even if it returns fifty times, so the effort is not justified by its contribution.

A fund's minimum viable cheque size is set by its size divided by the number of companies its partners can actually serve. When the market's required cheque falls below that number, the fund cannot participate — regardless of how good the opportunities are.

So a gap opened between what companies needed and what the existing industry could provide, and it was filled by a new kind of investor: small funds, often run by one or two people, frequently former founders, writing small cheques into many companies with limited governance involvement.

The structural features of this new model:

• Small funds, where a modest exit is meaningful.
• Higher investment counts with less involvement per company.
• Speed and lower diligence, appropriate when the cheque is small and the information is thin.
• A different return model — accepting that most investments fail entirely, in exchange for enough shots at the outcomes that matter.

This is the origin of the seed ecosystem, and the US Venture Capital Report 2010 covers what happened when it scaled.

The supply and demand of capital, briefly inverted

2009's unusual configuration is worth naming precisely, because it is rare and recognisable.

Normally, capital supply and investment opportunity move together. Good conditions attract capital and generate opportunities simultaneously; bad conditions remove both. This correlation is why timing an entry into private markets is hard — the cheap years are cheap because there is less worth buying.

2009 broke the correlation.

Capital supply contracted sharply, for reasons described above and in the US Venture Capital Report 2008 — reasons entirely internal to the institutions providing it, and unrelated to the quality of what venture funds could buy.

Opportunity expanded simultaneously, because the cost collapse meant more could be built for less, and because a genuine platform shift was beginning.

A configuration where capital falls while opportunity rises is the definition of a favourable vintage, and it does not require any forecasting skill to identify — both variables are observable at the time.

What made it unactionable in practice was not information but structure:

• Institutions were mechanically constrained by allocation policy.
• The evidence to justify a contrarian commitment would not exist for a decade.
• And the career risk of committing into a collapse is asymmetric — being wrong alone is far more costly than being wrong with everyone.

The Private Markets Outlook 2016 covers how some institutions built processes specifically to overcome this, principally by removing the discretion — committing a fixed amount per vintage year regardless of conditions, which mechanically forces counter-cyclical behaviour without requiring anyone to make a brave call.

What the vintage actually bought

The 2009 and 2010 vintages deployed into the early stage of a genuine platform transition, and it is worth being clear about what that means, since the term is used loosely.

A platform shift changes where software runs and how it is distributed, which resets the competitive position of every incumbent. The transition beginning in this period moved computing to devices that were carried rather than sat at, always connected, aware of location, and distributed through a controlled store rather than the open web.

Why platform shifts are the dominant source of venture returns:

• Incumbents are structurally disadvantaged, not because they fail to see the shift but because their existing business, distribution and organisation are optimised for the old platform. Adapting means damaging the thing currently paying the bills.
• New categories become possible that were not merely impractical before but incoherent — a service depending on knowing where the user is has no meaning on a desktop.
• Distribution resets. Established relationships with customers do not transfer, so a new entrant and an incumbent start closer to level than at any other time.

The timing feature that matters for allocators: the shift was clearly visible by 2009 and the returns from it were realised between roughly 2012 and 2021. The gap between observability and realisation is the whole problem — the shift was not a secret, and knowing about it was worth nothing without the willingness to fund into it for a decade.

The US Venture Capital Report 2010 and 2011 cover the resulting deployment, and the Global Investment Outlook 2023 examines the same structure in a later platform transition, where the same observability-versus-patience gap appeared again.

What an allocator could act on

Commit on a fixed pacing schedule rather than a discretionary one. The mechanical approach forces counter-cyclical commitment without requiring anyone to make a brave call in the year when everything looks worst.

Weight competitive intensity above entry price when assessing a vintage. Fewer funded competitors in a market improves the winner's economics for its entire life; a lower entry valuation is a one-time effect and a smaller one.

Watch for capital supply and opportunity moving in opposite directions. They normally move together, so the divergence is rare, observable at the time, and the clearest available vintage signal.

Check whether a fund's size permits the cheques its market requires. A fund's minimum viable cheque is its size divided by the companies its partners can genuinely serve, and a fund writing below that number is either over-owning or under-attending.

Separate cyclical from structural change. The crisis was cyclical and reversed within years; the cost collapse was structural and never did. Allocation decisions made on the first were wrong, and on the second were the decade's most consequential.

Accept that the evidence will arrive too late to be useful. A vintage's returns are known eight to twelve years after the commitment, so any process requiring evidence before acting will systematically commit to the wrong years.

What 2009 established

• Crisis vintages outperform through four independent channels — entry price, competitive intensity, input costs and selection — of which the middle two are the durable ones.
• The cost of building a software company fell by roughly an order of magnitude, structurally and permanently.
• Fund size sets a minimum viable cheque, which opened a gap the existing industry could not serve.
• The seed ecosystem emerged to fill that gap, with different economics, diligence and involvement.
• Capital supply and opportunity diverged, a rare and observable configuration that institutional machinery was structurally unable to act on.