Jul 16, 2025 · 41m · american-optimist
How AI Robots Are Taking Over Construction · Joe Lonsdale
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this episode of American Optimist, host Joe Lonsdale interviews robotics pioneer Boris Sofman, former executive at Waymo and co-founder of Bedrock, to explore how physical AI and autonomous machinery are revolutionizing heavy construction. They discuss Sofman's background, the technical challenges of autonomous excavation, and how physical automation addresses critical labor shortages while revitalizing American infrastructure.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Joe holds 22.5% of the talking time here. How this is scored →
Joe as informed peer 4.8 Guest teaching 2.4 Guest disagreement 0.2 Joe pushing back 0.7
speaking balance: gold is Joe, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 12:44 Boris rejects simplistic truck data generalization
Boris immediately counters the host's premise that scaling to trucks is straightforward, emphasizing that highway driving carries severe reliability constraints where stopping is not an option.
Hardest push from Joe ▶ 12:36 Joe pushes back on autonomous truck complexityJoe directly challenges Boris's earlier remark about simply adding data for trucks, pointing out intuitive differences in braking physics and maneuverability.
Biggest teaching moment ▶ 13:03 Boris explains the highway autonomy long-tail problemBoris provides a deep technical breakdown of why autonomous highway trucking requires near-perfect reliability before shipping, educating the host on edge-case sparsity.
Joe holds their own ▶ 34:11 Joe cites Milton Friedman on economic productivityJoe demonstrates his deep grasp of economic theory by citing Milton Friedman's shovel versus spoon parable to dismantle arguments against labor-saving robotics.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Joe as informed peer | Guest teaching | Guest disagreement | Joe pushing back | Why |
|---|---|---|---|---|---|---|
| Episode Preview: AI Autonomous Robotics in Construction | 3 | 1 | 0 | 0 | Joe introduces Boris Sofman, outlining his pedigree at CMU and Waymo, and frames the potential of autonomous construction robotics against Luddite opposition. | |
| Immigration Story and Appreciation for American Opportunity | 3 | 1 | 0 | 0 | Boris shares his family's immigration story from the Soviet Union and founding consumer robotics startup Anki, discussing the brutal seasonality of toy manufacturing. | |
| Scaling Autonomous Driving Stack at Waymo | 5 | 2 | 0 | 1 | Boris discusses leading Waymo trucking and freeway driving, while Joe references his discussions with Elon Musk regarding end-to-end ML replacing traditional heuristics. | |
| Safety Standards and ML Generalization in Autonomous Vehicles | 5 | 4 | 1 | 4 | Joe challenges Boris's assertion that car driving stacks easily transfer to big rigs with just a bit more data. Boris clarifies the immense long-tail safety challenge of high-speed highway operations where vehicles cannot safely stop in traffic. | |
| Founding Bedrock to Solve Construction Labor and GDP Bottlenecks | 5 | 3 | 0 | 1 | Boris outlines the rationale for Bedrock, highlighting construction labor aging and GDP bottlenecks. Joe compares the semi-controlled environment advantages against terrain irregularity challenges. | |
| Technical Deep Dive: Physics, Underground Utilities, and Machine Interaction | 6 | 4 | 0 | 0 | Boris explains imitation learning and sensor feedback for underground modeling and utility avoidance. Joe contextualizes the collaboration within the 1960s framework of man-machine symbiosis. | |
| AI Tech Stack: Multimodal Embeddings and Rapid Machine Retrofitting | 5 | 4 | 0 | 0 | Boris details Bedrock's AI architecture, comparing vision embeddings and action trajectory generation to large language models. Joe draws parallels to AlphaGo's decision modeling. | |
| Operational Deployment and High-Volume Earthwork Efficiency | 4 | 2 | 0 | 0 | Boris explains targeting high-volume repetitive earthmoving where razor-thin bidding margins make automated continuous operation commercially transformative. | |
| Economic Impact: Addressing Labor Shortages and Lowering Infrastructure Costs | 7 | 1 | 1 | 1 | Joe recounts Milton Friedman's famous spoon versus shovel parable to argue against protectionist resistance to automation, while Boris explains macroeconomic capital unlocking. | |
| Overcoming Regulatory Friction and the Demographics Imperative | 5 | 2 | 0 | 0 | Boris and Joe discuss overcoming regulatory hurdles, noting that 85% of GDP resides in physical industries where demographic decline makes automation essential. |
Statements from this episode (18)
Sofman: Early CMU robotics talent seeded Waymo, Aurora, and Uber ATG
“A lot of the people around that time ended up being the seeds that then propagated to all the, Grand Challenge, Urban Challenge, and then Waymo, Aurora, Uber, ATG.”
Sofman: Anki Reached Nearly $100M Revenue With 85% Concentrated in Q4
“It was 85% Q-four, and we were, like, kind of close to a hundred million in revenue, and so your cash flows are spiking up and down like crazy, and you're locking in forecasts and end of the summer for what you think is going to be the demand.”
Sofman: Anki's $150 Product Once Resold for $1,000 on eBay
“I remember one year we were the number one selling product on eBay, and the 150 dollar product was going for a thousand on eBay.”
Sofman: Autonomous driving AI must optimize for worst case, not average case
“This is one of those where oftentimes you think of like AI or products where it's like, yeah, get it out there and iterate. This is one of those where you're optimizing for the worst case, not the average case. This isn't search where you give some bogus resul…”
Sofman: Waymo has driven 70M+ miles and is 5x safer than humans
“Today, like there's over seventy million miles and statistically it's five times safer than a human. It's like genuinely like it is saving lives from a, on a, from a statistical basis.”
Sofman: Up to 40% of construction workers will retire within 10 years
“The average age go above 50 the up to 40% of construction workers are retiring in the next 10 years”
Sofman: US heavy machinery logs 1.25 billion operator hours annually
“There's one and a quarter billion hours of operator time operating these heavy machines just in the US. So one and a quarter billion hours a year.”
Sofman: US excavation contracts total around $120 billion annually
“There was a study we saw that there's something like a hundred and twenty billion dollars of excavation contracts per year in the US.”
Sofman: Bedrock can retrofit existing heavy machinery without building vehicles
“These machines are actually beautifully designed to where we can upfit retrofit existing machines and go to market without a vehicle.”
Sofman: Autonomous earthmoving does not require perfect earth modeling
“The interesting thing is about these physics problems is you don't even have to like perfectly model the earth. You just want to be able to throw a diversity of circumstances and be able to kind of like test out your system.”
Sofman: Autonomous construction machinery expands net jobs
“And in fact it's expansionary because you're actually creating jobs that wouldn't be done otherwise, but you have an ability to kind of move higher and higher level and create almost like a operating system for a general contractor to, Define the work they wan…”
Sofman: Bedrock can retrofit heavy machines with autonomy in under four hours
“In a way that's totally reversible, it's non-invasive, and in less than four hours we can take a machine and upfit it to be, you know, to have the system on it.”
Sofman: Excavators take 4–5 years to master due to seven degrees of freedom
“So the interesting thing is for humans, like excavators are one of the toughest machines. That's why it's like very high dimensional. There's like seven degrees of freedom and like, you know, you like all these subtle controls on the bucket, the body, the trea…”
Sofman: Bedrock is moving 600,000 cubic yards of earth for a paper factory
“We're collecting, we're working right now in a factory where you're, they have to move 600,000 cubic yards of earth to build a paper factory.”
Sofman: Earthwork execution makes or breaks profits on 1% margin DOT bids
“When you have like general contractors bidding on you know, a project for Department of Transportation in Texas or Arizona or something like that. They're bidding, they're winning the project by maybe like one percent. And the contingencies are very heavy on l…”
Sofman: Every general contractor Bedrock speaks to turns down excess work
“Every single GC we talked to has more jobs that they could take on that they just, like, physically can't, and so when that happens, market forces come into play.”