Autonomous vehicle pioneer Sebastian Thrun has unveiled his latest venture, a stealth robotics startup named Dulo that aims to revolutionize hardware design and manufacturing through artificial intelligence (AI).

Thrun dropped the name during the closing moments of a keynote address at Actuate, a robotics conference in San Francisco. While keeping specific details under wraps, he confirmed the company’s domain to an audience of industry founders and engineers.

“I am not speaking about the company yet,” Thrun told attendees. “It’s under stealth, it’s very small. But it’s in robotics.”

Despite the brief revelation, early details hosted on a Stanford University website indicate Dulo is developing “foundation models for hardware design” to enable “manufacturing at lightspeed.”

While foundation models have radically transformed software, text, and image generation, physical hardware development remains a notoriously slow, traditional process. Dulo appears positioned to apply advanced AI techniques directly to the blueprint and assembly phase, targeting the tools used to build machines rather than manufacturing consumer-facing robotics directly.

The startup’s early lineup signals significant technical ambition. According to the university site, Dulo’s team includes leaders from Waymo, Google Brain, and the Stanford Artificial Intelligence Laboratory (SAIL), which Thrun previously directed.

The venture arrives amid a historic boom in physical AI investment. Driven by severe labor shortages, falling component costs, and an aggressive push toward supply chain reshoring, capital has flooded the sector.

Physical AI startups raised a record $16.3 billion across 492 deals in the first quarter of 2026 alone. Massive funding rounds and strategic acquisitions, such as SoftBank’s $200 million investment in autonomous construction firm Gravis and ON Semiconductor Corp.’s $7 billion acquisition of Synaptics Inc., highlight the market’s current appetite for hardware-focused automation. Western founders are also feeling increased pressure to innovate as Chinese manufacturers dominate global production, currently building nearly every humanoid robot shipped worldwide.

Thrun’s track record places him in an advantageous position to capitalize on this wave. He first gained national prominence in 2005 when his Stanford team won a landmark 132-mile autonomous vehicle race across the desert. Google co-founder Larry Page subsequently recruited Thrun to launch the tech giant’s self-driving car initiative, which eventually spun out as Waymo, now the leading commercial robotaxi operator in the United States.

Beyond autonomous vehicles, Thrun co-founded Google Brain, the moonshot incubator Google X, and the online learning platform Udacity. He also headed Kittyhawk, an electric flying-car venture that ultimately shut down operations—a stark reminder of the massive risks inherent in ambitious hardware bets.

For now, Dulo remains largely behind closed doors. With no disclosed funding totals, product demos, or customer launch dates, the venture rests on a simple premise: bringing software-speed foundation models to the physical world. Whether Thrun can replicate his early success in autonomous driving within the manufacturing sector remains the multibillion-dollar question.