Credit: @lightspeedvp / x.com
The firm calls the deal Project Mercury, Bloomberg reported, citing people familiar with it.
It gathers assets from two Lightspeed funds, Select V and Opportunity II, plus a separately managed account. OpenAI is the anchor.
Coller Capital is the lead buyer. UBS is advising. Lightspeed did not reply to Bloomberg’s messages and Coller declined to comment.
What a continuation fund actually does
The mechanics matter more than the number, so here they are without jargon. A venture fund has a life, usually about ten years, and at the end it must sell what it holds and hand back the cash.
A continuation fund breaks that clock. The manager moves chosen holdings into a new vehicle, a buyer funds it, the original investors get their money, and the firm keeps the position.
Investors in the old fund can take the cash or roll into the new one. That choice is the product.
The problem it solves is real. Companies now stay private far longer than a decade, so the fund expires before the investment does.
Lightspeed sits on both sides of it
Here is the part that makes these deals contentious in private equity, and it arrives in venture unchanged. Lightspeed is selling the assets on behalf of one set of investors, and it will manage the vehicle that buys them.
Somebody has to price OpenAI for that transaction. There is no share price to consult, because the company is private.
The reference point is the last primary round, which valued OpenAI at $852bn post-money in March after a $122bn raise, per Tech Funding News.
Buyers set that number. Sellers did not. A continuation fund has to turn it into a transfer price, and the manager benefits from getting that price right for its own side.
Coller is the discipline here. It is an arm’s-length buyer with its own investors, and it has no reason to overpay.
The Anthropic line is not a rollover
One detail separates this from a straightforward extension. Bloomberg describes the Anthropic exposure as a new commitment rather than an existing holding.
So investors asked to roll are not only keeping what they had. They are funding a fresh position.
The entry point is steep. Anthropic’s last round valued it at $965bn, three months after a round at $380bn.
Lightspeed knows the company well, having backed it across three consecutive rounds including a $3.5bn Series E in 2025. Familiarity is not the same as a discount.
What else is in the box
The other assets spread the risk away from the frontier labs. Verkada makes AI physical security systems, and a CapitalG investment in December lifted it to $5.8bn.
Rippling, the workforce platform, raised at $16.8bn in May 2025. Glean, which builds enterprise AI search, raised at $7.2bn a month later.
Reflection AI has moved fastest. It raised at $8bn in October 2025. In March it reportedly sought fresh money at $25bn.
The desk has covered where its compute comes from. Reflection signed a $6.3bn compute deal with SpaceX for Nvidia chips running to 2029.
Why the structure is spreading
Follow the arithmetic rather than the fashion. AI valuations compound every few months. Defending a stake against dilution in later rounds costs hundreds of millions.
An ageing fund does not have that money. Its investors expect distributions, not more capital calls.
A continuation fund answers both problems at once. It raises new money against old assets and pays the original backers out of it.
The volumes show the trend. Venture-led secondary deals reached $35bn in 2025, roughly double the 2023 figure, according to PJT Partners.
That sits inside a secondaries market worth $121bn of transactions in the first half of this year alone.
The concentration underneath it
Read this against where venture money is already going. Megadeals took 87.5% of the dollars deployed in US venture in the second quarter.
A structure designed to hold winners longer will concentrate a market further, not loosen it. Lightspeed is not causing that, and it is responding to it rationally.
Other firms are restructuring at the same time. Accel raised $3.5bn across four funds in a single process this week.
And a single private stake can now define a whole firm. Thrive Capital’s first investor letter showed how completely OpenAI has come to define it.
Europe owns the buyer
The European angle is on the other side of the table. Coller Capital is a London secondaries specialist, and EQT, the Swedish investment group, recently acquired it.
So a Swedish-owned, London-based firm is the lead buyer of Silicon Valley’s most valuable private AI positions.
That is a quiet reversal of the usual direction. European capital normally shows up in these stories as the thing being bought.
It is also a reminder of what secondaries firms do. They are the market of last resort for illiquid assets, and they are now the price-setters for private AI.
What would settle it
Three things, and the first is the price. The transfer valuation for the OpenAI stake would show whether Coller paid the March mark, a premium or a discount.
That number is the closest thing to an independent read on OpenAI available today. It is also the number nobody will publish.
The second is how many investors take the cash rather than roll. A high rollover says the LPs share the conviction. A low one says they wanted out and the structure gave them a door.
The third is the exit that ends the argument. Investors expect Anthropic to list in October at $2trn or more, and OpenAI would follow at some point.
Until one of those happens, every valuation in this deal is a private agreement between people who benefit from it being high. That is not an accusation. It is the definition of a private market, and a continuation fund is what you build when you want to stay in one.
Get the TNW newsletter
Get the most important tech news in your inbox each week.
