Summary

Sebastian Thrun, famed for pioneering self‑driving cars at Google and later launching ventures like Google Brain, Udacity and the short‑lived Kittyhawk, has quietly opened a new robotics venture called Dulo. The company was introduced at the end of a keynote at the Actuate robotics conference, where Thrun said it operates under “stealth” and is “very small.” Dulo’s public footprint is limited to a stripped‑down website hosted by Stanford that promises “foundation models for hardware design” with the lofty goal of enabling manufacturing at lightspeed. The roster of talent behind the startup reads like a who’s‑who of AI and robotics: alumni from Waymo, Google Brain, and Stanford’s Artificial Intelligence Laboratory, the same circles that produced Thrun’s earlier breakthroughs.

The push behind Dulo appears to be part of a broader industry movement to apply large‑scale AI models – originally popularized for language and vision – to the physical design of hardware. By treating hardware blueprints as data that can be generated or optimized by foundation models, the company hopes to slash design cycles that traditionally take months into seconds. This could dramatically lower the cost and time required to bring new physical products to market, a prospect that excites investors betting on the next wave of AI‑driven hardware innovation. Yet, the venture remains embryonic; no products, revenue streams, or concrete applications have been disclosed, and the team size is reportedly tiny.

Why Dulo is emerging now ties together several threads. First, the explosive growth of foundation models in software has opened eyes to their potential beyond text and images, especially in simulation‑heavy fields like robotics and manufacturing. Second, Thrun’s own history of high‑profile projects – both successes (Waymo) and failures (Kittyhawk) – suggests he is now pursuing a more focused, perhaps lower‑risk approach to AI‑hardware convergence. Third, venture capital continues to pour money into AI‑hardware startups eager to capture the “design acceleration” narrative, making a stealthy entry strategically attractive. Finally, the broader macro trend of supply‑chain re‑shoring and rapid prototyping creates a fertile environment for technologies that can compress product development timelines.