🔥 Sebastian Thrun's New Startup Raised $130 Million Before Building Anything

The man who taught a car to drive itself just talked some of the richest people in tech into funding a company with no product, no customers, and no disclosed valuation.

Let’s get the number on the table first: $130 million. That’s the seed round Dulo, a San Francisco startup co-founded by Sebastian Thrun and Adrian Aoun, pulled in before it has shipped a single thing (The Information via Dealroom). Aoun announced it on October 29, and the backer list reads like a Rolodex of Silicon Valley royalty: Eric Schmidt, John Doerr, Marc Benioff, Yuri Milner, and Ram Shriram, plus firms including Construct Capital and Lowercarbon Capital (TokenPost).

A seed round is supposed to be the small one — the napkin-sketch money. Most companies don’t touch nine figures until a Series C, if ever. Dulo got there on a thesis and two résumés.

The pitch, in Aoun’s own words: “foundation models for hardware design.” That’s the whole company for now — the idea that the thing ChatGPT did for text, Dulo can do for the physical engineering of planes, cars, and complex machines.

🧠 Why This Matters

Software ate the world because software is cheap to iterate. You type, you compile, you ship, you fix. Hardware is the opposite — every physical prototype costs weeks and money, and a mistake can mean a recalled product or a grounded aircraft.

AI has barely touched that world. Large language models are fluent in Python and prose; they are hopeless at guaranteeing that a bracket won’t shear off at 30,000 feet. Dulo’s bet is that the next frontier models won’t just write code — they’ll design the thing the code runs on, and validate it against physics before anyone cuts metal.

If that works, it compresses the most expensive, slowest part of building real products. If it doesn’t, it’s a very expensive way to find out that atoms are harder than bits.

📊 Deep Dive

The founders are why this round happened at seed scale. Thrun is one of the most recognizable names in applied AI: he founded Google X, led the self-driving car project that became Waymo, co-founded online-education company Udacity, and won the 2005 DARPA Grand Challenge with Stanford’s “Stanley.” Aoun founded Forward, the AI-powered doctor’s-office startup — which is also a cautionary tale we’ll get to.

Here’s how Dulo’s seed stacks up against the funding it skipped over:

• Dulo seed: $130M — raised before a product exists (Dealroom)
• A typical seed round: low single-digit millions, enough to build a first version and hire a handful of engineers
• Bedrock Robotics Series C: $600M at a $4.1B valuation — raised with autonomous excavators already working on US job sites
• Serval Series C: $150M at a $3B valuation — raised as a shipping enterprise-automation product

Notice the pattern: everyone else raised big rounds after proving something. Dulo raised a near-Series-C sum at the idea stage. No lead investor has been named, and no valuation has been disclosed (TokenPost).

“On a mission to create an Engineering Superintelligence, building foundation models for hardware design.” — Adrian Aoun, co-founder, describing Dulo (i-scoop)

The ambition is to automate the design and validation of complex hardware — planes, cars, the hard physical stuff — and leave the actual manufacturing to others (i-scoop). Former Autodesk CEO Carl Bass, who ran the company behind the design software engineers already use, is an adviser.

⚠️ The Catch

Start with what Dulo hasn’t said. No valuation. No lead investor. No product. No customers. No launch date. No public demo. The company was founded earlier in 2026 and surfaced with a nine-figure check and a mission statement (i-scoop).

Then there’s the Aoun track record you should weigh honestly. Forward, his last big swing, shut down in November 2024 after raising hundreds of millions to put AI-driven “CarePod” health kiosks in malls. The vision was bold, the execution ran into the real world, and the doors closed. That’s not disqualifying — plenty of great founders have a failure behind them — but it’s a reminder that a dazzling thesis and famous backers don’t bend physics or economics.

And hardware is where AI’s weaknesses bite hardest. A language model that hallucinates a fake citation is annoying. A design model that hallucinates a load tolerance is dangerous. Validation — proving a design actually holds up — is the hard half of the problem, and it’s the half that doesn’t tolerate a confident guess.

🎯 What Happens Next

Watch for a product, any product. $130 million buys a lot of runway and a lot of GPU time, so the pressure now is to show something real — a design tool engineers actually use, a validation result that holds up, a named customer.

Watch the hiring, too. Foundation models for physical engineering need people who understand both transformers and torque, and that talent pool is thin. The round is big enough to pay for it; the question is whether Dulo can assemble a team that spans both worlds.

And watch whether a valuation ever gets attached. Seed rounds this large usually come with one eventually. When it lands, it’ll tell you how much of this is conviction and how much is a bet that Thrun’s name alone moves the market.

🧩 Bigger Picture

Dulo is the clearest sign yet that the AI money is moving from screens to the physical world. The same week Dulo surfaced, Bedrock Robotics raised $600M for self-driving construction gear and Foundry Robotics took $100M to build autonomous factories — all chasing the idea that AI’s next act is atoms, not just bits (Crunchbase News).

It’s also a referendum on founder premium. In a market this flush, a proven track record can raise more at a napkin sketch than most startups raise with revenue. That’s a vote of confidence in Thrun — and a test of whether reputation can carry a company through the part where it has to actually work.

The design software engineers use today can tell you if your bridge will stand. Dulo wants to build the thing that designs the bridge in the first place. $130 million says it’s worth a shot. Physics gets the final vote.

---

Sources

• Dulo raises $130M seed — Dealroom (reporting The Information)
• Dulo Raises $130 Million in Seed Funding for Hardware Design AI — TokenPost
• Dulo’s $130 million bet on AI that designs and validates hardware — i-scoop
• The Week’s 10 Biggest Funding Rounds: Physical AI, Predictions, Robotics — Crunchbase News

❓ Quick answers

How much did Sebastian Thrun's startup Dulo raise?

Dulo raised a $130 million seed round, announced October 29, 2026, according to The Information and Dealroom.

What does Dulo do?

Dulo is building 'foundation models for hardware design' — AI to automate the design and validation of complex hardware like planes and cars. As of early November 2026 it has no public product.

Who invested in Dulo?

Investors include Eric Schmidt, John Doerr, Marc Benioff, Yuri Milner and Ram Shriram, plus firms such as Construct Capital and Lowercarbon Capital (TokenPost, Oct 30, 2026).