The Robots Already Working in American Factories | Tutor Intelligence

Josh Gruenstein · Tutor IntelligenceBearish

ABB's robots are built for integrator-style repetitive work, not co-designed with frontier AI models.

Gruenstein groups ABB with Fanuc. He says ABB's robots are mostly not built in the US, and even when they are, they are designed for systems integrators and fast repetitive motions rather than for the trade-offs needed to co-design with frontier models. He presents this as a reason Tutor had to design its own hardware, so the view is a competitive critique rather than a financial forecast.

“the robots that Fanuc and ABB are building... they're building these objects that are designed for systems integrators, that are designed for fast repetitive motions. They are not cut from the set of trade offs that you would want to co design with frontier models.”at 17:51

Josh Gruenstein · Tutor IntelligenceBearish

Fanuc's robots are built for repetitive integrator work, not co-designed for frontier AI models.

Gruenstein says the robots Fanuc builds are mostly not made in the United States. Even where they are, they are designed for systems integrators and fast repetitive motions, not for the trade-offs needed to co-design hardware with frontier models. He implies this leaves Fanuc's robots poorly positioned for the general-purpose, AI-driven robotics that Tutor is building, so the claim is a competitive critique rather than a financial forecast.

“the robots that Fanuc and ABB are building... they're building these objects that are designed for systems integrators, that are designed for fast repetitive motions. They are not cut from the set of trade offs that you would want to co design with frontier models.”at 17:51

Josh Gruenstein · Tutor Intelligence

No robot startup has yet been massively economically successful, Gruenstein argues.

Gruenstein says that to his knowledge no robot startup has become really massively economically successful. He gives two reasons. First, robots lacked the core primitive of general intelligence, which is what lets unit economics reach a viable level. Second, the market is highly disaggregated, with many players trying to work with each other, which blocks the cross-stack trade-offs robotics requires. He draws on the failure of several academically oriented supply-chain robotics startups that built custom systems for single big customers and then died.

“I would argue no robot startup has been like really massively economically successful. Why is that? And I think it's because one, you don't have this core primitive, which is intelligence”at 6:43

Martin Casado · Andreessen Horowitz (a16z)

Generative-AI-style data scaling may not carry to robots; some tutors may stay needed.

Casado contrasts generative AI, where training on one data distribution yields outputs people want, with robots, where the economically valuable tasks may be high-dimensional and require data slightly outside the training distribution. He argues it is not clear this carries over to robots. He suggests that even if 90% of tasks can be solved, the last 10% may need human tele-operators for a while, because collecting enough data for that tail would take too many tutors over time.

“It's not clear that this is actually the case with these robots, right? You could train it on a distribution of data, but then actually what you need is just slightly out of that distribution.”at 13:03

Josh Gruenstein · Tutor Intelligence

Frontier labs post-training on 3D spatial data is transferring to robot spatial reasoning.

Gruenstein says frontier models knew a lot about the physical world but could not express it, lacking the last mile of spatial reasoning needed for robot actions. He says the labs are now post-training on 3D spatial data such as CAD and 3D modelling, because there is a direct market they can monetise. That is exactly the spatial reasoning Tutor's robot models need, so he argues Tutor has been catapulted years ahead by drawing on general intelligence from big frontier models.

“The labs are now post training their models on three d spatial data like CAD and like three d modeling, which, by the way, they're doing because there's this direct economic incentive”at 15:49

Josh Gruenstein · Tutor Intelligence

Incumbent-made robots aren't designed for frontier-model co-design, so they lag tailored physical AI.

Gruenstein says the robots that incumbents build are mostly made outside the US, and even US-made units are built for systems integrators and fast, repetitive motions. Those designs were not chosen from the trade-offs you would make to co-design hardware with frontier models. This is why Tutor says it had to design and manufacture its own hardware from scratch, while sourcing cameras and actuators from OEMs.

“they're building these objects that are designed for systems integrators, that are designed for fast repetitive motions. They are not cut from the set of trade offs that you would want to co design with frontier models.”at 17:51

Josh Gruenstein · Tutor Intelligence

US factory robot adoption is tiny, so the market is almost entirely greenfield.

Gruenstein cites the US deploying only about 38,000 industrial robots in 2025, a tiny number. He adds that nine out of ten US factories and warehouses have no robots. From this he concludes that almost all of the market is greenfield, with most buyers having never purchased a robot. He says these buyers assume robots don't work for them, because they picture the old, custom automation industry.

“If The United States is only deploying 38,000 robots a year, if nine out of 10 US factories and warehouses don't have any robots, then almost all of the market is greenfield.”at 23:56

Josh Gruenstein · Tutor Intelligence

General-purpose robots let the US leapfrog China, whose older automation investments may become a mistake.

Josh argues that generally capable robots will be a great equalizer between the US and countries like China. China invested heavily in the previous generation of industrial automation, which made Chinese manufacturers successful, but he thinks that could prove to be an enormous mistake in hindsight. The US is doing greenfield builds that automate industrial capacity with general-purpose robots. China will have to tear down and rebuild its existing automation to make the same shift, while the US will not have to do that in the same way.

“this is going to be a great equalizer between The United States and other... countries, like China, invested in the previous regime of industrial automation.”at 39:06

Josh Gruenstein · Tutor Intelligence

Cheap, flexible robot labor will multiply factories, as cloud software multiplied businesses.

Josh argues the number of factories will grow a lot, in the same way that cloud computing and software tooling produced far more businesses. If anyone can access an infinite, flexible labor substrate through robots, then anyone can start or scale a business in the physical world. The argument rests on the analogy to cloud and software and on the idea that robot labor is flexible enough to lower the barrier to building a factory.

“I think another thing that's going to shift, which is going to be really interesting, is we're just going to have way more factories in the same way that we have way more businesses, because the cloud and all of the software tooling let people build new businesses.”at 39:40

Josh Gruenstein · Tutor Intelligence

Factory-built homes only need to be about 1% cheaper to shift developer demand from on-site builds.

Gruenstein relays a conversation with a worker on a housing factory line in Wisconsin. He says those factory-built houses cost about 5% more than traditional on-site construction. He argues that if they were even 1% cheaper for equivalent quality, developer demand would move to factory construction immediately. He sees the same dynamic in reclaiming materials from landfills and other markets, which is part of why he thinks robots could transform the physical economy.

“the houses that they're building are about 5% more expensive to build them in a factory than to do traditional construction on-site. But if these houses were even just 1% cheaper... all of the demand from developers would immediately go to that other shape”at 32:17