Venture Capital's Growing Funds, Shrinking Returns: The VC Crisis Behind the AI Boom

VC industry crisisfund size inflationuncertainty theoryAI boomVenture Capitalearly-stage investing
1 hour agoSource: blockweeks.com
Venture Capital's Growing Funds, Shrinking Returns: The VC Crisis Behind the AI Boom

Author: Jerry Neumann, former professor of entrepreneurship at Columbia University and veteran retired venture capitalist

Compiled by: Shenchao TechFlow

Shenchao Introduction: Benchmark and USV, once known for being "small and beautiful, strictly disciplined," have both recently substantially expanded their fundraising scale, with capital under management per person reaching an all-time high. This shift away from their original aspirations reflects the harsh reality that the entire venture capital industry is being swept up by the AI arms race and the concentration of massive capital.

Veteran investor Jerry Neumann, drawing on Knight's theory of "risk and uncertainty" and comparing it with the historical crisis of the 1980s, points out that when the growth rate of capital far exceeds the cognitive capacity of partners, institutions will inevitably abandon uncertain ventures and degenerate into pursuing certainty through leveraged buyouts or growth equity investments.

The excess returns of venture capital essentially stem from underwriting genuine uncertainty that no one else dares to touch. When the leading giants become addicted to the defensive war of capital scale and premiums, the traditional venture capital model is once again heading toward partial extinction, and the hope of nurturing the next technological era will fall back into the hands of those agile small funds that dare to bet on obscure unknowns.

Venture Capital

Abandoning Restraint: The Massive Pivot of Two Early-Stage Icons

In 2024, Benchmark, one of the most successful early-stage venture capital firms in history, raised a $425 million fund, while the firm's partners themselves contributed an additional $170 million. This means each of the five partners would on average need to manage nearly $120 million.

In their letter to limited partners (LPs) at the time, the partners explicitly stated that they deliberately kept the fund size at its previous level in order to "force discipline and accountability, thereby focusing on what truly matters."

However, in June of this year, Benchmark raised a total pool of up to $2 billion: including a $750 million early-stage investment fund and, for the first time in the firm's history, a $1.25 billion growth fund. Also managed by the same five partners, the per-person management scale suddenly surged to $400 million. What was once "what truly matters" seems to have quietly undergone a qualitative change.

Ten years ago, Fred Wilson, founder of Union Square Ventures (USV), a highly regarded early-stage investment benchmark over the past two decades, publicly wrote: "If USV were to manage a billion-dollar fund, I'm afraid I wouldn't sleep a wink at night."

Until 2024, USV's core fund size had always been strictly restrained between $125 million and $200 million. However, in October of this year, USV announced the completion of a $900 million new fundraise, comprising a $500 million core fund and a $400 million opportunity fund, steered by five investment partners. Wilson is dangerously approaching his former insomnia nightmare.

Benchmark and USV are notable because their claim to fame and their enormous returns are entirely built on precisely empowering nascent companies at the earliest stages. Choosing to raise large funds far exceeding previous scale is tantamount to testing the edge of betraying the very skill they are best at.

But this is not an isolated incident; they are merely riding the wave of the entire industry's sprint toward mega-funds in recent years. In 2025, the top 10 venture capital funds absorbed one-third of the industry's capital, far higher than 13% in 2021.

This trend of concentration at the top is still intensifying. So far in 2026, mega-funds exceeding $1 billion have swallowed 72% of total U.S. venture capital fundraising, compared with 34% in 2025 and only 16% a decade ago. But the institutions that received massive capital have not expanded their partner teams proportionally. More and more capital is being highly concentrated in the hands of a very few, and this has become a common picture across the entire industry.

Venture Capital

Chart: Data source: 2026 NVCA Yearbook, Pitchbook

To a large extent, venture capital firms' compensation is directly tied to the scale of capital they manage. For top-tier names with brilliant track records like Benchmark and USV, securing more assets under management for each partner based on past achievements is commercially rational.

But the underlying logic of venture capital has always rested on two pillars: first, investing in nascent micro-enterprises whose capital absorption capacity has an objective physical ceiling; second, each partner must personally and actively manage portfolio companies that constantly face the risk of early-stage mortality. When capital under management per person surges, one of these two pillars must break. After which critical point is it no longer even venture capital?

Risk and Uncertainty: The Fundamental Cornerstone of Venture Capital

The venture capital pioneers of the 1960s developed an effective strategy: build a portfolio of small trial-and-error bets and make subsequent large follow-on investments in companies that have proven their viability. Both elements are indispensable, because in the venture world, returns strictly follow an extreme power-law distribution, and startups face deep-rooted extreme uncertainty in their daily operations.

Frank Knight, the scholar who first proposed the economic theory of entrepreneurship, gave a classic definition of "uncertainty": when you know the probabilities of winning and losing, you face "risk"; when you have no knowledge of the probabilities at all, you face "uncertainty."

Mature, well-functioning conventional companies bear risk, while startups face pure uncertainty. When certainty exists, companies can apply to banks for loans; when facing calculable risk, secondary stock markets and private equity funds will step in to price it.

But when genuine extreme uncertainty arrives, when a company's foundation is built on variables completely unknowable to current human cognition, only venture capital can provide the solution. In short, this is the entire truth of how the venture capital model works. By diversifying across many micro-teams and controlling losses in stages, venture capitalists are able to harness uncertainty.

Well-managed established giants are not afraid of taking risks; insurance companies and banks buy and sell risk every day, but they require statistical predictability. Therefore, large companies systematically eliminate, at the institutional level, uncertain projects whose expected returns no one can rationally calculate.

This means that only those with no other choice will pursue fringe ideas full of uncertainty. These people with no other choice are entrepreneurs, whose very calling is born to bear uncertainty. This exploration avoids head-on confrontation with resource-rich giants and secures a narrow window of development for startups.

The most successful entrepreneurs in history have often harnessed uncertainty by creating entirely new markets that never existed before. The upside of entering an existing market is capped by the known size of the pie; but a brand-new market represents an entirely new space that no one has ever entered, and its ceiling cannot be estimated. When Apple was founded in 1976, no one knew how large the personal computer market would be.

This kind of unexpectedly positive surprise creates a long-tail distribution of extremely attractive business outcomes. The rational strategy in the face of extreme uncertainty is by no means to bet everything on one company upfront, but to build a diversified portfolio and retain the right to make follow-on investments through staged financing.

When There Is Too Much Money: The Double Curse of Partner Time and Returns

Entrepreneurs cannot juggle multiple uncertain ideas at once; fully polishing a single project already exhausts them. This is precisely the social division of labor that gives venture capital its meaning: venture capital firms manage uncertainty at the portfolio level, allowing frontline entrepreneurs to focus entirely on their own specific businesses.

Venture capital partners must manage their own investment uncertainty while also helping startup teams withstand external risks. But a partner's energy has an unbreakable physiological limit. The traditional rule of thumb is that the maximum number of companies a partner can deeply engage with as a lead investor and serve on the board of is typically only 10 to 12.

To increase the investment volume under each partner's name, institutions generally have only two paths: either increase the number of projects invested in, or cram more capital into each project.

If a firm chooses to increase the number of investment targets, partners will have to step back from the deep lead-investor role and instead rely on the due diligence judgment of external lead investors, drifting with the tide. But this free-riding behavior will fundamentally dismantle the mechanism of controlling the risk of company failure through deep empowerment.

Another approach is to increase the size of bets on individual deals: either pay a higher valuation premium at any cost in the early stage, or push capital into later rounds. However, whether paying abnormally high valuations in the early stage or buying mature targets in the later stage whose uncertainty has been substantially diluted, both will irreversibly erode the long-term return rate of capital.

Later-stage growth investing has in essence evolved into a completely different financial category. It manages conventional risks with known probabilities, rather than exploring pure uncertainty of the unknown. This also requires partners to expend a great deal of energy negotiating with external private equity buyers, severely crowding out the effort originally devoted to early-stage cultivation.

Echoes of History: The Sudden Death Crisis of Venture Capital in the 1980s

The phenomenon of each partner managing too much capital leading to a collapse in overall returns has already played out a complete and painful cycle in the history of venture capital.

In 1979, the United States amended the prudent man rule in the Employee Retirement Income Security Act (ERISA), clearing the way for the first time for massive pension fund capital to flood into the venture capital industry. Within just five years, the annual fundraising scale of U.S. venture capital soared sixteenfold, rocketing from $218 million in 1978 to $3.6 billion in 1983.

However, the pace of cultivating professional investment talent across the industry simply could not match the astonishing torrent of capital inflow. From 1978 to 1987, the assets under management per capita of venture capital industry practitioners tripled, surging from $4 million to $12 million.

The most direct consequence of this imbalance was soaring valuation bubbles. The Wall Street Journal wrote in 1983: "Valuations of young companies have risen so fast that traditional early-stage venture capital has found it extremely difficult to earn the 5 to 10 times returns it must." The industry's overall median net internal rate of return (IRR) fell in free fall from above 30% in 1982 to about 8% in 1988.

The bloated fund sizes triggered a more fatal chain reaction: there simply were not enough early-stage startups in the market to carry such enormous capital. The Wall Street Journal recorded in 1988: "The enormous size of the funds has in many cases reduced seed-stage investing from a core business to a dispensable marginal appendage."

Early seed-round investing as a share of total U.S. venture capital was halved, plummeting from 25% in 1980 to 12.5% in 1988. In response, many venture capital firms simply turned to leveraged buyouts (LBO). This category had no foothold at all in 1980, yet by 1986 it had swallowed 23% of all U.S. venture capital funds.

Multiple batches of established venture capital firms, including J.H. Whitney, Patricof, and Warburg Pincus, either substantially ventured into or completely transformed into LBO private equity institutions; even Sequoia Capital once attempted a leveraged buyout business.

Those firms that still kept the venture capital name began to move away from uncertainty and risk, turning instead to funding the consolidation of existing mature markets, homogeneous copycat competitors, and traditional retail businesses with no technological barriers. Office supply supermarket Staples was once a good investment, but in the following two years, investors rushed to pour nearly $200 million into at least 16 similar supermarket chains.

This quickly devolved into vicious price wars and a wave of closures. Because venture capital sought low volatility by attempting to eliminate uncertainty, it ultimately ended up with a complete mess overall.

As capital was massively withdrawn from the source of technology and poured into buyouts and mature industries, new companies genuinely engaged in core technological innovation plummeted precipitously. The number of new technology product companies established in the U.S. peaked in 1982 and then fell 28% by 1991; the number of new independent software developers plunged 76% over five years.

Facing the predicament of having no targets to invest in, committed capital flowing to venture capital was halved by 71% between 1989 and 1991. Marc Andreessen, describing the scene when he first arrived in Silicon Valley in early 1994, put it bluntly: "It was completely dead there, a dead stagnant pool of water."

In short, the once glorious venture capital industry, because of deregulation, received excessive capital far beyond its capacity to digest; when money grew faster than people who understood investing, returns began to collapse; institutions, in response, retreated and abandoned uncertainty, thus falling into a vicious cycle of even lower returns.

Repeating the Same Mistakes Under the AI Arms Race

Today, we stand at exactly the same historical crossroads. Based on the size of Benchmark's new 2026 fund, each partner would need to invest $400 million, more than three times what it deemed two years ago to be a size that could maintain focus. Each partner at USV now also needs to manage $180 million, likewise several times the capital managed in the past.

This is by no means an isolated case. After U.S. venture capital fundraising peaked at $223 billion in 2022, it plummeted to $67 billion in 2025. Although this year's data has rebounded somewhat, massive capital is irresistibly concentrating into a very small number of mega-funds. If history offers any warning, we are very likely repeating the eve of the painful period of the early 1990s.

No one deliberately destroys venture capital. The enormous expansion before our eyes is essentially an arms race that cannot be stopped. The explosion of generative AI has pushed capital thresholds and market valuations to extremely absurd heights, and it seems that only mega-funds with heavy capital are qualified to get an entry ticket.

Frontier labs routinely need billions of dollars to build data centers and train foundation models, and many emerging AI startups set sky-high prices in their first round of financing, such as Thinking Machines' first round of financing in 2025, which reached as much as $2 billion. But no matter how noble the original intention, this has fundamentally violated the basic commandments of venture capital.

OpenAI in its early days faced genuine uncertainty that no one knew. This pure unknown gave it years of unhurried time to explore underlying technology and the commercial closed loop, and it consumed only about $130 million in its first few years. In contrast, most AI companies emerging today have already stepped into a highly saturated, mature arena surrounded by giants.

They are able to raise huge sums because the underlying uncertainty has already been explored by pioneers; but at the same time, they must face strangulation by the world's most resource-rich technology giants. Venture capitalists seem to have returned to the 1980s, keen to gamble on known risks in red oceans, yet afraid to truly embrace unknown uncertainty.

Using sky-high valuations that underwrite uncertainty to buy conventional-risk companies caught in homogeneous fierce competition, the real return rate upon eventual exit is destined to be unable to make up for the expensive entry ticket. Just as General Motors acquired Cruise for more than $1 billion at the peak of the autonomous driving hype in 2016, only to be forced to shut it down in 2024, when the prosperity fades, the true colors are finally revealed.

Big Money Goes Left, Venture Capital Goes Right: The Next Generation of Pioneers from the Ashes

Venture capital is essentially about obtaining excess profits in the right-side long tail of a power-law distribution by underwriting extreme uncertainty. When the capital allocated to each partner in a fund is severely overloaded, they are either forced to compromise and make bad premium deals, or can only seek a sense of security by reducing uncertainty. The former is a bad business deal, while the latter is no longer venture capital at all.

In this sense, venture capital is once again entering a cycle of partial extinction. Devoting oneself to growth equity investment or leveraged buyouts is inherently blameless; the market does indeed need such asset movers. But the world equally cannot do without true venture capital. When large funds collectively drift away, small agile funds will surely take up the torch of exploration.

The rhyme of history is clearly echoing at this moment: Lightspeed has established a fund of more than $750 million specifically for acquiring existing old enterprises and forcibly wrapping them in AI; Thrive Capital spent $7.6 billion to leverage-buy the 150-year-old traditional home security company ADT.

The top former venture capital giants are spending a great deal of energy on asset packaging, restructuring, and mergers and acquisitions, while the pure venture capital forces focused on seed and early stages are undergoing severe shrinkage.

At the same time, the pure venture capital forces focused on seed and early stages are undergoing severe shrinkage. In 2025, first-time funds as a share of total U.S. venture capital plummeted to a historic low of 10.9%; early-stage micro-funds under $100 million also shrank to only 11.7%.

But remember that during the Great Depression of the 1980s, it was independent clear streams like Kleiner Perkins and Accel, which held fast to early-stage uncertainty, that ultimately transformed into the venture capital legends that dominated the era in the subsequent 1990s internet wave, leading humanity's magnificent innovation revival.

Venture capital did not truly become extinct in the 1980s, and it will not this time either. Eye-catching massive capital will eventually occupy media headlines, but the truly surging future pulse of the venture capital industry will always belong only to those few investors who, with clear-headed acuity, silently accompany entrepreneurs in exploring entirely new markets at the edge of the unknown.