I think that more than physics the bottleneck for this is political (at least in the US). All of the local large projects around me are expensive because of massive amounts of red tape (environmental studies, zoning, planning), and political patronage systems. After the kick backs, political donations, promises to only work 8 hours a day, only use union labor, hire x police officers for y hours in overtime security positions a month, use xyz contractor etc. a small cost seems to be the actual labor and materials. Hell these robots if they work will be made illlegal.
I'm the CTO and one of the founders of Bedrock. I was very pleasantly surprised to see the excitement from this crowd! Happy to answer any questions about us (and will look through the comment threads here). We're looking for really really awesome MLEs and software engineers, so if you're interested take a look at our careers page https://bedrockrobotics.com/careers
One big barrier I haven't seen mentioned is all the OEM competition they are going to face.
Caterpillar, John Deer, etc. already have remote operation vehicles. And a lot of provisions on what types of kits can be retrofitted onto their equipment without violating their terms/warranties.
I'm sure this is already something they've taken into consideration, but it seems like this will be more focused on partnerships with existing OEMs rather than selling add on kits to current fleets.
I'm an architect. Well really now an architecture professor who (elevator pitch) writes computational tools to implement sophisticated geometries in advanced fabrication...
That's all to say I want advancements in jobsite automation desperately. But it's WAY harder than people from other domains think. Imagine driving on a road while you're also building it while others are doing both around you too...
These folks seem to be concentrating at the moment on excavation which (without looking) if I recall is already a pretty active and developed in terms of automation. But get out of the ground and you hit some pretty big issues pretty quick. To get a sense, heres's one of my go-to articles when people wonder about jobsite automation...
Unrelated apart from the words "rock" and "construction": I wouldn't be surprised if we see dry stone construction becoming very practical thanks to advances in computer vision and robotics.
Dry stone construction is incredibly durable -- it doesn't rely on mortar which can weather away -- but it is limited by needing to reshape stones to fit together tightly (often by making flat surfaces). A human stonesmith can look at a handful of stones and find one which is close to fitting in the necessary spot; but a computerized system could scan thousands of stones and build tightly-fitting stonework with minimal need for reshaping.
I didn't think open field construction was hampered by the humans in the loop? Quite the contrary, I was under the naive impression that the heavy machinery was already largely doing the vast majority of the work. Even when operated by a human.
Will be neat to see where this goes. But I'm reminded of some Amazon guys that were supposed to revitalize the supply chains. My memory is that that didn't work out so well.
This is tangential, but I recommend Katsushiro Otomo's (of Akira fame) dystopian take on large-scale automated construction, the short movie Construction Cancellation Order, part of an anthology known in the west as Neo Tokyo[1], released in 1987 (Akira was released in 1988).
I won't link to it here, but it seems someone uploaded it to archive.org (most likely illegally).
Civil engineering is already a field where the very largest projects are done by humans planning and building the roads and bridges for the robots to move in (such as things rented from Mammoet [1] with extra control systems), but it does require significant human oversight (typically a metaphorical red button).
It's all very one off and specific, and given how big those projects are that seems unlikely to change. The manufacturing of suburbs though would be a whole different ballgame.
This feels more plausible for long-term success than Waymo for a number of reasons:
* Construction sites are smaller in footprint
* They’re more easily covered in sensor networks to support autonomous operations
* They’re typically controlled-access environments, which reduces potential interruptions to automated routines
* Since they have higher risk profiles than public infrastructure, the expectation is that workers will be more aware of autonomous operations and any deviations before they cause serious harm
Honestly, I’d wager closed-site autonomy takes off before anyone nails national or global self-driving on existing infrastructure.
I could see this really accelerating strip mining or deforestation in places where there either isn't a lot of local opposition or there is a favorable regulatory environment. It makes me wonder if Tinto is an investor.
I've been building a voice+vision AI assistant for tradesman, industrial services, and DIY consumers the last 8 months.
It streams vision+voice+text+spatial data to a multimodal llm, to help users solve problems on the job. Basically a "Cursor" for tradesman. Most of my work now has been figuring out what to put in the context window, when, and how to provide the best deterministic response.
This assistant is a bridge between where we are now (little to no tech for these guys) and where the future will be (robotic automation). I think the in between stage will be a significantly longer timeframe than people realize, and I hope my app can provide value to these guys while they work. www.camerasearch.ai
Looked at the linkedIn career paths in the profiles.
Nobody there worked in construction before this.
They may have hired some domain experts, but i doubt they will get it together.
Especially, as automated construction has very little to do with "copy human" constructors and replicate and all todo with abstract the environment away and get it done. Meaning, you will want classic autonomy with excavators and trucks and concrete trucks - until you have poured the base plate for a house.
Then the best approach still is, to have basically a factory for pre-built components and assemble them on site. The "doing constructor things" with machines on site- does not scale.
The wasted time and money in construction comes entirely from two places: a small percentage of crooked builders (and their local council mates), and the bureaucracy that is trying to protect the citizens from same. Big brother puts in a lot of hoop-jumps standards and supposed checks and balances that end up creating massive delays and costs for the consumer, but the actual standards (while usually quite sensible) are easily sidestepped by the crooked builders, so the war continues, and the overhead constantly increases with the usual expansion-only government regulation ratchet.
None of these things are susceptible to "AI" and other such automation. We have had prefab construction for decades.
> Bedrock Robotics is focused on developing a self-driving kit that can be retrofitted to construction and other worksite vehicles
> Bedrock is “upgrading existing fleets with sensors, compute, and intelligence that understands project goals, adapts to changing conditions, and executes work around the clock,”
I can also imagine this applying to all kinds of mining too, where there's already all the heavy equipment to mine and transport resources and we're just turning it into a robot so they don't have to employ a human anymore.
What’s interesting here isn’t just the move from self-driving cars to construction, it’s how Bedrock is betting that autonomy will take off faster in industries where safety, labor shortages and cost pressures already demand it.
I'm skeptical of the plan to provide a "kit" that can be put on arbitrary construction vehicles. I think Cruise tried something similar with personal vehicles and quickly learned it really could only build a system for a specific vehicle. There are tons of complexities associated with mounting of sensors and interfacing with the vehicle specific control systems.
Not saying it can't be done, but I wouldn't be surprised if the company has to pivot to supporting a subset of their goal and then gets bought out by some larger company to work on their line of construction vehicles.
There are already so many specialized machines in construction. When I was a kid I thought we needed exo-skeletons. Then I watched modern construction happen and realized we already have all these wild Bobcat like machines doing all the heavy lifting that humans have done in the past.
I guess it's cheaper to use a computer to drive a Bobcat than a human driver. But I wonder if the cost is worth the loss in immediate knowledge and speed. There's no way a computer is going to be faster than a human while also being safe. So I have my doubts.
From the imagery, they seem to be focused on earthmoving. Some cursory googling indicates an excavator operator can cost on the order of $1000/day (to the end customer, presumably less to an owner-operator). For some loose context, a slab-on-grade warehouse project I was distantly involved with a couple years back budgeted some $3.5M of sitework (includes other things than just earthmoving), which was about 10% of the total construction cost.
Plug in your own longevity and uptime assumptions, but if that roof-mount kit costs on the order of $50k, it seems viable.
To build a commercially successful autonomous bulldozer requires building a commercially successful bulldozer. That’s hardware and hardware is hard. Probably harder than the autonomous part because bulldozers are a century past the proof-of-concept era.
My cynical take is this is financial engineering more than construction engineering. YMMV.
Definitely echo the concerns about bureaucratic red tape (looking at you, California high-speed rail fiasco) that kind of environment makes innovation in infrastructure extra hard. Still, there's something compelling about Bedrock’s approach if they can genuinely deliver a system that adapts in real-time to the complex nature of construction sites. The big question is whether they can win through retrofitting existing fleets or by locking in tight partnerships with OEMs, adding to that the competition is going to be pretty tough
Is this a scenario where offshore operators doing remote equipment control would be 90% as good as a US union worker for 15% of the price and could work in shifts 24/7 (e.g. for mining operations)? Sensor + control data would be great training for future AI.
Jumping straight to autonomous operations seems expensive/hard.
Eg. if bricklayers could talk to their machine the way we can with coding agents, and say “yep, wall here please, check the blueprints to confirm how high and where the holes for the windows go”, retired & injured tradespeople could choose to come back. Less injuries means cheaper insurance & better margins. People could work in multiple parts of a site by supervising several robots, and not be exhausted at the end of the day. The list of benefits to individuals and the industry is long.
First, real estate in US designed to be the sink for excess money globally gained by nefarious means, sell and make the money clean.
Second, all the prep work before actual construction costs the same regardless of the size of the home. Its profitable for builders to build large homes (cheaper per square foot) and sell them at a premium.
Third, zoning laws that make anything thats not a SFH mostly illegal to build.
Fourth, manufactured homes are illegal in most areas.
Finally, The building codes are localized to each county, there is no federal/state building code.
As long as these perverse incentives exist, costs are not going to come down.
Automation will break the back of capitalism. That's a good thing. But it's going to get very bad before that happens.
There are a ton of jobs that should be automated like working in an Amazon warehouse. Automation can go one of two ways: it can make all of our lives easier or it can displace the workforce and suppress the wages of those who remain so those are the very top can have $250 billion isntead of $200 billion.
Those displaced will be told to find new jobs and that will work for a time with considerable personal hardship for those affected. We've seen the impacts of this with manufacturing jobs being shipped overseas from the Rust Belt, for example. But at some point we'll start to run out of jobs for people to go. And then things will get very, very bad.
I imagine a distant future where food is grown with automation, houses will be built that way, robots will pick up the trash and so on. This will leave people to find more meaningful pursuits. But this will require the ultra-wealthy to share and history has taught us that this will get bloody.
Instead we'll face a future where companies will fail because there simply aren't customers because nobody can afford anything.
I welcome less labor intensive constructions but construction is a significant (~20%?) sector of the economy. The potential for negative impacts from such mass layoffs is enormous.
and as others have mentioned: there's a ton of cost built in that really comes down to corruption and overcoming NIMBYism.
Ex-Waymo engineers launch Bedrock Robotics to automate construction
(techcrunch.com)518 points by boulos 16 July 2025 | 386 comments
Comments
Caterpillar, John Deer, etc. already have remote operation vehicles. And a lot of provisions on what types of kits can be retrofitted onto their equipment without violating their terms/warranties.
I'm sure this is already something they've taken into consideration, but it seems like this will be more focused on partnerships with existing OEMs rather than selling add on kits to current fleets.
That's all to say I want advancements in jobsite automation desperately. But it's WAY harder than people from other domains think. Imagine driving on a road while you're also building it while others are doing both around you too...
These folks seem to be concentrating at the moment on excavation which (without looking) if I recall is already a pretty active and developed in terms of automation. But get out of the ground and you hit some pretty big issues pretty quick. To get a sense, heres's one of my go-to articles when people wonder about jobsite automation...
https://www.construction-physics.com/p/where-are-the-robotic...
Dry stone construction is incredibly durable -- it doesn't rely on mortar which can weather away -- but it is limited by needing to reshape stones to fit together tightly (often by making flat surfaces). A human stonesmith can look at a handful of stones and find one which is close to fitting in the necessary spot; but a computerized system could scan thousands of stones and build tightly-fitting stonework with minimal need for reshaping.
Will be neat to see where this goes. But I'm reminded of some Amazon guys that were supposed to revitalize the supply chains. My memory is that that didn't work out so well.
I won't link to it here, but it seems someone uploaded it to archive.org (most likely illegally).
1: https://en.wikipedia.org/wiki/Neo_Tokyo_(film)
Civil engineering is already a field where the very largest projects are done by humans planning and building the roads and bridges for the robots to move in (such as things rented from Mammoet [1] with extra control systems), but it does require significant human oversight (typically a metaphorical red button).
It's all very one off and specific, and given how big those projects are that seems unlikely to change. The manufacturing of suburbs though would be a whole different ballgame.
[1] Specifically https://en.wikipedia.org/wiki/Self-propelled_modular_transpo...
* Construction sites are smaller in footprint
* They’re more easily covered in sensor networks to support autonomous operations
* They’re typically controlled-access environments, which reduces potential interruptions to automated routines
* Since they have higher risk profiles than public infrastructure, the expectation is that workers will be more aware of autonomous operations and any deviations before they cause serious harm
Honestly, I’d wager closed-site autonomy takes off before anyone nails national or global self-driving on existing infrastructure.
It streams vision+voice+text+spatial data to a multimodal llm, to help users solve problems on the job. Basically a "Cursor" for tradesman. Most of my work now has been figuring out what to put in the context window, when, and how to provide the best deterministic response.
This assistant is a bridge between where we are now (little to no tech for these guys) and where the future will be (robotic automation). I think the in between stage will be a significantly longer timeframe than people realize, and I hope my app can provide value to these guys while they work. www.camerasearch.ai
Especially, as automated construction has very little to do with "copy human" constructors and replicate and all todo with abstract the environment away and get it done. Meaning, you will want classic autonomy with excavators and trucks and concrete trucks - until you have poured the base plate for a house.
Then the best approach still is, to have basically a factory for pre-built components and assemble them on site. The "doing constructor things" with machines on site- does not scale.
None of these things are susceptible to "AI" and other such automation. We have had prefab construction for decades.
> Bedrock is “upgrading existing fleets with sensors, compute, and intelligence that understands project goals, adapts to changing conditions, and executes work around the clock,”
I can also imagine this applying to all kinds of mining too, where there's already all the heavy equipment to mine and transport resources and we're just turning it into a robot so they don't have to employ a human anymore.
> adapts to changing conditions
The real play here is starting a business that specializes in getting construction equipment unstuck from the mud.
And an $80M round sounds sane these days
Not saying it can't be done, but I wouldn't be surprised if the company has to pivot to supporting a subset of their goal and then gets bought out by some larger company to work on their line of construction vehicles.
I guess it's cheaper to use a computer to drive a Bobcat than a human driver. But I wonder if the cost is worth the loss in immediate knowledge and speed. There's no way a computer is going to be faster than a human while also being safe. So I have my doubts.
Plug in your own longevity and uptime assumptions, but if that roof-mount kit costs on the order of $50k, it seems viable.
To build a commercially successful autonomous bulldozer requires building a commercially successful bulldozer. That’s hardware and hardware is hard. Probably harder than the autonomous part because bulldozers are a century past the proof-of-concept era.
My cynical take is this is financial engineering more than construction engineering. YMMV.
https://en.wikipedia.org/wiki/Self-replicating_spacecraft#Vo...
Jumping straight to autonomous operations seems expensive/hard.
Eg. if bricklayers could talk to their machine the way we can with coding agents, and say “yep, wall here please, check the blueprints to confirm how high and where the holes for the windows go”, retired & injured tradespeople could choose to come back. Less injuries means cheaper insurance & better margins. People could work in multiple parts of a site by supervising several robots, and not be exhausted at the end of the day. The list of benefits to individuals and the industry is long.
Second, all the prep work before actual construction costs the same regardless of the size of the home. Its profitable for builders to build large homes (cheaper per square foot) and sell them at a premium.
Third, zoning laws that make anything thats not a SFH mostly illegal to build.
Fourth, manufactured homes are illegal in most areas.
Finally, The building codes are localized to each county, there is no federal/state building code.
As long as these perverse incentives exist, costs are not going to come down.
Cool idea, but a lot cooler to be able to attract talent with QSBS stock
There are a ton of jobs that should be automated like working in an Amazon warehouse. Automation can go one of two ways: it can make all of our lives easier or it can displace the workforce and suppress the wages of those who remain so those are the very top can have $250 billion isntead of $200 billion.
Those displaced will be told to find new jobs and that will work for a time with considerable personal hardship for those affected. We've seen the impacts of this with manufacturing jobs being shipped overseas from the Rust Belt, for example. But at some point we'll start to run out of jobs for people to go. And then things will get very, very bad.
I imagine a distant future where food is grown with automation, houses will be built that way, robots will pick up the trash and so on. This will leave people to find more meaningful pursuits. But this will require the ultra-wealthy to share and history has taught us that this will get bloody.
Instead we'll face a future where companies will fail because there simply aren't customers because nobody can afford anything.
I welcome less labor intensive constructions but construction is a significant (~20%?) sector of the economy. The potential for negative impacts from such mass layoffs is enormous.
and as others have mentioned: there's a ton of cost built in that really comes down to corruption and overcoming NIMBYism.
Seems sensible a project. $80m raised also seems a sensible amount. And the guy has a background in this field. Good luck