Justin Ernest's firm uses single deal SPVs to place family offices into the rounds they cannot get into directly. It works, and it concentrates risk in one operator's hands.
Justin Ernest spent the last twelve months writing checks into Anthropic, Anduril, PsiQuantum, SpaceX, and a handful of other contested late-stage companies. He did it without raising a traditional venture fund, without a long-term capital commitment, and without ever telling limited partners what his next deal would be. According to TechCrunch's Marina Temkin, his firm Sabertooth VC has moved roughly $400 million into ten companies in that window. What makes the number worth reading is the structure behind it.
Ernest's vehicle of choice is a special purpose vehicle, or SPV: a single-deal fund that wraps one approved financing round and resells it to a small, fixed group of outside investors. Each Sabertooth deal is, in effect, its own fund. The structure is not new. The scale and the deal selection are. As Temkin reports, Sabertooth has used the format to route family offices and smaller institutional allocators into companies that have largely closed their official cap tables to anyone outside the founders' existing networks. The roster reads like a list of the rounds institutional investors most often hear about and most often cannot get into: Anthropic, Anduril, Databricks, PsiQuantum, and SpaceX.
The mechanics matter. A traditional venture fund commits a manager to a multi-year investment plan and locks in limited partners for the life of the vehicle. Raising one takes twelve to eighteen months for a first-time manager, per Ernest's own framing in the TechCrunch piece, and the time cost is real for anyone trying to chase rounds that close in days. SPVs invert the order. Ernest identifies a deal, lines up a group of roughly thirty family offices and smaller institutions, and deploys capital into that single round. No multi-year lockup. No commitment to write the next check. Each transaction stands alone.
That ordering changes who gets access and on what terms. Family offices with five to fifty million dollars to place into late-stage AI and deep-tech deals have, until recently, had three options: pay a premium on the secondary market, build direct relationships with founders, or accept whatever trickle of capacity a multi-stage fund is willing to allocate. None of those scale well. Sabertooth offers a fourth path: a packaged single-deal product, sold to allocators who want the deal more than they want a fund relationship. The package is not free. SPVs typically layer a management fee and carry on top of whatever the underlying company charges, and the smaller the deal, the more that overhead bites. Temkin's reporting flags this honestly, calling the wider market of single-deal SPVs and small allocations a "shady world" where the paperwork is thinner and the disclosures looser than a traditional fund's.
Ernest's pitch to that market rests on access, not on returns. He spent more than five years at Playground Global, the deep-tech firm founded by Android creator Andy Rubin, working on investments in early-stage hardware and frontier software. That network is the asset. Founders who will not take a call from a fund manager they have never met will sometimes take one from an operator with a known track record and a fast process. In the TechCrunch piece, Ernest positions Sabertooth as a way to let the company side keep moving without spending weeks educating a new limited partner base, and to let the allocator side write a check into a deal that would otherwise be closed to them. One family-office chief investment officer quoted in the piece describes the dynamic as something close to a routing service: Sabertooth exists because the company does not want to spend its fundraising cycle on fifty small checks, and the family office does not have a way in otherwise.
The structural picture is bigger than Sabertooth. Single-deal SPVs have become a quietly significant layer of late-stage venture plumbing, especially as AI-era rounds have grown larger and closed faster. Allocators with real capital but no network of their own now sit downstream of a small set of operators who can place them. That concentration is a real risk. The same person who picks the deal also vets the allocator base, sets the fee, and often sets the carry. TechCrunch's reporting makes clear that this is the part of the model the most thoughtful allocators are watching: the appeal of access is obvious, but the absence of a fund means there is no vehicle-level governance, no LP advisory board, and no diversified exposure to soften any one bad outcome. If Ernest picks wrong on a single deal, his investors have no other deals to balance it against. Key-person risk is not a footnote here. It is the business.
What to watch next is whether the SPV intermediation layer hardens into a permanent feature of the late-stage market, or fades once founders reopen direct allocations to outside capital. The answer depends less on Sabertooth's deal flow than on whether founders keep treating the formal cap table as a closed network, and whether family offices keep preferring a packaged single-deal product to the slow work of building direct relationships. If both hold, expect more operators to copy the model. The $400 million number will not be the headline for long.