A San Francisco startup called Dulo has raised $130 million in seed funding to build artificial intelligence models aimed squarely at engineering and hardware design, betting that the same kind of AI reshaping software can be pointed at cars, planes and other physical products. The company did not disclose its valuation, and its founders are hardly strangers to Silicon Valley's biggest swings.
A Famous Founding Duo Behind the Round
Dulo was co-founded by Sebastian Thrun, the computer scientist known for founding Google X and helping establish Waymo, and Adrian Aoun, who previously founded healthcare startup Forward, according to citybiz. Thrun is also an adjunct professor at Stanford University and founded the online education company Udacity, which Accenture acquired in 2024, per the same outlet. Aoun's resume includes founding Sidewalk Labs, a Google-backed urban technology venture, and co-founding Torch, an AI healthcare startup that was later acquired by OpenAI.
The round drew a notably starry investor list. As reported by SiliconANGLE, Salesforce CEO Marc Benioff and venture capitalist John Doerr joined the round, alongside more than half a dozen startup funds. Among the backers named in citybiz's reporting are DST Global, which has offices in Silicon Valley, New York, London and Hong Kong; Construct Capital, founded by Dayna Grayson and Rachel Holt in 2020 and focused on early-stage manufacturing, supply chain, food and transportation investments; and Lowercarbon Capital, which invests in commercial solutions to reduce carbon emissions and address climate change.
From Writing Code to Designing Physical Products
Dulo's own website describes its goal as developing foundation models optimized for hardware design tasks, and Aoun has written in a LinkedIn post that the company aims to create what it calls an “Engineering Superintelligence” to automate technical stages of product design, according to citybiz. One report, from huggingnews.com, describes the software as converting natural-language inputs into valid engineering designs for cars and planes. According to citybiz, the company plans to invest the funding in high-performance GPU clusters, expand its research and engineering teams, and recruit machine learning researchers and systems software developers.
Engineering design involves constraints that differ fundamentally from the world of software development, and AI engineering models face the challenge of translating digital outputs into systems that have to operate under real physical constraints, per citybiz's reporting. Dulo plans to expand beyond pure software generation into the design and automation of physical systems, the outlet reports — a leap that goes well beyond writing code that runs on a server.
A Crowded Field Chasing the Same Problem
Dulo is far from alone in trying to fuse AI with physical-product engineering. Forbes reports that AI systems are increasingly being developed to accelerate complex simulations and automate engineering workflows across industries. In September, Mistral opened an industrial AI research hub in Munich focused on physics AI and engineering applications, Forbes reported. It followed Mistral's May 2026 acquisition of Emmi AI, a company specializing in AI models for fluid dynamics, structural mechanics and multiphysics simulation.
Other comparisons are even starker. SiliconANGLE notes that several other engineering-focused AI startups are also working to optimize simulations, including Vinci4D, which has built a platform automating the meshing phase of simulation projects, and Prometheus, which is using AI to speed up development of complex hardware systems such as robots. Prometheus raised $12 billion earlier in 2026 from Jeff Bezos and a group of financial institutions, according to the same report — a sum that dwarfs Dulo's seed round and underscores how much capital is chasing this category.
Data Readiness Remains the Industry's Bottleneck
Even as investment pours in, survey data suggests the industry still has real gaps to close before AI-driven engineering becomes routine. A March 2026 report from SimScale found that 80% of surveyed engineering leaders' organizations were experimenting with AI pilots, while only 9% had a mature, scaled AI program, even though all respondents rated AI as important to their design and simulation strategy. The same report found that 74% of organizations still cite data preparation and availability as the primary barrier to scaling engineering AI.
Where AI workflows have taken hold, the SimScale data points to real productivity gains: organizations using AI-enabled workflows evaluated an average of 56 design variants per program, compared with 17 for teams using conventional workflows. A related writeup from Manufacturing Dive on the same survey found that about 69% of respondents had adopted AI-enabled copilots and assistants, and that companies using AI-enabled workflows turned around requests for quote and technical bid responses roughly three times faster than those relying on conventional methods.
Adjacent research is also pushing in similar directions. Separately, TechCrunch reported that Physicl raised $10 million in seed funding led by Inovia Capital to help robot builders generate more realistic training simulations, releasing a product called Studio with 10,000 3D environments and 100,000 assets for synthetic data generation. Whether Dulo's approach to hardware design proves as capable as its founders' ambitions remains to be seen, since the available reporting describes the company's plans and goals rather than a shipped product or verified engineering results.
