US Venture Capital Report 2010 — The Funnel Nobody Sized

2010's defining feature is a mismatch between two stages of the same market.

The cost collapse described in the US Venture Capital Report 2009 meant a company could reach a working product on a few hundred thousand dollars. A new class of investor emerged to write those cheques, and the number of companies receiving a first institutional investment rose sharply.

The capital available at the next stage did not rise correspondingly, and could not have, for reasons that are structural rather than a matter of sentiment:

• Series A is written by larger funds whose cheque sizes are set by fund size and partner capacity, per the arithmetic in the 2009 report.
• Those funds' capacity is limited by partner time, since a Series A investment normally comes with a board seat and sustained involvement.
• And fundraising for those funds had contracted following 2008, so the pool was flat or shrinking while the stage before it expanded.

The result was a funnel with a widening mouth and a fixed neck. If seed investments double and Series A capacity is unchanged, the proportion of seed companies that raise a Series A must halve. This is arithmetic, not a forecast, and it was available to anyone who cared to divide one number by the other.

It was not widely acted upon, because the consequence lags the cause by roughly eighteen months — the time between a seed round and the attempt to raise the next one. The companies funded in 2010 discovered the problem in 2012, which is when it acquired a name.

The year's second theme is instrument design. The convertible note became the standard seed instrument because it let both sides avoid negotiating a valuation. This was genuinely efficient — valuing a company with no product and no revenue is close to arbitrary. It also meant nobody knew how much of the company had been sold until a later round converted the notes, and the accumulated dilution surprised founders repeatedly.

The arithmetic of the funnel

The bottleneck is worth setting out formally, because its predictability is the point.

A venture market has stages with independent capacity constraints:

• Seed capacity is set by the number of seed investors multiplied by their investments per year. Because cheques are small and involvement is light, one investor can do many.
• Series A capacity is set by the number of Series A funds multiplied by investments per partner per year. Because cheques are large and involvement is heavy — usually a board seat — this number is small and hard to increase.

The graduation rate is the ratio between them, and it is determined by capacity, not by company quality:

If seed-funded companies double and Series A slots stay fixed, the graduation rate halves — regardless of whether the companies are better or worse than last year's.

This has a consequence that is easy to state and was widely misread at the time. A company that failed to raise a Series A in 2012 was not necessarily worse than one that succeeded in 2010. The bar had moved because the ratio had moved. Attributing the outcome to company quality alone confuses a capacity constraint with a selection judgment.

Why Series A capacity is genuinely hard to expand:

• Partner time is the binding constraint and cannot be scaled by adding capital. A partner can serve on a limited number of boards, and this number has not changed materially in decades.
• Adding partners is slow and dilutes the firm's decision quality and culture, which is why most firms grow partner count very cautiously.
• And raising a larger fund makes the problem worse, not better — a bigger fund needs bigger cheques, which pushes it later, further from the bottleneck.

The eventual resolutions took years and none of them expanded partner capacity: more seed funds grew into Series A funds, the definition of Series A drifted to larger and later rounds, and a new "seed extension" stage appeared to hold companies that were not ready. The US Venture Capital Report 2011 and 2012 cover this playing out.

The instrument that deferred the question

The convertible note — and later the simple agreement that replaced it — became the standard seed instrument for good reasons and with real consequences.

What it does: the investor provides capital now, and instead of buying equity at an agreed price, the investment converts into equity at the next priced round, usually at a discount and often subject to a valuation cap.

Why it was adopted so quickly:

• It avoids valuing an unvaluable company. A business with no revenue, no product and two people has no defensible valuation, so negotiating one consumes time and goodwill to produce an arbitrary number.
• It is cheap and fast. A note can be a short document executed in days; a priced round involves legal work costing a meaningful fraction of a small seed round.
• It allows a round to close incrementally. Investors can be added over weeks without renegotiating, which suits a founder raising from many small sources.
• And it defers the governance question — no board seat, no protective provisions, no formal control.

The costs accumulated quietly:

Dilution became invisible. A founder raising several notes over eighteen months, each with a different cap, often could not say what percentage of the company had been sold — the answer depended on a future valuation that had not happened yet.

Caps became de facto valuations anyway. The mechanism intended to avoid pricing produced a price, negotiated with less rigour than a priced round would have received.

And the conversion could surprise. Several notes at low caps converting at once could take substantially more of the company than anyone had modelled, occasionally leaving founders with less than they believed after a successful round.

The instrument did not remove the valuation question. It moved it to a moment when nobody was negotiating and everybody was celebrating.

The later standardisation of the instrument — with clearer conversion mechanics and post-money caps that make dilution calculable at signing — fixed most of this, and the US Venture Capital Report 2019 covers the eventual settlement.

What an accelerator actually solved

Accelerators expanded substantially in this period and are frequently described as a cultural phenomenon. They were a structural solution to a specific economic problem.

The problem: at a cheque size of tens of thousands of dollars, the cost of evaluating and supporting an investment exceeds any plausible return from that investment individually. Diligence, negotiation, legal work and ongoing help cost roughly the same whether the cheque is $50,000 or $5 million. At the small end the economics do not work.

The accelerator model addresses this by batching every expensive activity:

• Selection is batched. Applications are reviewed in cycles against each other, which is faster and produces better relative judgments than evaluating companies one at a time.
• Support is batched. Standard advice — incorporation, hiring, pricing, fundraising mechanics — is delivered once to a cohort rather than repeatedly to individuals.
• Terms are standardised. No negotiation, identical documents, near-zero legal cost per company.
• And introductions are batched into a demo event, which is the genuinely clever part: it converts the hardest and most time-consuming service — access to the next round's investors — into a single scheduled occasion, and creates competitive pressure by presenting companies simultaneously.

The peer effect is a real and underrated input. Founders in a cohort learn more from each other than from mentors, because their problems are contemporaneous and specific. This is a service that costs the accelerator nothing to provide and is difficult to replicate outside a batch.

The return model is explicitly portfolio-based. Most companies in a cohort will not matter. The model works if a small number produce large outcomes, which requires enough companies to make that plausible — hence large cohorts and standard terms.

The structural consequence is that accelerators became a significant contributor to the funnel widening described above, and the US Venture Capital Report 2012 covers what happened when the model was widely copied without the network that made the demo event valuable.

Signalling risk

A cost specific to the new structure emerged in this period, and it is a genuine information problem rather than a matter of etiquette.

The setup: a large fund that normally leads Series A rounds also makes small seed investments. For the fund this is cheap optionality — a small cheque buys information and a relationship.

The problem arises at the next round. When the company raises a Series A, other investors observe whether the existing investor is leading it.

• If it leads, that is a strong positive signal. The best-informed investor, with full access to the company for a year, is committing at scale.
• If it does not lead, that is a strong negative signal — and it is negative regardless of the reason.

The asymmetry is what makes it costly. The fund may decline for reasons entirely unrelated to the company: portfolio construction, a conflicting investment, a partner leaving, the sector falling out of favour internally. But outside investors cannot distinguish these from a negative judgment, and the safe inference is the negative one.

The seed investment gave the company a small amount of capital and an option that only its holder could exercise. When the option is not exercised, everyone learns something — whether or not there was anything to learn.

The rational responses that emerged:

• Founders became warier of seed money from large funds, or negotiated explicit understandings about the next round.
• Some large funds separated their seed activity into distinct vehicles with distinct branding to weaken the signal.
• And dedicated seed funds could credibly claim an advantage — a fund that never leads Series A rounds sends no signal by not leading one.

This is a general property of staged financing with informed insiders, and the US Venture Capital Report 2023 describes the same dynamic operating at much later stages when insiders declined to support flat and down rounds.

More companies, different distribution

The expansion in company formation had two distinct effects that are routinely conflated, and separating them is necessary to read the period correctly.

Effect one: average quality fell. With a lower capital barrier, more marginal companies were started. This is close to definitional — if the cost of attempting falls, attempts with lower expected value become worthwhile, so the mean of the population declines.

Effect two: the number of exceptional outcomes rose. A larger population, even with a lower mean, contains more observations in the extreme tail — provided the tail is not itself thinned by the expansion.

These effects are not in tension, and both were real. The confusion in contemporaneous commentary came from treating "quality is falling" and "this is a great period for venture" as contradictory claims. For an asset class whose returns come almost entirely from the tail, only the second effect matters.

Why the tail was not thinned: the cost collapse did not make exceptional companies harder to build. It lowered the barrier at the bottom without lowering the ceiling, so the distribution widened at one end.

The implication for investors depends entirely on strategy:

• A concentrated fund making few investments needs to identify the tail in advance, and a larger, lower-mean population makes that harder — more noise around the same signal.
• A high-volume seed fund benefits from the larger tail count, since it is buying exposure to the distribution rather than selecting from it.

Both strategies became viable, which is why the industry bifurcated in this period rather than converging — a split the US Venture Capital Report 2014 traces through to its consequences.

What an allocator could act on

Divide seed investment counts by Series A capacity. The graduation rate is set by that ratio and is calculable eighteen months before it bites. A widening mouth on a fixed neck guarantees a shortage.

Do not read a failure to raise as a quality judgment when capacity has moved. A company rejected in a tight year may be stronger than one funded in a loose one; the bar is a function of supply.

Ask what a seed instrument's conversion actually does. Multiple notes at different caps can convert to substantially more dilution than modelled, and the calculation should be run at signing rather than discovered at the next round.

Price signalling risk when a large fund takes a small position. The seed cheque carries an implicit option, and the failure to exercise it is read as negative regardless of the real reason.

Match strategy to the shape of the population. A larger, lower-mean company population makes concentrated selection harder and high-volume exposure more attractive. These are opposite conclusions from the same fact.

Separate average quality from tail count. For an asset class where returns come from the extreme, a falling mean alongside a rising tail count is a favourable development, and treating the two as one number gets the sign wrong.

What 2010 established

• Stage capacity constraints are independent, so seed expansion without Series A expansion mechanically lowers the graduation rate.
• Partner time, not capital, is the binding constraint at Series A, which is why the bottleneck could not resolve quickly.
• Convertible instruments deferred rather than removed the valuation question, accumulating dilution surprises.
• Accelerators made the smallest cheque viable by batching selection, support, legal cost and introductions.
• Signalling risk became a structural cost of taking capital from an investor who could lead the next round.