Sovereign AI You Can Own: Vantora's $100M Round Bets Corporations Want AI Ventures All to Themselves

The most interesting check in enterprise AI this week was not written for a model, a chip, or a data center. On September 16, Vantora — the corporate venture builder formerly known as UP.Labs — announced more than $100 million in growth investment from Silversmith Capital Partners, a Boston-based growth equity firm with over $5 billion under management. It is the first outside capital the company has ever taken. Four years after launching with Porsche as its inaugural partner, Vantora is doubling down on a provocative thesis: the most valuable AI companies of the next decade will not be sold to everyone. They will be owned, outright, by a handful of industrial giants.

What Vantora actually builds

Vantora is not an incubator, an accelerator, or a venture fund in the conventional sense. Its teams of seasoned founders and senior AI engineers embed inside a partner enterprise — Porsche, Alaska Airlines, J.B. Hunt, Wabash, and TDG, the parent of Ashley Furniture, among others — and work with the partner’s own operators and data to identify the problems worth the most money to the core business. The company says those problems typically represent $50 million to $100 million of annual EBITDA contribution. The output is not a consulting deck: it can be a new venture, a proprietary capability, or another strategic asset, built as an AI-native operating company from day one.

Founder and CEO John Kuolt, who spent years building corporate ventures at BCG X before starting the firm in 2022, frames the model as an answer to the innovator’s dilemma solved from inside rather than outside. “The world’s largest enterprises are sitting on enormous value disguised as problems,” Kuolt said in the announcement. “There is a conundrum: there has never been a way to align the world’s best tech entrepreneurs with the massive problem sets of the world’s most important companies. Equity is what aligns them.”

That equity structure is the heart of the pitch. The corporate partner invests in each venture from formation, serves as its first customer, and holds equity from the start. As the venture scales, the partner shares in the upside — and, critically, holds the option to spin the venture into the core business entirely once results reach the P&L. Vantora calls this “Sovereign AI”: if the intelligence layer of your business is going to be built on your operations and your data, you must own it.

Why the rebrand signals a strategic shift

The rename from UP.Labs to Vantora tracks a real change in strategy, not just marketing. Under its old identity, the firm built startups designed to solve problems for corporate customers and for the outside world alike. Vantora now builds startups solely for its corporate customers — what Kuolt calls a “proprietary M&A pipeline.” The company’s new focus areas include industrial manufacturing and oil and gas partners it declined to name.

The reason, Kuolt told TechCrunch, is that the old model forced the firm to kill its best ideas. “We were missing on the biggest value problems, which had the biggest upside because of that,” he said. “Imagine you’re a Fortune 100 industrial company and you need to retrofit all of your hardware and machines for autonomy. You need to own that, it needs to be sovereign, and you can’t rely on a third party to go do that for you. You need to own that intelligence layer. They’re never going to let us go sell that to their competitors.”

One concrete example: Vantora came up with an idea to use AI to advance the business of logistics partner J.B. Hunt. Under the old model, the plan would have meant taking the product to the broader market. “They said there is no way you can take this out to the world, and so we passed on it,” Kuolt said. The proprietary model now lets Vantora pursue exactly that category of idea.

The shift has also pulled the firm toward physical AI — systems that touch machines, vehicles, and physical operations rather than just software workflows. Silversmith principal Danielle Waldman said the firm “was doing Physical AI before there was even a name for it.” Vantora’s ventures range from automating product configuration and quoting for made-to-order manufacturing to rebuilding maintenance planning for airlines.

The numbers behind the bet

Vantora’s operating record is unusually concrete for the venture-building category. The company has launched 17 ventures to date and is targeting 20 by the end of 2026, according to Kuolt and Silversmith partner Sid Shah, with what the firm describes as a strong pipeline beyond that. Revenue has grown 79% year over year, and the company says it has been profitable throughout — a claim almost no AI-native company of its vintage can make. The Wall Street Journal reports that Silversmith’s investment is a majority stake, making it the first institutional money into the startup.

The capital will fund expansion of corporate partnerships, continued development of Vantora’s data ontology product, and hiring across AI and commercial roles. Todd MacLean, Danielle Waldman, and Annie Cory of Silversmith will join Vantora’s board of directors.

That ontology product, called COSMOS, is the technical foundation of the whole model. It unifies an enterprise’s operating data and makes it usable by AI-driven processes and autonomous agents — the unglamorous prerequisite for any serious physical AI deployment, and a moat that compounds with every partner added.

Why this matters beyond one funding round

The timing is not accidental. Vantora’s pitch lands in the middle of an enterprise AI market suffering from a stubborn value realization gap: 94% of organizations still do not derive a significant share of their earnings from AI, according to McKinsey data cited by the company — and that percentage has not moved in a year. Generic tools layered onto legacy systems rarely reach the operations where the value actually sits. Boards have spent three years buying AI; many are now asking why so little of it shows up in EBITDA.

Vantora’s answer inverts the standard enterprise AI go-to-market. Instead of selling software to many customers, it builds one company per problem, with the customer as anchor investor and eventual owner. The model addresses the two loudest complaints in enterprise AI simultaneously: vendors who do not understand the domain, and pilots that never convert. If a venture only exists to serve one operator’s core business, neither failure mode applies.

It also stakes out a position in the emerging “sovereign AI” conversation that governments and chipmakers have dominated until now. Nvidia has spent two years selling the term to nation-states wanting national compute and national models. Vantora is applying the same logic one level down: to individual corporations whose competitive survival may depend on owning the intelligence layer of their own operations — and who will not tolerate that layer being resold to the competitor across the street.

There are real risks. Venture building has a long graveyard, from the corporate incubator busts of the early 2000s to the more recent struggles of standalone venture studios. Concentration risk is structural: each venture’s fate is tied to a single anchor customer’s procurement cycles, strategy shifts, and balance sheet. And the model’s economics depend on corporate partners being willing to acquire their ventures at prices that reward fourteen years of building — a bet that has historically been easier to promise than to collect. Vantora’s profitability and 79% growth suggest the model is working so far; how it performs through a corporate capex downturn is the open question.

What to watch

• Whether Vantora hits its target of 20 launched ventures by year-end, and whether the unnamed oil and gas partners materialize into public case studies.
• Whether any partner exercises the fold-in option — the first full acquisition of a Vantora venture by its anchor customer would be the model’s proof point.
• Whether the “Sovereign AI” framing spreads: if more corporations start demanding ownership rather than subscriptions, the enterprise AI market’s entire revenue model gets renegotiated from the top down.

The frontier labs are racing to raise at trillion-dollar valuations. Quietly, in parallel, a Boston growth firm just put $100 million behind the idea that the next wave of valuable AI companies will never be sold to the public market at all — because their owners built them that way from day one.