RJ Scaringe of Rivian thinks we’re getting robots wrong

If you haven’t heard, Rivian founder and CEO RJ Scaringe has another company — the third one by our count. This time we focused on robots.

The serial entrepreneur is best known for Rivian, the buzzy electric vehicle maker that was thrust into the spotlight in 2018 after debuting an electric truck with its now-famous gear tunnel.

The publicly traded company is now a few months away from delivering its most important electric vehicle yet, a midsize SUV called the R2. This electric car is cheaper to build than its flagship R1 counterpart (“about half the cost,” Scaringe told me), yet it is more technologically capable in many ways. It will also be cheaper for buyers. The R2 First Edition will start at $57,990, 20% cheaper than the base R1T pickup.

Then there’s also a micromobility startup that started as a skunkworks program inside Rivian and launched last year with backing from venture capital money. Also, in which Rivian maintains a minority stake, launched a modular pedal-assisted electric bike and charging quad bike last October. Scaringe is deeply involved in that startup as well.

Now, Scaringe has robots on his mind. He called his new startup Mind Robotics, which I learned was not its original name. (More on that below.)

Mind Robotics wasn’t born within Rivian’s walls, although the automaker might one day become a customer. Mind Robotics is a private company founded by Scaringe that recently raised a $500 million Series A round co-led by venture firms Accel and Andreessen Horowitz. The company, which has raised $615 million since its founding in November 2025, is valued at about $2 billion.

I sat down with Scaringe on the sidelines of South by Southwest, of which his company is a major sponsor, to talk about everything he’s working on. The Q&A below, which has been edited for brevity and clarity, focuses our discussion on robotics and how his new startup is taking a fresh approach.

Have you finished separating companies?

Three companies? (Pause) Maybe.

Are you sure about that?

Yes. I mean, yes. (Nods in agreement) Also [the ebike company] Here, so this is great. Nice to see. It’s great to see brands coming together, which makes me happy.

mind [Robotics] It’s something I’ve been working on for a while. It was also unique because it was built inside Rivian, and we took it out. Mind you… I’ve started a brand new company, and Rivian is a partner. I’m very optimistic about that. It really benefits Riven, but I think it has the potential to become a very big business, given the scale of what’s going on.

When did you start working on Mind?

So, the mind is an interesting story and it actually goes back to R2.

For about two years, we’ve been building a lot of confidence in R2. So, if we are confident about R2 it means we think there will be high volume. If we think there will be a lot of volume, that means we think we will need a lot of plants [factories].

And effectively, I said, “Boy, if we have to build four or five plants over the next decade, that means we’re going to spend many, many billions of dollars in capital expenditures.” What will these plants look like? We don’t want to build a collection of plants and then make them immediately obsolete.

I started an effort to study the future of manufacturing, and I basically came to the view that classic industrial robots would continue to exist like what you see in our factory today, or in a Tesla factory or a Ford factory. But the idea of ​​robots with human-like skills is going to be really important.

We already had a strong thesis about artificial intelligence in the physical world. Every company I’ve met is working on some form of robot that can perform human-like skills.

I would describe it broadly as there are companies that fall into the current category of industrial robotics companies. And I came to the view that they would continue to exist, but they wouldn’t be the kind of company that made robots that could do human-like tasks.

Then there is a new group of companies that all effectively resemble startups. Surprisingly, most of them focus on the home – which we can spend hours in. A very large number of these companies focus on folding towels and washing dishes, a subset of those looking to the industry.

We just figured out that the ingredients needed to be successful, none of them had all of them. I felt like I could build a better company.

I’m not going to base Rivian’s future manufacturing dependency on companies that have never made a product or understood or spent time understanding industrial processes, or that don’t have a data flywheel for training models, or that don’t have supply chains because all of those companies, we’re going to need to help them learn how to become a real company.

I made the decision to start a new company, and then I had to figure out what the management structure would look like.

The day I launched the company, I raised a seed round, and then we just did our Series A funding.

When you originally founded or registered it, was its name something else?

What was the name of the project? I’ll tell you, no one asked me to do that. (Pauses) So it was Project Synapse.

Why?

There are a lot of reasons. This is a great question, no one has asked it. So the company’s name is “Mind”, which kind of links to the synapse and links to the brain.

Like shooting and missing.

Yes, exactly. Both were due to the brain connection. At that time, my children were in a school called Synapse, and my children inspired me and I thought about business, so I connected them. I actually almost called the company Synapse, but that’s a very funny word; I wanted something simpler, but yeah, it was almost tangled.

Funny enough, one of the board members wrote me a note and still refers to it as Project Synapse.

Who’s on your board?

I wish I could [Behl] from Eclipse, Sameer from Accel, and Rivian have a representative on the board. It’s a private company, so it’s much easier than a public company.

You talked about the different classes, which is what you plan to do very closely related to what Boston Dynamics is working on, which is the Human Atlas, but coupled with the AI ​​lab. [Google DeepMind]?

We build models too. We build the models, robots, and infrastructure to deploy them at scale.

Then in terms of mechatronics and robotics themselves, there has been a surprising focus on mimicking human biomechanics, or in some cases, going further, to make more complex mechatronics.

I think that’s what was missed in industrial [robotics] This is one of the things that we see clearly, is that the work is done with the hands. So, hands are very, very important. Everything else, from the point of view of the automated system, is putting the hands in the right place. And so the ability of robots to do really complex movements, like, let’s say, like a backflip actually means that the robot has a lot of unnecessary complexity in the vast majority of tasks.

I understand that the purpose of showing this is to demonstrate the flexibility and capability (of humanoid robots). But if you go to a Rivian facility, you’ll see very few people with the kind of flexibility to do a backflip. Therefore, when you think about a large-scale deployment, you want to reduce complexity, reduce the number of failure modes, and reduce power consumption.

So I think many robotic systems, especially human systems, that are being considered for manufacturing are too complex for what they do. They’ll be human, they’ll have hands, and there’s a perception model at the top. There will be the ability to edit, X, Y, Z; There will be movement to allow it to move. But I think that mimicking human biomechanics in a manufacturing environment misses some of the fundamental points of manufacturing, which is that everything depends on the hand.

I’m trying to imagine what this will look like in the end or what the different variations will be.

So, hands are the most complex part of robots.

(jumps quickly) But it’s also where you focus your energy, your dollars, your spending, everything that should be in your hands.

And I think the other thing to note is, if you’re building a business that’s going to serve a lot of different industrial sectors of manufacturing, there’s not going to be one set of hands that’s going to be perfect.

If you think about challenging the hands it is torque amplification. If you’re picking up a large four-inch steel tube, this is a different solution than screwing an M4 fastener into an aluminum die. They’re just very different skills and very different levels of dexterity.

One of the other things that happens in robotics, which is very surprising, is that in the biological world we clearly realize that evolution has allowed us to create, like, very different ideal sets of biomechanics for different things. The best thing for swimming looks very different from the best thing for running and the best thing for climbing looks different from the best thing for lifting. For this reason, humans are not particularly good at swimming. For example, if you put me against a dolphin, I will lose. If you put me against a cheetah, I’ll lose.

We are, of course, stronger in our brains. And so in manufacturing, the assumption that the shape and form factor of humans is the optimal shape and form factor for all types of work, I think misses the fact that we didn’t evolve in plants. Now plants have evolved around us, and this is an important point. There is huge infrastructure built into the brown domain that the robots have to connect. But the part they have to fit in is that they have to fit roughly into the space that humans allow, and they have to have hands. I point this all out because we have a different point of view than a lot of robots that often start by saying we’re going to work at home, which is very different. [operational design domain].

But like all these use cases that say we’ll do everything a human would do. It’s a different ODD, so it’s going to result in a different form factor than if you said, I’m just going to focus on being a manufacturer.

And in fact [in an industrial factory] You don’t have to deal with stairs, you don’t have transitions from carpet to tile, and you don’t have to worry about tripping on a cat. You don’t have to worry about stepping on a child. Therefore, you have constraints that are more suitable for large-scale deployment. And you can map the environment, and the environment doesn’t change, right?

The only thing that would change the dynamic is any interaction with humans [in a factory].

This is an important point.

That’s why the UI [user interface] Very important. We haven’t shown what it will look like, but I’ll give you the synopsis [sic]: She must feel friendly, because she will be working for a long time with humans. But he doesn’t want to feel stupid and end up with a “kick me” sticker on his back, right?

A lot of robots have also moved towards these very lean, muscular shapes that look like the Terminator. i don’t know why. I think it was science fiction movies or something like that that prompted the industrial design world to do this. There’s a robot version and they all blend together. These are very athletic female runners’ body shapes. I think a different form factor could look friendlier, but not stupid.

I’m starting to see a thread that explains what else you’ve done, and your thinking, regarding modularity with the mind.

Some of it, yes.

The other thing is Rivian. If you asked someone in 2018 what a 1,000 horsepower truck looked like, and you closed your eyes, you wouldn’t think of the R1 [truck]. And so, I think the idea is that you can have all this ability, but it doesn’t have to look scary or intimidating. It can be friendly – ​​it’s a design and aesthetic belief that will carry over to robots as well.