Lightspeed is divesting its stake in OpenAI to a fund that it will also manage.

      The firm refers to the deal as Project Mercury, according to Bloomberg, which cited sources familiar with the matter. It involves assets from two Lightspeed funds, Select V and Opportunity II, in addition to a separately managed account, with OpenAI as the primary asset. Coller Capital is the main buyer, while UBS is providing advisory services. Lightspeed did not respond to Bloomberg's inquiries, and Coller chose not to comment.

      Understanding what a continuation fund does

      The mechanics are more important than the terminology, so here’s the straightforward explanation. A venture fund typically has a lifespan of around ten years, at which point it must liquidate its holdings and return the funds to investors. A continuation fund changes this timeline. The manager transfers selected assets into a new fund, a buyer finances it, the original investors receive their money, and the firm retains its position. Investors from the previous fund can either take the cash or invest in the new one. This choice is the product being offered.

      The problem it addresses is significant. Companies are remaining private for longer than a decade, meaning the fund often expires before the investment matures.

      Lightspeed operates from both sides

      This is where the deals become contentious in private equity, and it remains unchanged in venture capital. Lightspeed is selling the assets on behalf of one group of investors while simultaneously managing the fund that acquires them. Someone must determine OpenAI's value for this transaction; there is no public stock price, given the company's private status. The reference point is the most recent primary funding round, which valued OpenAI at $852 billion post-money in March following a $122 billion raise, according to Tech Funding News. The buyers set that figure, not the sellers. The continuation fund must convert it into a transaction price, and the manager benefits by accurately pricing it for their side.

      Coller acts with discipline in this scenario. As an arm’s-length buyer with its own investors, it has no incentive to overpay.

      The Anthropic aspect is not a rollover

      One detail that distinguishes this from a simple extension is that Bloomberg characterizes the exposure to Anthropic as a new investment rather than a pre-existing one. Thus, investors being asked to roll over are not only retaining what they previously held but also financing a new acquisition. The entry point is substantial; Anthropic's last round valued it at $965 billion, just three months after it was valued at $380 billion. Lightspeed is well-acquainted with the company, having supported it through three consecutive funding rounds, including a $3.5 billion Series E in 2025. However, familiarity does not guarantee a discount.

      Additional assets involved

      Other holdings help diversify risk beyond the cutting-edge laboratories. Verkada produces AI-driven physical security systems, and a December investment from CapitalG valued it at $5.8 billion. Rippling, a workforce platform, raised $16.8 billion in May 2025. Glean, which develops enterprise AI search technologies, secured $7.2 billion a month later. Reflection AI has progressed rapidly, raising at $8 billion in October 2025 and reportedly seeking additional funding at $25 billion in March.

      The desk has clarified its computational sources. Reflection has signed a $6.3 billion agreement with SpaceX for Nvidia chips effective until 2029.

      Reasons for the structure’s expansion

      It is wise to follow the math rather than the trend. AI valuations increase every few months, making the cost of maintaining a stake against dilution in later rounds extraordinarily high—amounting to hundreds of millions. An aging fund does not possess such capital, as its investors anticipate distributions rather than additional capital calls. A continuation fund addresses both issues simultaneously by raising new funds against existing assets while reimbursing the initial investors.

      The figures illustrate the trend: venture-led secondary transactions hit $35 billion in 2025, nearly double the 2023 total, as reported by PJT Partners. This fits within a secondaries market that was valued at $121 billion for transactions in the first half of this year alone.

      Underlying concentration

      This trend should be viewed alongside where venture capital is currently directed. Large deals accounted for 87.5% of the capital deployed in U.S. venture in the second quarter. A structure designed to hold successful investments longer will likely further concentrate the market rather than loosen it. Lightspeed is merely responding to this situation rationally rather than initiating it. Other firms are concurrently restructuring; for instance, Accel raised $3.5 billion across four funds in a single process this week. Moreover, a single private investment can now define an entire firm, as evidenced by Thrive Capital’s first investor letter showcasing how significantly OpenAI has come to represent it.

      European buyers

      The European perspective features on the opposing side of the table. Coller Capital, a secondary specialist based in London, has recently been acquired by the Swedish investment group EQT. Thus, a Swedish-owned, London-based firm emerges as the leading