A Deal-By-Deal Route Into AI Winners

Sabertooth Capital has invested nearly $500 million in 10 late-stage companies over the last 12 months, using single-deal vehicles instead of a traditional venture fund.

Justin Ernest, a former Playground Global investor, built the firm for family offices and smaller institutional investors that want exposure to fast-growing AI and deep-tech companies but often cannot enter those rounds directly.

The firm secures company-approved allocations and then offers each deal to about 30 smaller institutional investors through a special purpose vehicle.

The names cited for its portfolio include Anthropic, Anduril, Base Power, Databricks, PsiQuantum and SpaceX.

The structure makes access itself the product: investors are not buying a broad blind-pool fund, but a defined allocation in a specific private company.

Why The Structure Matters

The timing advantage is important.

A new venture fund can take 12 to 18 months to form, while an SPV can be organized around one company and one financing event.

For late-stage AI companies, where allocations are scarce and large investors dominate many rounds, that flexibility can make smaller institutions more relevant without forcing the company to run a wide fundraising process.

The numbers show real scale.

Sabertooth has written checks ranging from $10 million to $275 million, and Ernest says the firm participates in official, company-approved rounds.

That difference changes the story because the model depends on primary access and company permission, not merely trading private shares after the fact.

Access Does Not Remove Risk

The SPV model also concentrates exposure.

Each vehicle is tied to one company, so an investor gets less diversification than a conventional venture fund would provide.

That puts more pressure on valuation discipline, round terms, company approval and the timing of any eventual exit.

For AI financing, the signal is that capital is being reorganized around access bottlenecks.

Large model developers, defense-tech startups, quantum firms and data infrastructure companies can attract demand from more investors than they can easily admit to a round.

Sabertooth's next test is whether nearly $500 million across 10 companies becomes a repeatable allocation engine, or remains a strong run around a small set of highly sought-after names.