Who will recharge all those robotaxis? More robots, one CEO says.

• Humans are responsible for plugging robotaxis into charging stations and maintenance.
• Rocsys, a startup based in the Netherlands, aims to automate the charging and cleaning processes.
• Crijn Bouman, CEO of Rocsys, said operations are an oft-overlooked aspect of the robotaxi business.

How many humans does it take to plug a fleet of robotaxis into a charging station?

Crijn Bouman, the CEO and cofounder of Rocsys, did the math on that.

"Both in China and the US, it's roughly 1:12 to 1:14 ratio of head count versus vehicles in a depot," Bouman told Business Insider in an interview. "It's crazy."

Robotaxis aim to take the human out of the driver's seat, but there's another less-glamorous side of the driverless business that Bouman says should be automated: charging and maintaining the vehicles.

"The operations are a completely overlooked area, which, if you don't get it right, breaks the business case," the CEO said.

Rocsys, co-founded in 2019, aims to turn every charging port into a robocharger by essentially adding a robotic arm and software on top of the existing infrastructure.

The goal, Bouman said, is to replace humans in a process he views as wholly inefficient and unprepared for the mass adoption of robotaxis.

This interview was lightly edited for length and clarity.

What did you see that led you to start Rocsys and focus on automated charging?

I previously founded a company called Epyon, which made fast chargers for electric cars. That company was acquired by ABB, and I spent six years there as the head of the product line for electric-vehicle charging.

In my final year at ABB, we had an experience that was somewhat unusual: There was a customer from San Francisco who bought a lot of fast chargers, but they did not want to tell us what they were doing with them.

It turned out to be Cruise, the self-driving company. This was in 2017. In South San Francisco, Cruise had a big white-box warehouse where they'd built a complete indoor test track, because they didn't yet have a permit to go on public roads. The vehicle would drive for about two hours and then park itself; a person would then walk over and plug it in.

That triggered the idea: If you take out the human driver, the interface to the infrastructure is actually broken. If you want autonomous vehicles to scale, there needs to be a company that fixes the interface to the built world — such as charging, entering a parking garage, or going into a car wash. All the infrastructure is built for human drivers. That was the big idea behind Rocsys.

We first began focusing more on ports and logistics yards because, five years ago, there were few autonomous vehicles beyond those used in logistics operations and port terminals. Then, last year, robotaxis became mainstream in the US and China.

Last year, we established a dedicated team to focus on the robotaxi market in the US. We won our first very large customer contract in robotaxis — I can't talk about it publicly — and we are about to sign the second one. That contract is for automated charging at service depots for robotaxis.

Rocsys's thesis is that manual charging is prone to errors, time-consuming, and costly to operate. How does automated charging solve those three problems?

We've been in robotaxi depots in the US and China, using stopwatches to determine how much time each task takes. With charging, you walk to the vehicle, plug it in, and then half an hour later, you walk back and unplug. It may seem small, but upon examining the metrics, it's a substantial amount of work.

All the metrics add up to that 1:12 or 1:14 ratio of vehicles per head count in the depot. If you take a city like Los Angeles or the San Francisco Bay Area, you might need roughly 10,000 vehicles to operate a sizable fleet, which means you'd need to hire roughly 800 to 1,000 people just to keep that fleet running.

The reality is, it's not really a career. This job — it's just walking around in a depot outside, plugging a vehicle in, and wiping a screen. The average tenure is about three months.

The hassle is insane. These tech-savvy companies building autonomous vehicles need to establish a comprehensive parallel infrastructure of gig workers — including HR, facilities, and OSHA requirements — everything. It breaks the scale and is very costly to operate.

Our product means you don't need to build that whole parallel organization, which is a complete distraction from their core business.

Looking closely at depot operations, what do the numbers behind the actual labor look like?

Plugging in, inspecting the vehicle, performing a quick interior clean, and then walking back to unplug lasts between 300 and 400 seconds.

There are a lot of interruptions in between because charging is so important for the throughput of the depot that it gets priority. So, if someone is cleaning a vehicle, they have to stop when a new one arrives, walk over, plug it in, and then walk back. There's a lot of inefficiency.

If you automate charging, you take out roughly half the head count. Then you can optimize the other tasks so they're more sequential instead of in an ad-hoc manner. The efficiency gains are huge.

If the ratio for current depot operations is one human per 12 robotaxis, then what does the ratio look like with automated chargers?

If cleaning and inspection are still manual processes, then you will more than double the number of vehicles served per person.

We also have a road map to automate those other processes. We're currently focused on charging because that's the primary customer request, but we have developed a proof of concept for automated inspection and have already built a working system for automated interior cleaning.

What are the labor-cost differences between a depot that uses human labor versus a depot with Rocsys?

The cost savings range from 30% to 70% in the first year, depending on the dynamics of each depot. It's really significant.

A couple years ago, there was some buzz around Tesla's robot charger. What's to stop Tesla from reviving that idea and doing automated charging themselves?

Elon Musk will do everything himself, so they might. It would make a ton of sense for them. The whole pitch for Tesla is that it can upgrade the entire fleet to robo­taxis, and those vehicles need to be charged.

To upgrade a couple of million Teslas to be autonomously driven, they need an automated charging option. So I think it would be very logical for them to bring robotic charging to market. It's likely to be a captive model. Tesla has pitched inductive (wireless) charging for the dedicated Tesla robotaxi.

Current robotaxi fleets don't have that. None of the vehicles have it.

But before we even talk about robotaxis scaling, doesn't the US still have issues with its EV charging infrastructure that need to be solved first?

I think for the robotaxi use-case, it's sort of a parallel world because the charging stations are very dedicated to a fleet. So, robotaxi operators build out their fleets, but in parallel, they also develop the charging infrastructure with partners or independently. It's sort of bespoke to the robotaxi operation. Currently, the hubs are dedicated to one operator — Waymo has its own hubs, while Zoox has its own.

These differ from public charging infrastructure, so I'd say they are parallel worlds. There's no reason to solve one problem before the other.

Are robo­taxis scaling fast enough today for Rocsys to build a real business on automated charging?

Absolutely. Over the next five years, we can expect a substantial market for us.

Robotaxis are a limited fleet, but usage is extremely intense. The vehicles are running 24/7, three to four times per day. There is already a substantial market for us, with a fleet of several thousand vehicles.

The ratio of vehicles to charging bays — the parking spots where a vehicle can be charged — is about one to six. For 6,000 vehicles, you need around 1,000 charging bays.

Public information suggests that there are somewhere between 3,000 and 4,000 robotaxis on US roads, if you add up Waymo, Zoox, and others. That means the current market is a few thousand bays to serve, which is already a very nice starting market. But it's scaling so fast now. There's the Uber-Nuro announcement with a 20,000-vehicle robo­taxi fleet.

What we're seeing now already feels like a land-grab dynamic. There's Waymo, there's Zoox. Tesla — let's see how far they actually are. There's also Nuro and Uber. Wayve from the UK, which is also pretty impressive.

It really feels like a land grab in the next two years.

Are you approaching Uber, Lyft, and all the other players?

You can be sure that we talk to every one of them.

When will you be able to announce your first big robotaxi customer?

We don't have a date yet, but I'm fairly certain it'll be sometime next year.