How Venture Capital Exits Actually Happen: IPO vs M&A vs Secondaries, Year by Year

Venture capital exit routes determine when and how founders, employees and limited partners get paid, and this article explains how each route actually works year by year: the split between M&A and IPOs by count and by value, the rise of secondaries and tender offers at Stripe, SpaceX, OpenAI and Databricks, continuation funds, the SPAC boom and bust, direct listings, acqui-hires and PE buyouts, how holding periods lengthened and what the 2025–2026 reopening means. It completes the Sectors & Exits by Year pillar of the Kurums Venture Capital hub; the deal-level record is in our table of the biggest venture-backed exits by year, and the capital that needs exiting is quantified in where venture capital went in 2024–2026.

How is the exit mix split between M&A and IPOs, year by year?

M&A accounts for the large majority of venture exits by count in every year since 2015, typically 85–90% in the US, while IPOs and listings produce most of the value in open-window years: about half of 2025’s $549.2 billion global exit value came from roughly 10% of exits, according to PitchBook.

The pattern is structural. Thousands of venture-backed companies are acquired each year, usually for undisclosed sums below $100 million; Crunchbase counted more than 1,500 acquisitions of venture-backed companies worth $149 billion in 2020, with 41 of them above $1 billion. IPOs are rarer and lumpier. PitchBook-NVCA data show US exit value of $312.2 billion in 2019, $338.7 billion in 2020 and $841.5 billion in 2021, each driven by a handful of listings (Uber, Snowflake, Airbnb, Coinbase, Rivian), then $148.0 billion in 2022 and about $120 billion in 2023 when the window closed and M&A was the only route. 2024 was still M&A-led ($149.2 billion across 1,259 US exits), and 2025 saw both routes recover, with 1,636 US exits worth $297.8 billion.

The route also varies by geography. In Germany, trade sales make up about 92% of exits, as our analysis of Germany’s scaleup gap shows, and Europe’s exit environment remained soft through Q3 2026 even as its funding grew (KPMG). Asia’s exits are listing-heavy because of Hong Kong, Shanghai STAR and India’s NSE; Asian exit value exceeded $143 billion in 2023, more than the US and Europe combined that year. The US is the only market where both routes are deep, which is one reason US venture firms, reviewed in our guide to the top US venture capital firms, dominate global returns.

Sources: PitchBook-NVCA Venture Monitor (US exit value; 2023 restated), SPAC Research (SPAC counts), Crunchbase News, PitchBook Venture Secondary Market Watch, company announcements. Route shares are approximate.

How do IPOs and direct listings work as a venture exit, and when is the window open?

An IPO sells new (and sometimes existing) shares to public investors through underwriters, while a direct listing (Spotify 2018, Slack 2019, Palantir 2020, Coinbase 2021) lists existing shares without raising capital; both are available only when public-market multiples and volatility permit, which in the last decade meant 2019–2021 and 2025–2026.

For venture funds an IPO is rarely the moment of liquidity. Lock-ups typically hold insiders for 180 days, and funds then distribute shares in kind or sell down over quarters, which is why 2021’s record exit value turned into 2022 distributions of stock that had already fallen. The IPO pricing valuation is what determines whether preference stacks pay out and whether a company “exited above its last round”; Instacart (listed at about $9.9 billion against a $39 billion 2021 mark) and Klarna ($15 billion against $45.6 billion) show how that can disappoint late investors even in a successful listing.

Direct listings solved a narrower problem: companies with no need for primary capital and a large employee shareholder base (Spotify, Slack, Coinbase) could list without an underwriter discount or lock-up. They are uncommon because most companies want the primary raise and the price support of a book-built IPO. The 2025–2026 reopening has been conventional: CoreWeave (March 2025), Circle (June), Chime (June), Figma (July), Klarna (September), Netskope, StubHub and Figure in 2025, then SpaceX’s $75 billion IPO in June 2026, the largest ever, and Cerebras. Stout counts 186 US IPOs in H1 2026 raising $134.2 billion, with biotech listings alone ($4.5 billion across 13 deals) exceeding all of 2025.

The SPAC route deserves its own note. SPAC Research counts 248 SPAC IPOs raising about $83 billion in 2020 and 613 raising about $162 billion in 2021; most merged with venture-backed targets (Lucid, SoFi, Grab, Opendoor, Clover) at valuations that mostly collapsed in 2022, with redemption rates above 90% and SEC rule changes in 2024 closing most of the loophole. SPAC issuance in H1 2026 is the highest since 2021, according to Stout, but the structure is now used mainly by smaller, often non-venture companies. Founders considering a listing should prepare the equity story and data room well ahead; our reviews of virtual data room software and pitch deck software cover the tooling.

Why does M&A dominate, and how have acquisition structures changed?

M&A dominates because there are far more buyers than IPO windows: strategic acquirers, PE firms and, increasingly, other venture-backed companies buy thousands of start-ups a year, and the structure has shifted from cash acquisitions toward all-stock mergers, minority stakes and licensing-plus-hiring deals that avoid antitrust review.

The classic form is a strategic cash acquisition: Microsoft–GitHub ($7.5 billion, 2018), Palo Alto–Chronosphere ($3.4 billion, 2025), ServiceNow–Moveworks ($2.9 billion). Crunchbase counts 36 unicorn acquisitions worth about $67 billion in 2025, nearly nine times 2024’s roughly $7 billion, with cybersecurity and AI infrastructure the dominant categories. Below the unicorn tier, the typical acquisition is a sub-$100 million tuck-in that returns capital to seed investors and often little to anyone else.

Since 2022, regulators have reshaped the top end. Adobe’s abandoned $20 billion Figma deal (terminated December 2023 after UK and EU objections) and the FTC’s posture under both administrations pushed large acquirers toward structures that do not trigger merger review: Microsoft paid about $650 million to license Inflection’s models and hire most of its team (2024); Google paid about $2.7 billion for a Character.AI licence and its founders, and about $2.4 billion for Windsurf’s leadership and licence (2025); Amazon did the same with Adept; Meta bought 49% of Scale AI for $14.3 billion rather than the whole company (2025). Investors in those companies received cash through the licence payment and a shell company that kept the remaining business, which is an exit in economic terms but not in form.

2026 added a new structure at the top: the all-stock merger between venture-backed companies. SpaceX absorbed xAI at a $250 billion valuation in Q1 (announced in early February with SpaceX at $1 trillion), then agreed to acquire Cursor for $60 billion in stock in June after holding an option with a $10 billion break fee. Nvidia’s $12.9 billion bid for Hugging Face, AMD’s announced $8.2 billion deal for World Labs and Stripe’s $7.5 billion purchase of OpenRouter in Q3 2026 show that AI infrastructure and developer platforms now have the deepest strategic buyer pool. For founders, the terms that govern these outcomes (drag-along, liquidation preference, escrow, earn-outs) are set at the venture round, which is why reading the term sheet before it reads you matters at Series A, not at exit.

How did secondaries and tender offers become the third exit route?

Secondaries let existing shareholders sell to new investors without a company exit; PitchBook estimates US venture secondary volume at $106.3 billion in 2025 ($91.7 billion direct and $14.6 billion GP-led), and company-run tender offers at OpenAI ($6.6 billion, October 2025), SpaceX ($2.6 billion, December 2025), Stripe (at a $159 billion valuation) and Databricks have become routine.

There are three forms. Company-sponsored tender offers let employees and early investors sell a fixed amount to pre-selected buyers at a company-set price; Stripe pioneered this at scale with a roughly $6.5 billion employee tender in 2023 and has repeated it, most recently at a $159 billion valuation, up 49% from $106.7 billion about five months earlier (Augment). SpaceX ran tenders roughly twice a year before its IPO, and OpenAI’s $6.6 billion tender in October 2025 was the largest of that year. Direct secondaries are bilateral sales of shares or SPV interests on platforms such as Forge, Hiive, Caplight, Nasdaq Private Market and Augment; Caplight data cited by Chronograph show 83% of Q3 2025 trading volume concentrated in just 15 companies, and Hiive’s top-20 list is dominated by SpaceX, OpenAI, Anthropic, Stripe, Databricks and Anduril. LP-led fund secondaries, where an LP sells its fund interest, accounted for 22% of all LP-led secondary activity in H1 2025, up from 8–12% in the prior three years (Jefferies, cited by Chronograph).

Why it grew is simple arithmetic: the largest private companies chose not to list. Stripe, Databricks, SpaceX (until 2026), OpenAI and Anthropic each had enough late-stage and sovereign capital available that a listing was optional, so secondaries became the only liquidity for their employees and early funds. Dedicated venture-secondary funds (StepStone’s $3.3 billion vehicle is the largest cited, with Industry Ventures, Pinegrove and G Squared also active) raised capital to buy at discounts of 20–60% to last-round price for most companies and at premiums for the top names. The practical rule for founders: a tender offer with a company-set price, a cap on each seller’s percentage and a right of first refusal keeps the cap table clean; an unmanaged secondary market creates price signals you do not control.

What are continuation funds, PE buyouts and acqui-hires, and when do they apply?

Continuation funds let a venture firm move late-stage holdings into a new vehicle with fresh investors so old LPs can cash out (Lightspeed, NEA and Insight have used them); PE buyouts take profitable venture-backed companies private at modest multiples; and acqui-hires or licensing-hires return capital when a product fails but the team is valuable.

GP-led transactions, mostly continuation funds, reached $14.6 billion of US venture secondary volume in 2025 (PitchBook). The structure, borrowed from private equity, works when a fund is at the end of its 10–12-year life but holds positions (an Anduril, a Databricks) that the GP expects to appreciate; existing LPs can sell or roll, and the GP resets carry on the new vehicle. The conflict of interest is obvious, which is why LP advisory committees and fairness opinions are required, and why the clauses that govern these deals are covered in our guide to venture fund terms and LPA clauses.

Private-equity buyouts of venture-backed software companies became common from 2022 when public multiples fell: Thoma Bravo, Vista, Francisco Partners and others took former venture darlings private (Anaplan, Zendesk, Coupa, Qualtrics) and bought private growth companies with positive cash flow at 5–8x revenue. For founders of profitable, slower-growth software businesses, a PE buyout is often the most realistic exit above $500 million, and it increasingly comes with rollover equity. At the other end, acqui-hires value a start-up at roughly $1–3 million per engineer and typically return little to investors beyond their preference; the AI licensing-hires of 2024–2025 were the same mechanism at a scale (hundreds of millions to billions) that did return capital. Founders in that position should read why VCs say no, because a weak exit usually reflects an issue investors flagged earlier.

How long do venture-backed companies take to exit now?

Holding periods have roughly doubled since the 2000s: the 2025–2026 IPO cohort listed at 10 to 24 years of age, median time from first venture round to exit is now well past eight years on PitchBook’s data, and funds routinely extend beyond their 10-year terms to hold the largest positions.

The causes are the ones traced across this pillar: mega-rounds from 2015 (Vision Fund, crossover funds) made staying private cheaper than listing; tender offers from 2021 removed the employee pressure to list; and the 2022–2024 window closure added three years to every 2021-vintage company’s timeline. The results show up in fund performance. Carta’s Q1 2026 data report median realised DPI just above zero for 2019 and 2020 vintage funds and only a small fraction of 2017–2018 funds at 1x DPI, even though paper TVPI has risen for six straight quarters; Carta’s Q4 2025 report found about a third of 2021-vintage funds and under a quarter of 2022–2023 vintages had recorded any distribution at all.

Long holds change the economics for everyone. For LPs, a 1x DPI at year 12 instead of year 7 cuts IRR roughly in half for the same multiple, which is why distributions rather than marks drive re-up decisions in 2026 and why investing in venture funds as an LP now requires scrutiny of a manager’s secondary and tender-offer practice. For founders, it means employee liquidity must be engineered (tenders, structured secondaries, RSU refreshes) rather than waited for, and for early investors it means the first opportunity to sell, often a Series C secondary at a discount, may be the best one.

What does the 2025–2026 reopening mean, and how should founders and LPs decide?

The reopening is real but narrow: IPO proceeds in H1 2026 were more than five times H1 2025, yet the value is concentrated in SpaceX, AI infrastructure and a few fintechs, so the decision framework for most companies is still M&A first, secondaries for interim liquidity and an IPO only when scale, profitability and the window align.

A decision framework

Ask four questions. First, scale: public-market investors in 2026 expect revenue above roughly $300–500 million, growth above 25% and a credible path to free cash flow; below that, a strategic or PE buyer is the realistic counterparty. Second, buyer depth: security, developer tools, AI infrastructure, payments and vertical software have many strategic buyers who have paid premium multiples (Google, Microsoft, Cisco, Palo Alto, ServiceNow, Nvidia, AMD, Stripe, SpaceX); consumer and marketplace companies have few. Third, cap-table pressure: if early funds are past year 10 or employees have waited more than five years, a tender offer or structured secondary should precede any exit process. Fourth, window: run a dual-track process (IPO preparation alongside a banker-led M&A outreach) only when public comparables are stable, because the 2021–2022 record shows a six-month delay can halve a valuation.

For LPs, the framework is about the manager, not the company. Ask how a firm has handled liquidity in 2023–2026: has it sold into secondaries, run continuation funds, distributed stock in kind or held through lock-ups, and what DPI resulted per vintage. The pattern across this pillar, from the 2015–2020 build-up through the 2021 peak to the 2026 SpaceX IPO, is that exit value is cyclical and concentrated, so a manager’s ability to realise gains in the narrow windows matters more than its paper multiples. Browse the full Sectors & Exits by Year pillar for the year-by-year data behind this analysis.

Frequently Asked Questions