Coller Leads $600M GP-Led Deal Anchored by OpenAI and Anthropic Stakes
Lightspeed Venture Partners is the GP behind the vehicle, which pairs the continuation stakes with a fresh primary check, according to Bloomberg.
Lightspeed Venture Partners is putting together a roughly $600 million secondaries process internally dubbed “Project Mercury,” according to Bloomberg. The goal: move positions from two of its funds (Select V and Opportunity II) and a separately managed account into a continuation fund, with Coller Capital as lead buyer and UBS advising.
The package bundles five assets: OpenAI, Verkada, Rippling, Reflection AI, and Glean, plus one detail that stands out: fresh capital to (re-)enter Anthropic, right after the round that valued the company at $965 billion, Bloomberg reported.
We don’t normally cover individual continuation vehicles here at Secondary Scoop. This one is different, and not just because of the headline. Lightspeed has run this exact strategy before, and the differences between the two deals tell you something real about where VC secondaries are headed.
This isn’t Lightspeed’s first continuation vehicle
Back in January 2024, the Financial Times first reported that Lightspeed was exploring a continuation fund of roughly $1 billion, covering stakes in about 10 portfolio companies, around 70% of them enterprise tech. The firm had a term sheet from a lead investor and was targeting a close by July 2024. Notably, that structure came with no additional fees or carry, a term clearly designed to make it easier for existing LPs to swallow, since fee-stacking is one of the standard LP objections to continuation vehicles.
Lexington Partners ended up leading that 2024 deal, part of what Buyouts Insider called Lexington’s “GP-led deal tear” at the time. Per Lazard’s H1 secondary-market report cited in that coverage, growth/venture deals made up only around 14% of GP-led market volume that half, up markedly year over year, but still a minority of the broader GP-led market.
The vehicle eventually closed, PitchBook lists it as the “Lightspeed Multi-Asset Continuation Vehicle,” a 2024-vintage fund that closed at $1.5 billion.
What’s different about Project Mercury
1. It’s a “hold” dressed up as a “sell.” A continuation fund isn’t a real exit, it’s a way of saying “I don’t want to give up this position, but I do want to give my current LPs liquidity.” Lightspeed is using the vehicle to return cash to Select V and Opportunity II investors without having to let go of OpenAI too early. It’s the classic move when a GP believes the best of the markup is still ahead, and it’s the same logic that drove the 2024 deal.
2. The buyer changed: Lexington to Coller. Lexington led in 2024; Coller (freshly owned by EQT) leads in 2026. Worth watching whether this becomes a rotating cast of the same handful of large secondaries funds (Coller, Lexington, Ardian, Pantheon) each taking a turn anchoring VC-led multi-asset CVs, or whether Lightspeed is deliberately spreading its buyer relationships deal to deal.
3. This time there’s a new-money component. The 2024 vehicle was framed purely as a continuation/liquidity mechanism for existing stakes. Project Mercury explicitly bundles in fresh primary capital for Anthropic alongside the continuation piece, a hybrid structure that’s more aggressive than what Lightspeed did the first time around.
4. Concentration went way up. 2024’s ~10 assets were diversified, mostly enterprise tech. 2026’s vehicle is anchored by the two largest foundation-model companies on earth. For the buyer, that’s a much higher single-name concentration bet than a diversified basket, and a bet on two directly competing labs at once.
5. It fits, and amplifies, the macro trend. The underlying driver hasn’t changed since 2024: the IPO window for large private tech companies has stayed mostly shut, marquee names (OpenAI, Anthropic, Stripe) keep staying private well past a decade, and traditional 10-12 year VC fund lifecycles don’t match that timeline. Lightspeed’s own chief business officer put it plainly in 2024: VCs “need to take a page out of the private equity playbook” on liquidity. echoing how continuation vehicles went from roughly 41% of PE sponsor-led secondary volume in 2019 to about 80% by mid-2023. The numbers back it up: venture-led GP-led secondaries volume hit $35 billion in 2025, roughly double the 2023 figure, according to a PJT Partners report cited by Bloomberg, against a global secondaries market of $121 billion in H1 2026 transactions alone.
Lightspeed isn’t the only one doing this
Insight Partners closed its second continuation fund at $1.3 billion, spanning six funds, one of the largest venture-specific CVs to date.
NEA has been quietly developing its own continuation vehicle.
RockPort Capital moved two portfolio companies into a CV.
Shasta Ventures tried to move nearly all the holdings from its last fund into a CV, LPs rejected the pricing, which had been offered at roughly 65% of the Q3 2023 valuation. A useful reminder that these deals don’t automatically clear.
Reported median performance for continuation funds sits around 1.4x MOIC, roughly in line with dedicated secondaries funds and ahead of typical buyout returns, one reason LP appetite for CV exposure keeps growing even when individual deals, like Shasta’s, don’t.
The open question
How do you price an OpenAI or Anthropic position inside a continuation fund when neither company is transparent about its cap table or deal terms? That’s the real risk (and the real opportunity) for Coller and any buyer stepping into a structure like this. It’s also worth watching whether other large multi-stage VCs with heavy OpenAI/Anthropic exposure follow Lightspeed’s lead: doing this twice in under three years, each time bigger and more concentrated, looks less like a one-off and more like a template.
About Lightspeed Venture Partners
Lightspeed Venture Partners is a Menlo Park-based VC firm founded in 2000 that now manages $40+ billion in AUM (as of its December 2025 fund close, up from roughly $35B in 2023), spread across early-stage, growth, and opportunity vehicles, its latest raise brought in over $9 billion across six new funds, including Fund XV ($2.18B combined), Select VI ($1.8B), and Opportunity Fund III ($3.3B). It’s one of the more AI-forward large VCs today, having backed 165 AI-native companies with $5.5B+ deployed since 2012, and its current marquee holdings include Anthropic, OpenAI (via its recent secondary/continuation activity), xAI, Databricks, Mistral AI, Glean, Rubrik (where it’s the largest shareholder), Navan, Netskope, Abridge, and Skild AI. Historically, its standout exits include Snap (early backer, IPO 2017), Affirm (IPO 2021 at ~$30B), MuleSoft (acquired by Salesforce for $6.5B), AppDynamics (acquired by Cisco for $3.7B), Nest (acquired by Google for $3.2B), and Nutanix, a track record that’s helped it become one of the go-to names for founders building category-defining AI companies