Google X co-founder Sebastian Thrun raises $130M for hardware AI startup Dulo

A new heavy hitter in AI

Sebastian Thrun, a prominent figure in the autonomous vehicle space, has launched a new venture called Dulo. Thrun is well-known for co-founding Google X and leading the programs that eventually became Waymo. He also played a major role in establishing Google Brain, which is now known as Google DeepMind.

Dulo is joining a growing trend of engineering-focused AI startups. These companies are moving beyond simple chatbots to solve complex industrial problems. The startup recently confirmed it raised $130 million in a funding round to build these specialized tools.

The investment group includes some of the most recognizable names in technology. Former Google CEO Eric Schmidt and search giant investor Ram Shriram joined the round alongside Salesforce CEO Marc Benioff and venture capitalist John Doerr. While the company did not disclose a specific valuation, the backing suggests high confidence in Thrun's vision.

The engineering of hardware

Dulo's primary goal is to develop foundation models specifically optimized for hardware design tasks. A foundation model is a large-scale AI system trained on vast amounts of data that can be adapted to a wide range of downstream applications. In this case, the focus is on the physical creation of products.

One way the company might achieve this is by optimizing engineering simulations. These are the digital tests hardware teams use to check for flaws before they actually manufacture a product. By using AI to speed up these simulations, companies can identify errors much faster than traditional methods allow.

The startup's website mentions a goal of facilitating "manufacturing at lightspeed." This suggests that the AI might automate parts of the design cycle that currently require manual labor or heavy computing power. The company also listed robotics as a key focus area, hinting at applications in industrial automation.

How the technology works

To understand how Dulo might work, it helps to look at similar technologies in the field. Some startups are already automating the "meshing" phase of simulation projects. Meshing is the process where engineers turn a blueprint into a collection of small, three-dimensional shapes to reduce the hardware needed for a simulation.

Other engineering AI models focus on speeding up partially differentiated equations. These are the math formulas that simulators use to predict how hardware will behave in the real world. By solving these equations faster, engineers can see how a robot or vehicle will react to stress or movement in seconds rather than hours.

Dulo appears to be positioning itself to handle these types of complex calculations. By applying foundation models to these specific engineering hurdles, the startup hopes to streamline the path from a digital concept to a physical machine. This could be particularly useful for creating autonomous vehicles or complex industrial robots.

The competitive landscape

Dulo is entering a space that already has some very large players. Its most prominent rival is Prometheus Inc., which was co-led by Amazon founder Jeff Bezos. Prometheus raised a massive $12 billion earlier this year to use AI for developing complex hardware systems like robots.

The competition highlights how much capital is flowing into the intersection of AI and physical engineering. While Dulo is smaller in terms of funding, it benefits from Thrun's deep history with Google's most advanced research labs. The presence of multiple startup funds in Dulo's round suggests a crowded but high-interest market.

Other players like Vinci4D Inc. are also carving out niches by automating specific parts of the simulation pipeline. As more companies build tools to optimize engineering designs, the goal is to make the transition from idea to prototype as seamless as possible. Dulo aims to be a primary driver of this transition for hardware teams.

What comes next for Dulo

The next step for Dulo will be defining exactly which hardware design tasks its models will automate first. While the company has mentioned robotics and manufacturing, the specific software tools they release will determine their market position. They will need to prove that their AI can handle the precision required for industrial-grade engineering.

Success for Dulo would mean a significant shift in how products are designed. Instead of engineers spending months on simulations, they could use Dulo's models to get near-instant feedback on their designs. This could lead to faster production cycles for everything from consumer electronics to heavy machinery.

As the company begins its work, observers will watch to see how it differentiates itself from the Bezos-backed Prometheus. With $130 million in the bank, Dulo has the resources to build a sophisticated platform. The ultimate goal is to make the physical world as programmable and predictable as the digital one.