Lola Vision Systems targets Nvidia's edge-AI grip with $1M seed round

Washington, D.C.-based Lola Vision Systems has raised just over $1 million to date and signed one paying customer as it tries to break Nvidia's hold on running artificial-intelligence models on-device.

The company, founded in 2024 by Tayo Adesanya, sells what he calls a "compiler toolchain" — software that converts an AI model into instructions a specific chip can run. It is also developing its own semiconductor. The pitch targets a bottleneck Adesanya pegged at "roughly 200 hours" of manual engineering just to begin testing an AI model on new hardware.

Lola Vision was selected for this year's TechCrunch Battlefield 200. TechCrunch Disrupt runs October 13-15 in San Francisco.

What does Lola Vision actually sell?

A client hands Lola Vision its code and an AI model — custom-built or open source — and the company's software compiles both into machine-readable instructions for the client's chip. "Starting Lola Vision Systems was a bet on where the world was headed and what I was seeing," Adesanya said. He spent nearly 12 years advising large manufacturers on chip selection before launching the company.

Lola Vision plans to license its software on third-party hardware in the near term rather than wait for its own silicon. "To get revenue sooner, we will now license our software on existing hardware," Adesanya said. He added that chips remain part of the long-term roadmap.

How big is the gap the company is trying to close?

Most buyers, Adesanya said, start with Nvidia's Jetson line of compact AI modules or with open-source AI models. Those often arrive broken or running poorly. "Teams spend days or weeks getting them to run at all, then even more weeks debugging until the models are usable," he claimed.

Power budgets are the next pain point. "Power consumption often blows edge computing budgets, or the board can't deliver enough compute for the medium to large models the product actually needs to run successfully," Adesanya said. The downstream effect, he added, shows up as "recognition models lagging behind targets or misreading objects."

"Speed is only part of it," Adesanya said. He framed the longer-term payoff for aerospace and "other mission-critical companies" as the ability to "run more accurate models on their own data, at a lower power."

What's the funding picture, and who is buying?

According to the company, twelve corporate customers have signed letters of intent to buy Lola Vision's chips once production units are available, and one customer has signed a purchase agreement. Total capital raised sits at slightly more than $1 million — a modest seed by edge-AI standards.

Lola Vision has also partnered with SCALE, a microelectronics workforce development program, to extend its reach into more semiconductor labs. That gives the startup a talent and tooling channel at a scale a one-year-old company could not build on its own.

Why does Adesanya think a compiler is a moat?

The founder's bet is that the field has obsessed over chips while leaving the surrounding software brittle. "For these customers," Adesanya said, "accuracy and reliability aren't nice to have. They determine whether a product passes regulatory review and whether it works reliably in the field." In markets where a missed object detection can ground a drone or fail an audit, a faster, more reliable model-deployment pipeline carries a price premium.

What is the next milestone?

Lola Vision heads into Battlefield as a one-year-old startup with paying revenue, hardware-software licensing as a near-term lever and a proprietary chip on the back burner. Adesanya made his near-term priorities explicit: "making meaningful connections and learning as much as I can about what's happening in and around our space." He was also candid about the cash chase: "And, to be direct, I'm looking forward to investors writing checks."

If the software licensing model pulls in early customers before the first chip tapes out, Lola Vision will pressure Nvidia's Jetson not on silicon but on the unglamorous layer of tooling that turns a model into a working product.