How Venture Capital Funds Work

Most people assume a VC fund is one thing. One account, one entity, one pool of money. It's actually three separate legal structures, and the separation is completely intentional — like a Russian nesting doll where each layer exists specifically to protect the one inside it.

Nearly every U.S. fund is organized as a Delaware limited partnership. Delaware gets chosen for the usual reasons: favorable business law, decades of legal precedent, courts that have seen every possible partnership dispute. The LP structure solves two problems simultaneously. It keeps taxes at the individual partner level rather than collecting them at the fund level first, and it separates liability cleanly between the people running the money and the people supplying it.

The three entities break down like this:

The Fund LP. The actual fund. It holds the portfolio investments, admits both the LPs and the GP, and is governed by the Limited Partnership Agreement (the LPA). Think of it as the vehicle itself.
The GP Entity. A separate LLC that controls and manages the fund. The GP entity is the one that takes on liability. It also contributes a slice of capital to the fund. That slice is small, somewhere in the range of 0.39% for larger funds up to around 1.2% for smaller ones, but it's real dollars at risk. The GP isn't just playing with other people's money.
The Management Company. This is where the team actually works. It employs staff, pays salaries, and receives the management fees. Operationally separate from both the fund and the GP entity.

Why bother with all this? Liability stays with the GP entity, not with the LPs. Tax obligations pass through to individual partners. Everything stays in its lane.

Access is also restricted by law. Funds operate under SEC exemptions. The most common are 3(c)(1), which limits the fund to accredited investors and caps beneficial owners at 100, and 3(c)(7), which requires qualified purchaser status, a meaningfully higher bar. This isn't gatekeeping for sport. It's a legal requirement.

Setting all this up costs real money. Legal fees for a standard U.S. Fund I typically run between $35,000 and $75,000. And the LPA, the governing document that encodes every economic term, every governance right, every investment parameter, will shape everything that happens over the next decade-plus. It is genuinely worth getting right the first time.

How LPs and GPs Actually Divide Labor, Capital, and Risk

The structure creates two very different roles. The differences are substantial.

Limited Partners are the capital providers. Pension funds, university endowments, insurance companies, family offices, high-net-worth individuals. They supply the vast majority of the fund's capital, usually upward of 98% of the total. In exchange, they stay passive. No investment decisions, no day-to-day involvement, no board seats. Their liability is capped at the amount of their commitment. They cannot lose more than they put in.

General Partners run everything. Sourcing deals, conducting due diligence, making investment decisions, sitting on boards, supporting portfolio companies through the rough patches, and ultimately returning capital to LPs. They contribute a small fraction of the capital but carry unlimited liability for the GP entity. Their compensation comes in two forms: management fees to keep the lights on and carried interest, their share of the profits.

The asymmetry is intentional. LPs have large pools of capital they cannot actively manage themselves. GPs have expertise and deal flow they cannot self-fund at scale. The structure brings both together in a way that aligns their interests.

The GP's co-investment deserves more credit than it usually gets. Even at less than 1%, GPs lose real money if the fund fails. It's not just foregone upside. It's actual principal at risk. That distinction changes behavior in ways that are hard to measure but easy to feel when things go sideways.

Why LPs Don't Wire Their Full Commitment on Day One

When an LP commits to a fund, they sign a subscription agreement for a specific dollar amount. But they don't wire that money upfront. The full commitment sits on standby until the GP actually needs it.

As the GP identifies and closes deals, it issues capital calls (sometimes called drawdowns) to LPs. These typically require fund transfers within 7 to 10 days of notice. The GP asks, the LP sends. This repeats throughout the investment period.

Why does it work this way?

LPs keep their capital deployed in other assets until it's actually needed, which improves their own portfolio efficiency considerably.
Idle cash sitting in a fund earns nothing and drags down IRR.
Capital deployment aligns with actual investment activity rather than some arbitrary fixed schedule set years earlier.

Deployment pace varies with market conditions. Funds from the 2020 vintage deployed roughly 60% of committed capital within 24 months. The 2022 vintage dropped closer to 43%. That's not a structural change in how funds work. That's GPs being more deliberate in a tighter market where valuations had stopped making sense to a lot of people.

Failing to respond to a capital call is a serious LP default. The LPA includes penalty provisions, and they get enforced. This is a system with real consequences.

One more thing worth having in your head: LPs have to track total paid-in capital, not just committed capital, to know when they've actually been made whole. The distinction matters more than most people expect until suddenly it matters enormously.

The "2 and 20" Fee Model: What It Funds, What It Incentivizes, and Where It Gets Weird

"Two and twenty" is shorthand everyone in finance knows and almost no one outside it fully understands.

The Management Fee (The "2")

The management fee is typically 1 to 2% of committed capital per year, charged throughout the fund's life. It covers operational costs: salaries, office space, due diligence expenses, legal fees. Running a fund is expensive in ways that aren't obvious from the outside.

Quick example: a $50M fund at 2% generates $1M per year in management fees. That sounds like a lot until you price out a team of five to eight experienced investors, support staff, travel, and legal work stretched across ten-plus years. It's not a windfall. It's a budget that requires real discipline to manage.

Step-down provisions are common. Many funds reduce the fee to around 1.5% after the initial investment period ends. You're managing a portfolio at that point, not hunting for new deals. The workload is lighter and the fee reflects that.

The key thing about management fees: they don't depend on performance. They accrue whether or not the fund returns a single dollar to LPs. That's both a feature and a recurring source of criticism.

Carried Interest (The "20")

Carry is 20% of profits above a hurdle rate. This is the GP's real payday.

The hurdle rate is a minimum return threshold that LPs must receive before carry kicks in. GPs only participate in profit above that bar. The distribution waterfall works in tiers:

100% of distributions return LP principal
Preferred return goes to LPs
GP catch-up
Remaining profits split 80% to LPs and 20% to GPs

GPs don't touch profit until LPs have been made whole. That ordering matters.

Where the Model Gets Complicated

Management fees sustain GP operations regardless of performance. Carry is the primary incentive. But a firm running a very large fund can generate substantial fee income on its own. When that happens, the GP's dependence on carry decreases. Critics argue this creates misalignment. A GP who can live comfortably on fees alone will be less hungry on returns. LPs in larger funds have been pushing back on this for years, and honestly they're right to.

The Fund Lifecycle: From First Close to Final Distribution

The average VC fund lifespan is approximately 13 years. The standard 10-year baseline almost always extends. Plan for it.

Phase 1: Formation and Fundraising

Before a GP approaches a single LP, they define the investment thesis. Target sectors, geography, stage focus. Then come the offering documents: private placement memorandum, LPA, subscription agreements. The "first close" allows deployment to begin while fundraising continues. The "final close" ends LP admission.

Phase 2: Investment Period (Roughly Years 1 Through 5)

Active deployment. Deal sourcing, due diligence, Investment Committee approval, capital calls. The GP's stage focus shapes everything here. Seed funds write lots of small checks. Growth funds write fewer, larger ones into companies with proven traction. Each approach carries a different risk profile and a different expected return distribution, and conflating them is a mistake that's easy to make and painful to correct.

Phase 3: Portfolio Management (The Long Middle)

Board seats. Follow-on investment decisions. Helping portfolio companies hire key executives. This phase is quieter than the investment period but demands just as much judgment, maybe more. Reserve capital management becomes critical. GPs have to anticipate which companies will need more runway and which ones won't make it. Getting that allocation wrong is expensive in ways that compound over time.

Phase 4: Harvesting and Exits (Years 5 Through 10-Plus)

Average time to exit ranges from six to nine years. Exit routes include:

M&A. The most common path. An acquirer pays cash or stock, and the fund distributes proceeds to LPs.
IPO. High profile, but a minority of exits. The fund distributes shares or sale proceeds after lockup periods.
Secondary sales. The fund sells its stake to another investor. Increasingly common as companies stay private longer. Usually at a discount, but it provides earlier liquidity for LPs who need it.

One- to two-year extensions are routinely granted to avoid forced sales at bad valuations. For deep tech funds, the timeline can stretch further still. The underlying company timelines don't conform to a neat ten-year window, and pretending otherwise usually costs everyone money.

Why Most Portfolio Companies Fail and a Tiny Number of Winners Carry the Whole Fund

VC funds don't generate strong returns because most of their bets pay off. They generate strong returns despite most of their bets failing. That distinction is the whole thing.

VC returns follow a power-law distribution. Not a bell curve. A very skewed curve where the vast majority of outcomes cluster near zero and a tiny minority of outcomes are enormous. In a typical fund of 20 to 30 companies:

1 to 3 investments generate 50 to 80% of total fund returns
5 to 8 investments return 1 to 3x invested capital
10 to 15 investments return less than invested capital, many going to zero

More than 40% of portfolio companies in older vintages returned less than cost. Up to 20% were fully written off. Nobody talks about those at the LP meeting.

The pattern holds at the vintage level too. A broad analysis of VC vintage performance found that roughly 80% of returns across the industry were driven by just 22 to 30% of vintage years. Five to seven standout years out of roughly two decades drove the overwhelming majority of industry-wide gains. Pick the wrong year to deploy a fund and it shows up in the numbers for a long time.

What this means for how GPs actually behave:

Every check is underwritten as a potential fund-returner, not a probable moderate winner.
GPs pursue ownership stakes large enough that one breakout company can return the entire fund.
Pro-rata rights, the right to participate in follow-on rounds, get fought for aggressively. Doubling down on winners is how the math works.
A portfolio of 20-plus companies isn't diversification in the traditional sense. It's a search strategy for the one or two outcomes that actually matter.

How Portfolio Construction Puts Power-Law Thinking Into Practice

Fund size sets the table for everything else. A $100M fund targeting 20% ownership has to write checks large enough to achieve that stake at entry. The math isn't optional.

Reserve ratios are a key planning tool. GPs typically set aside 40 to 60% of the fund for follow-on investments. Not to spread capital more thinly across more companies, but to double down on breakout companies when they start showing early signs of being the one.

Ownership matters more than entry price. A 5% stake in a company that returns 100x is worth more than a 20% stake in one that returns 3x. GPs optimize for the former.

Stage discipline isn't a preference. It's a constraint with real consequences:

Seed funds write many small checks and accept high failure rates in exchange for early ownership in companies that are eventually enormous.
Growth funds write fewer, larger checks into companies with proven traction. Lower failure rate, lower multiple potential.

Each stage has its own power-law profile. A growth fund trying to generate seed-fund multiples is playing the wrong game with the wrong tools, and you can usually tell pretty quickly when that's happening.

The concentration effect plays out at the portfolio level too. The top companies in a given vintage generate a disproportionate share of total value, and those companies often represented a fraction of a percent of total investment cost across all underlying portfolio companies. The math is uncomfortable until you've seen it enough times that it just becomes how you think.

Portfolio construction is a probabilistic bet. The GP is engineering maximum exposure to the right tail of the distribution. Everything else is just the cost of getting there.

The Metrics LPs and GPs Use to Track Performance Before Exits Happen

The honest problem with VC metrics is that for most of a fund's life, you genuinely don't know how it's performing. Exits are sparse and slow. So the industry uses three proxies, each with blind spots worth understanding.

IRR (Internal Rate of Return) is the annualized return factoring in the timing of cash flows. It's sensitive to when capital was called and when distributions were made. A quick small win early in a fund's life can inflate the number before the real story plays out. IRR is useful context. It is not the verdict.

TVPI (Total Value to Paid-In) is the total value of the fund, realized plus unrealized, divided by total capital paid in. A TVPI of 2x means LPs are on track to double their money. But it includes unrealized marks, which are paper valuations that may or may not survive contact with an actual exit. TVPI can be managed by a motivated GP. It happens.

DPI (Distributions to Paid-In) is actual cash or stock returned to LPs divided by paid-in capital. This is the only metric that cannot be managed. Cash is cash. A DPI above 1.5x on a mature fund is a real signal. A DPI sitting near zero on a fund that's eight years old is a different kind of signal entirely.

Recent data on 2017-vintage funds puts median IRR around 11.5%, median TVPI around 1.72x, and median DPI around 0.27x. That gap between TVPI and DPI is the story: most of the value in even a mature fund is still locked in unrealized positions. Nobody's gotten paid yet — which, in VC, is a bit like saying the meal was excellent but the waiter still hasn't brought the check.

Liquidity is genuinely scarce right now. Significant portions of recent vintage funds had generated little to no LP distributions by the end of Q1 2025. That's an industry-wide condition driven by a slow exit environment that has persisted longer than most people expected, rather than a fund-level problem in most cases.

Dispersion between funds is enormous. The gap between a top-quartile fund and a median fund in the same vintage is far wider in VC than in almost any other asset class. Benchmarks are useful for orientation but they are not the whole story. Picking the right manager, in the right vintage, with the right thesis is still the actual job. The metrics just tell you how the current bet is tracking.