Building startups for one customer

Venture capital funds companies that might find a market. Vantora has raised $100m to fund companies that already have one customer and will never look for a second.

The firm, formerly UP.Labs, announced more than $100m from Silversmith Capital Partners, its first outside capital since it was founded in 2022 by John Kuolt, who previously built corporate ventures at BCG. It says it has launched 17 companies for partners including Porsche, Alaska Airlines, J.B. Hunt and Wabash, and that revenue grew 79% last year.

The model is a venture studio with the customer attached at the start. A corporation identifies a problem inside its own operations, Vantora builds a company to solve it, and that partner can absorb the result rather than watch it sell to the whole market. Kuolt describes the arrangement as a proprietary pipeline for acquisitions.

Read as economics, it is an argument about where the risk sits. A conventional startup carries product risk, market risk and distribution risk at once, and most fail on the last two. A company built for one industrial buyer eliminates distribution risk entirely and caps the upside in the same stroke.

That is a rational trade in the current market. Exits have been running through acquisition rather than the public markets, and if the likely end state is being bought by an incumbent, agreeing the buyer in advance removes years of uncertainty from the middle.

The declared focus is physical AI — models applied to machinery, logistics and plants rather than to text. It is the segment where the corporate partner's assets are the moat: nobody builds a useful predictive maintenance company without access to the machines, and the machines belong to the partner.

The obvious cost is on the other side of the table. A founder joining a venture built for one buyer is trading the improbable large outcome for a probable modest one, and the talent this model attracts will be different from the talent chasing a public listing.

There is also a question about what gets built. Solving the commissioning partner's problem exactly is the point, and it is also the definition of a product that cannot be sold to anybody else. If the partner's priorities change, the company has no second market to turn to.

For the industrial partners the appeal is straightforward and slightly damning. Large companies have spent a decade failing to produce software businesses internally, and paying an outside studio to do it — with outside money and outside operators — is an admission that the innovation-lab era did not work.

The interesting signal is the investor. Silversmith is a growth firm putting money into a venture builder rather than into ventures, which is a bet that the repeatable part of this business is the building, not any of the companies built.