An IPO should be Plan A. It’s the best outcome for almost everyone on the cap table.
But founders, employees and even a lot of VCs treat the IPO date as the liquidity date. For most of the cap table … most of the money comes years later. Start the clock at the first check and the lead investor in a great IPO is usually 10 to 13 years in and still holding stock.
Meghan Reynolds of Altimeter highlighted it on SpaceX: roughly 40% of private investors’ shares are distributable, and LPs have received far less. The filings back her up, and SpaceX is typical.
The Median Tech Company Takes 11.5 Years to IPO. A Fund Lasts 10.
PitchBook puts the median time to IPO for tech companies at 11.5 years. The standard venture fund term is still 10. And 45% of unicorns have already been held in portfolios for nine years or more.
Few companies get there. US VC-backed exits in H1 2025 totaled 649: 472 acquisitions, 150 buyouts and 27 public listings. That’s 4% going public.
First Check to Last Share: Six Clocks
In all five IPOs, the lead VC still holds stock.
Snowflake: 9.5 Years, and That’s the Best Case
- Sutter Hill incubated Snowflake in 2012.
- Mike Speiser was its CEO from 2012 to 2014.
- At the September 2020 IPO, Sutter Hill owned 20.3%, worth about $12.6 billion on less than $200 million invested.
- Its first filing after the IPO showed 32,399,403 shares, 11.4% of the company.
- The next one, as of December 31, 2021, showed 3,925,889 shares, or 1.3%.
This is one of the fastest examples of almost complete VC liquidity in the modern tech IPO era. 8 years to IPO, then 15 more months to get down to 1.3%. And Snowflake’s stock held up through the whole distribution window.
Rubrik: 12.4 Years In, 82% Out
- Lightspeed led Rubrik’s Series A in 2014, the year it was founded.
- Going into the April 2024 IPO it held 38,984,426 shares, 23.9% of the company.
- On December 31, 2025, Lightspeed funds held 9,525,413 Class B shares.
- On June 30, 2026, they held 7,144,060.
- Form 4s from this summer show more in-kind distributions.
Lightspeed is 82% out, 26 months after the IPO and 12.4 years after the first check. At the recent pace of 2.4 million shares per six months, the last 18% takes about 18 more months.
Founder Bipul Sinha has sold far less. He held 12,342,646 shares before the IPO. In September 2026 he still held 10,679,839 Class B shares directly, about 87%. That month he entered a prepaid variable forward on 500,000 shares, a structured product to get cash from a small slice, 12.7 years after founding.
Samsara: Still Distributing 45 Months After the IPO
Samsara was founded in 2015 and went public in December 2021, a fast six and a half years.
The distributions have still run long. a16z entities were making in-kind distributions and open-market sales in September 2025. That month’s filing showed 4,512,974 shares left, down from 19,889,079 in the prior one. That was 10 years after the first check.
Co-founder John Bicket reported 98,314,290 shares in February 2026, down 4.07% from 102,480,719 a year earlier. He’s 11 years in, four years public, and selling about 4% of his stake a year.
Figma: 13 Years In, 50.3 Million Shares Still in the Fund
- Figma was founded in 2012 and Index led the seed in 2013. It IPO’d July 31, 2025 at $33 and closed its first day at $115.50.
- A month later, holders of about 54.1% of Class A shares signed an extended lock-up through August 31, 2026. It released in four steps: 17.5%, 20%, 27.5%, then the remainder.
- That final block was 77.7 million shares, released in August 2026 with Figma at $23.97, about a $12.6 billion valuation. It closed Friday at $21.36. Adobe offered $20 billion in 2022.
- Index has barely started distributing. On August 11, 2026, Index Ventures VI distributed 2,758,691 shares and still held 50,293,428 directly. That one distribution was about 5% of the fund’s stake.
Dylan Field sold 2.35 million shares in the IPO and 3,029,063 more in the three months before the January 2026 unlock. That’s about 5.4 million shares, roughly 9% of his stake, for something like $190 million. A year after the IPO he still holds about 53 million shares, close to 90% of what he started with.
SpaceX: 24 Years to IPO, 8.5% Distributed
SpaceX was founded in 2002. It priced at $135 and started trading June 12, 2026. Instead of a 180-day cliff, it used a staggered schedule:
- Up to 20% after Q2 earnings, plus 10% more if the stock was up 30%
- 7% more at each of days 70, 90, 105, 120 and 135
- Up to 28% more after Q3 earnings, with everything released at 180 days in mid-December
That’s 41% unlocked so far, which is Meghan’s “~40% distributable.”
Valor Equity Partners, an early backer, reported distributing 42,790,223 shares in kind in a Form 4 dated September 11. It still held 460,624,307. Valor has distributed 8.5%.
The rest is slower:
- An extended lock-up covers about 7.8 billion shares, roughly 60% of the company, including all of Musk’s. It releases in slices tied to Q4 2026, Q1 2027 and Q2 2027 results.
- A first-layer SPV gets 30 days to distribute after it receives shares. Each layer below waits on the one above.
- Musk’s shares are locked until June 12, 2027 with no early release. He’ll be 25 years past founding before his first sale.
The price moved the whole time. By late July the stock was down about 50% from its post-IPO peak. It closed last week at $158.96, about 51.6% above its low.
Wiz: 6 Years From Founding to $32 Billion in Cash. Case Study of How M&A Can Be So Much Faster.
Wiz was founded in 2020. Google signed an all-cash $32 billion deal on March 18, 2025. It closed March 11, 2026.
That was a slow deal. The regulatory review alone ran 12 months across the US, EU, Australia, Israel and others. Founders, seed investors and employees still all got paid on the same day, six years in.
M&A holds some money back too, but far less. From SRS Acquiom’s 2026 study of 2025 deals:
- 88% of private-target deals had an escrow or holdback
- Median escrow was 10.0% of deal value without RWI and 2.8% with it
- 24% had an earnout, with median earnout potential at 34% of the closing payment
- Earnouts have paid about 21 cents per potential dollar
Deals also die. Adobe and Figma signed in September 2022 and terminated in December 2023, 15 months of limbo for a $1 billion breakup fee.
How Founders Actually Sell: 4% to 8% in the IPO, Then a Few Percent a Year
Founders have more ways to sell than most people think. They still end up selling slowly.
- They can sell in the IPO itself. Circle’s Jeremy Allaire sold 1.58 million shares in the June 2025 IPO, about 8% of his stake. Co-founder Sean Neville sold 11%. Field sold roughly 4% of his in Figma’s. In 2025, 13.8% of the shares in VC-backed IPOs came from existing holders, up from an 8.5% long-run average.
- They can sell in a follow-on if the bank waives the lock-up. Circle came back 71 days after its IPO. JPMorgan released the lock-up early, and holders sold 8 million of the 10 million shares at $130, against a $31 IPO price. Allaire sold another 357,812 shares, about $46 million. Between the two deals he sold about 10% of his stake in 71 days.
- They can sell on a 10b5-1 plan. Field adopted one in August 2025, the month after the IPO. Allaire’s sold 56,200 shares in July 2026 and 56,200 again in August. These run automatically after a 90-day cooling-off.
- They can sell before the IPO. Stripe’s 2026 tender was open to employees and shareholders. A founder who sells in tenders arrives at the IPO with money already out.
- And they can use structure. Sinha’s prepaid variable forward on 500,000 Rubrik shares gets him cash now on stock he delivers later.
Rule 144 gets blamed for the slow pace, but the cap is loose. It limits an affiliate’s open-market sales to the greater of 1% of the class or average weekly volume every three months. For Figma that’s about 4.5 million shares a quarter, and Field’s biggest quarter was 3 million.
Three other things slow founders down:
- The stock punishes it. Figma fell about 5% in June 2026 after a filing showed Field sold $4.36 million of stock. Circle fell on the day it announced its follow-on.
- The price has to be there. Field sold 3 million shares in the three months before late January 2026, when insiders were averaging about $38. In May 2026, with the stock at $25, his trust sold 174,430.
- The lock-up comes first. Standard is 180 days. Musk’s is 366 with no early release.
Here’s the pace so far:
A founder who sells in the IPO and has a stock that works can get about 10% out in the first year. After that it’s a few percent a year. Allaire is 13 years past founding and still holds 15.6 million Class B shares directly.
M&A vs. IPO: The IPO Path Has to Be Worth 2-3x the Offer
A cash offer pays in 6 to 12 months. The IPO path pays about four years later, in pieces, at whatever the stock is on those days. To compare the two, adjust for three things.
1. Compounding. Cash in hand grows. Four years at a 10% index-like return is 1.46x. The IPO path has to beat that just to tie.
2. Risk. You’re holding one stock, locked up, with insiders selling into a thin float. Nobody should hold that for an index return. At a 20% hurdle the bar is 2.07x. At 25% it’s 2.44x.
3. Dilution. Your percentage shrinks the whole time you wait:
- Rubrik sold 23.5 million shares at $32 in its IPO at about a $5.6 billion valuation, roughly 13% of the company. SpaceX sold only 4% to 5%.
- Grants follow every year. Median net dilution from stock comp at public growth tech companies is 2.3% a year, and the bottom quartile is above 3.2%. Newly public companies run hotter. Rubrik’s share count went from 200.1 million to 208.2 million in the six months to May 31, 2026, or 4% in half a year.
- Shares sold in the IPO go at the IPO price. Figma priced at $33 and closed day one at $115.50. Selling holders sold 24.7 million shares at $33.
Model 13% for the IPO and 3% a year for four years. A holder keeps 77% of their stake. The company has to be worth 1.30x more for that stake to be worth the same.
Multiply them together:
The bar is roughly 2x to tie an index fund and 2.7x to 3.2x to get paid for the risk. Measure it at the market cap on the days you can sell, after the unlocks have hit the float.
Founders run a slightly different version. They get new grants on the IPO path, so they dilute less than VCs do. But a founder selling 4% a year, like Bicket at Samsara, has an average liquidity date far past year four, so the compounding bar is higher.
M&A takes haircuts too, and they’re smaller: a median 10% escrow, or 2.8% with RWI, plus the risk the deal dies. A stock deal isn’t cash and isn’t always fully liquid. And founders in PE deals often re-vest some of their proceeds.
Here’s how the cases did against that bar:
- Figma missed. Adobe offered $20 billion in 2022. At the final unlock in August 2026, Figma was worth about $12.6 billion, or 0.63x against a bar of roughly 2x.
- Wiz cleared it with a bigger offer. It turned down about $23 billion from Google in 2024 and signed at $32 billion in March 2025, 1.39x in under a year.
- Snowflake cleared it by a mile. Sutter Hill’s stake was worth about $12.6 billion at the IPO on less than $200 million invested. No acquirer was paying that.
Are the Rules Changing? Stripe and Anthropic
Two companies don’t fit the pattern.
Anthropic is fast.
- It was founded in 2021. It confidentially filed for an IPO on June 1, 2026, and the Wall Street Journal has reported a November target. If it prices this year, it will have taken about 5.5 years from founding, against a median of 11.5.
- It also started liquidity before the IPO. Anthropic ran an employee tender from February to April 2026 at a $350 billion pre-money valuation, with about $6 billion of investor demand lined up.
Stripe is slow, but has done many emplopyee tenders.
- It was founded in 2010. Sixteen years in, it has no S-1 on file, and John Collison says an IPO would be “a solution in search of a problem.”
- But Stripe has run three tenders in twelve months: $91.5 billion in February 2025, $106.7 billion in September 2025 and $159 billion in February 2026. The last one was open to employees and early shareholders, with Thrive, Coatue and a16z buying and Stripe itself repurchasing shares.
A Stripe holder has had a liquidity window roughly every six months, with no lock-up and no public filing.
IPOs themselves now carry more secondary. In 2025, only 86.2% of shares in VC-backed IPOs came from the company, against a 91.5% long-run average. Existing holders sold the rest. Figma’s IPO had 24.7 million shares from holders and 12.5 million from the company.
What Secondary Changes in the Math
Secondary changes the math somewhat.
The shares you keep still have to clear the same bar. Selling 20% in a tender does nothing for the other 80%.
What changes is the return you need. A founder or fund with nothing off the table should want 20% to 25% a year to keep holding, which is the 2.7x to 3.2x bar. Once you’ve taken meaningful money out, a 10% to 15% hurdle is reasonable for the rest. That works out to 1.9x to 2.3x.
So secondary takes the bar from about 3x to about 2x.
It also buys time. A fund at year 12 with DPI from tenders can wait for the IPO. A fund at year 12 with none takes the M&A offer.
The catches:
- You sell early, at a discount to what comes next. Anthropic’s tender priced at $350 billion pre-money. Its Series H closed in May 2026 at $965 billion post-money. Sellers got roughly 40 cents on the dollar against the round one month later. Most employees saw it coming, and the tender fell short of the $6 billion because they wouldn’t sell.
- The company controls it. It sets the price, the cap, the buyers and who’s eligible. Stripe’s 2026 tender was open to early investors. Many are employees-only.
- Most of it happens at the biggest companies. It’s spreading: Intercom ran a $100 million tender at a $2 billion-plus valuation, and Decagon ran one at $4.5 billion. But a company with an average cap table and a strong M&A offer in hand usually doesn’t have a $159 billion tender as its alternative.
A fast IPO doesn’t shorten the lock-up, either. Banks have pitched Anthropic and OpenAI on phased lock-ups instead of the standard 90 to 180 days. SpaceX is the template: 15 release dates, with the largest holders locked into mid-2027. A November 2026 IPO on that structure means the biggest Anthropic holders are fully unlocked in late 2027 at the earliest. That would still be about seven years from founding, the fastest clock in this post apart from Wiz.
A Strong M&A Offer Vs IPO: Do The Math Carefully, and Thoughtfully. Again, An IPO Probably Has to Be Worth 2-3x a Strong M&A Offer.
Start the clock at the first check. The best case here was 9.5 years to 94% out. Rubrik is at 12.4 years and 82%. Figma is at 13 years with 50 million shares still in the fund. Wiz was 6 years to 100%.
Then run the multiple. If you’ve already taken meaningful money out in a tender, the bar is about 2x. If you haven’t, it’s about 3x. That’s public market cap, four years out, after dilution.
If you can’t see the company getting there, take the offer seriously.
An IPO is still Plan A, and Snowflake shows why. But most founders compare the offer to the expected IPO price. The IPO path has to beat 2 to 3 times the offer.
