US Venture Capital Report 2023 — The Reset

If 2022 was the year US venture capital deferred its adjustment, 2023 was the year the deferrals came due.

The mechanism was simply the passage of time. A company that raised in 2021 with two years of runway reached the end of it in 2023. A bridge note taken in mid-2022 with an eighteen-month horizon converted in late 2023. An extension round intended to buy time until conditions improved was exhausted before conditions improved. Each of these had converted a valuation problem into a time problem, and the time ran out.

The resulting year had a distinctive character. Headline funding totals fell, which was widely reported. Less widely reported, and more important, was that a large share of the year's actual activity did not appear in funding data at all: shutdowns, acquihires, quiet asset sales, and restructurings that never produced a press release.

Meanwhile, a constraint that had been secondary became primary. Venture funds raise from limited partners, and limited partners fund new commitments substantially from distributions returned by prior ones. With exits closed for a second consecutive year, distributions had collapsed. Institutions could not recycle capital, so new fund commitments slowed — and that flowed through to how much venture funds could deploy, regardless of how attractive the opportunities looked.

The one exception was artificial intelligence, which absorbed a share of deployment far out of proportion to its share of companies. That concentration is the origin of the bifurcated market described in the 2025 report.

When deferral expires

The 2022 report describes the bridge round as converting a valuation problem into a time problem. 2023 is where that trade was settled, and it is worth tracing precisely how.

A convertible instrument defers pricing but does not eliminate it. The note converts, typically at the next priced round or at a maturity date, and at that moment a price is set. The company's position in 2023 was determined by what had happened in the interval:

• If conditions had improved, the note converts at a better price than a 2022 down round would have produced, and the deferral was correct.
• If conditions had not improved, the note converts at a price no better — and often worse, since the company has now consumed more capital, is further from its 2021 milestones, and has less negotiating leverage.

Conditions had not improved. So a large share of 2022's deferrals resolved on worse terms than an immediate adjustment would have produced.

Several second-order effects compounded:

• Notes accumulate. A company that bridged twice converted both, and the combined dilution at conversion frequently exceeded what a single clean down round in 2022 would have cost.
• Discount and cap terms bite hardest in a falling market. A note converting at a discount to a lower price produces a lower conversion price still.
• The negotiating position had deteriorated. In 2022 the company had runway and could walk away. In 2023 it had weeks of cash.

A deferral is a bet that time will help. In 2023 that bet was settled, and it mostly lost. The companies that took the mark in 2022 generally ended up in better shape than those that waited.

That is a durable lesson and it runs against instinct. The cost of accepting a valuation correction rises the longer it is postponed, because postponement consumes the very resources — runway, leverage, optionality — that would have made accepting it survivable.

The activity that data does not capture

Funding totals for 2023 tell only part of the story, and the missing part is arguably larger.

Shutdowns. A substantial number of companies that had raised in 2020 and 2021 ceased operations in 2023. Shutdowns are poorly captured in venture data: there is no filing requirement, no announcement obligation, and companies often wind down quietly over months. The data records the absence of a new round, not the closure.

Acquihires and asset sales. Many companies were acquired for amounts below the capital invested, in transactions that were effectively an orderly wind-down with the team preserved. These are recorded as exits, which flatters exit statistics, but they returned little or nothing to investors.

Restructurings. Recapitalisations that wiped out prior preferred stacks, converted preferred to common, or reset option pools happened without generating a reported valuation.

Team reductions. Reductions in headcount extended runway without a financing event, changing the trajectory of thousands of companies invisibly.

The consequence is a systematic bias in how the year is remembered. A dataset built on announced financings measures the companies that raised money. It does not measure the ones that did not, and in 2023 the second group was where most of the change occurred.

For allocators, the practical implication is that portfolio-level analysis in a downturn should start from the number of companies still operating rather than from valuation marks. A fund whose marks held up while a third of its portfolio quietly ceased operating has not held its value.

The LP liquidity chain

The binding constraint on US venture capital in 2023 was not valuation and not opportunity. It was liquidity at the limited partner level, and the chain is worth setting out because it is the least visible link in the system.

The mechanism runs in a loop:

1. Venture funds return capital to LPs through exits — listings and sales.
2. LPs use distributions to fund new commitments, since most institutions operate with a target allocation rather than continuously adding new money.
3. New commitments fund the next generation of venture funds.
4. Those funds deploy into companies.

Break step 1 and the whole loop constricts. With exits largely closed since 2022, distributions had fallen far below the level needed to sustain commitment pacing. LPs faced a straightforward arithmetic problem: to make a new commitment, they had to either receive distributions, sell something, or find new money.

Compounding it was the denominator effect described in the 2022 slot-A report: with public portfolios having fallen while private marks lagged, private allocations were mechanically above target. Institutions were over-allocated to private markets on paper, at exactly the moment they were being asked to commit more.

The consequences were specific:

• Fund raising extended dramatically. Processes that had taken months in 2021 took a year or more.
• Re-ups were prioritised over new relationships, disadvantaging emerging managers regardless of performance.
• Fund sizes fell, and some managers skipped a fund cycle entirely.
• Deployment pacing slowed, with funds deliberately extending investment periods to preserve reserves for follow-ons.

This is why 2023 deployment fell even where investors saw attractive opportunities. The constraint was upstream of the investment decision, which is an unfamiliar situation for an industry accustomed to thinking of capital as the abundant input.

Concentration into a single category

Against a contracting backdrop, one category absorbed a share of deployment far out of proportion to its share of companies.

The immediate cause was the public emergence of large language models, which converted artificial intelligence from a long-standing research area into an investable category with visible product-market fit. But the degree of concentration had structural causes beyond the technology:

• Capital had somewhere to go. Funds with dry powder and a constrained ability to raise their next vehicle still needed to deploy. A category with obvious momentum absorbed that capital readily.
• The category justified large cheques. Model development is capital-intensive in a way most software is not, so a single company could absorb what would previously have been spread across many.
• Career risk ran in one direction. In a difficult year, backing the consensus category is defensible in a way that backing an out-of-favour one is not.

The consequence for everything else was severe. Capital is finite, and concentration into one category means withdrawal from others. Companies in unrelated sectors faced a market that had contracted more than headline figures suggested, because the headline included a category they were not part of.

This is the origin of the bifurcation described in the 2025 report. It began in 2023, deepened in 2024, and by 2025 had become sharp enough that market-level averages stopped describing any actual participant.

Secondaries become permanent

The most durable structural change of 2023 was the normalisation of secondary transactions and continuation vehicles.

Secondaries — the sale of existing fund interests or direct positions to a new buyer — had historically carried a stigma. A seller was assumed to be distressed. Continuation vehicles — where a manager moves an asset from an ending fund into a new one, allowing existing LPs to exit and new ones to enter — were viewed similarly.

2023 changed both, for a simple reason: with the primary exit route closed, these were the only routes available. Institutions needing liquidity sold fund interests. Managers holding good assets in ageing funds used continuation vehicles rather than forcing a sale into a bad market.

Volume grew enough that the stigma disappeared, and something more interesting followed. Having become normal, these instruments did not recede when conditions partially improved in 2024 and 2025. They became standard tools for managing duration and liquidity — used opportunistically rather than defensively.

That is a genuine structural change in how private capital works, and it originated in 2023 as a response to a problem rather than as a design decision.

What a portfolio review should have measured

2023's measurement gap — shutdowns, quiet sales and restructurings absent from funding data — has a direct consequence for how a private portfolio should be assessed, and it argues for a different starting point than the one most reviews use.

The conventional starting point is valuation. Total portfolio value, change since the last period, unrealised gain. This is what reporting systems produce and it is what LP communications lead with.

In a downturn it is the least informative number available, for reasons this report and the 2022 report establish: marks are set at events rather than continuously, positions that have not transacted have not been remeasured, and structure means a flat headline valuation can conceal a substantial economic change.

A more informative starting point is company status. Specifically:

• How many portfolio companies are still operating. Not marked, not valued — operating. A fund whose marks held while a third of its companies quietly ceased trading has not held its value.
• How many have raised in the last eighteen months, and at what. A company that has not raised in two years is either capital-efficient or unable to raise, and the distinction is the whole question.
• How many are at or past their expected runway. This predicts the next twelve months better than any valuation.
• How many raised a bridge, and how many raised more than one. Repeated bridging is the clearest available signal that a valuation problem has been converted into a time problem and the time is running out.
• How many have had a structured round. Preference stacks accumulate ahead of earlier investors, so a position's economic value can fall substantially while its headline valuation does not.

None of these requires a valuation judgement, which is their advantage. They are facts about the portfolio's state, and they were largely available to any investor who asked.

In a downturn, count the companies before valuing them. A portfolio's marks describe what was last agreed. Its operating status describes what is still there.

What an allocator could act on

Measure distributions, not exit announcements. Distributions as a percentage of net asset value measures capital actually returned and is the variable that determines commitment capacity. Announced transaction values do not. Bain publishes the figure free.

Recognise that the constraint moved upstream. In 2023, deployment fell in many cases despite investors seeing attractive opportunities, because the binding constraint was LP liquidity rather than opportunity quality. A constraint upstream of the investment decision is unfamiliar to an industry accustomed to treating capital as the abundant input, and it is not relieved by better selection.

Read category concentration as crowding out. Capital is finite. A category absorbing a disproportionate share means withdrawal from others, so companies in unrelated sectors faced a market that had contracted more than headline figures suggested. The headline included a category they were not part of.

Accept the correction early rather than late. The companies that took the mark in 2022 generally ended up in better shape than those that deferred, because deferral consumed the runway, leverage and optionality that would have made the adjustment survivable. The cost of accepting a correction rises the longer it is postponed, which runs against instinct.

Treat secondaries and continuation vehicles as standard tools. Having become necessary in 2023, they did not recede when conditions improved. They address a duration mismatch that exists in every market condition, which is why the change proved structural rather than a crisis adaptation.

Track first-time fund formation as a leading indicator of industry composition. NVCA/PitchBook reports it separately in its free quarterly summary. A manager who cannot raise a first fund now does not raise a third a decade hence, and the cost appears as a gap in the population of established managers long after the constraint has passed.

What 2023 established for US venture

• Deferral was shown to be costly, with the cost rising the longer the adjustment was postponed.
• Funding data was shown to miss most of a downturn's activity, because shutdowns and quiet sales are not recorded.
• The LP liquidity chain became the binding constraint, moving the bottleneck upstream of the investment decision.
• Category concentration began, initiating the bifurcation that defined the following two years.
• Secondaries and continuation vehicles became permanent infrastructure, the year's most durable change.