One recent breakdown puts the self-driving software at about 7% of what a robotaxi ride costs. Fleet operations, the cleaning, charging, repositioning, and the real estate under all of it, runs closer to 37%. The expensive part of this business is shifting from R&D to the ground the cars come home to.

Working with our AV customers, we've been learning what it actually takes to get them the properties they need, and the same shape keeps showing up, enough that it's become a kind of playbook. If you're operating an AV fleet, keep reading. We'll share what we've been seeing and how you can use this knowledge to boost your next market launch.

Fit for purpose lots

You know your operation better than anyone else. Orchestrating an AV fleet means that at any given hour some cars are driving, some are waiting for demand to pick back up, some need a charge, and some need to be cleaned or inspected. And every now and then a whole bunch of them need to get out of a bad situation at once, poor weather or something worse. Those are different jobs, and they don't want the same kind of site. None of that is news to you.

The part we've spent our time on is the ground underneath those states. Each of those needs maps to a different kind of property, priced very differently, and the right mix is hard to find. Parking for an AV fleet is a tiered real-estate problem.

The hub, and how to need less of one

The hub is the hard tier for one reason: it's the only one that needs serious power, and serious power is the most contested thing on the grid right now. Ask a utility for a few megawatts and you join a line, and the line is full of data centers. The AI buildout is the largest new draw on the grid in a generation, and a hyperscaler's campus outranks your depot in every queue it sits in. So the honest timeline to bring fresh power to a raw site keeps stretching past a year and a half, and it's getting longer, not shorter.

Power isn't the only thing pinning the hub down. The site that can charge, wash, and service cars has to be industrial, and industrial land sits on the edges of a metro, not in the dense core where the cars earn. So a centralized hub taxes you twice: once to build it, and again every day, because every car that leaves the revenue zone to charge or get cleaned runs empty miles to get there. Those empty miles are the enemy of utilization, and utilization is the whole robotaxi business case. The farther the hub sits from demand, the more uptime you quietly hand back.

So the useful work is figuring out how to need less of one big hub. A fast charger has to deliver a huge burst of power, and sizing your grid connection to that burst is what triggers the substation upgrade and the wait. A battery breaks that link. It trickles power from a small, ordinary connection all day, stores it, and dumps it fast into the cars when they plug in, so the grid never sees the spike. Picture filling a bucket from a thin hose and tipping it out in one go. Nobody plugs the battery in; it charges itself and runs on remote monitoring. NREL has measured this cutting a depot's peak grid demand by about two-thirds, and in one study a battery in place of a substation took a four-charger project from roughly four million dollars to about one and a half, live in weeks instead of a year. It doesn't beat physics, a hub run flat out all day still needs real grid power, but for most sites it changes what's possible.

It also changes where charging can live, and that's where the uptime comes back. Because a battery-backed point doesn't need a big grid connection or a full crew, you can put charging close to where the cars earn instead of all the way out at one industrial lot.

The heavy work, deep cleaning, inspection, repair, is a different animal. It needs bays, water, and people, but very little power. So that facility isn't stuck behind the same interconnection queue, which means it can go up faster and sit in a better spot. It can use a few chargers, but it doesn't need a full charging plant, because the battery-backed points out near demand are carrying most of that load. And a facility like that drops neatly into a place that's already been overbuilt.

Public fast charging got built well ahead of demand, and a lot of it is starving. Utilization in some places runs under 3%, against the 15 to 20% a site needs just to break even. For an AV operator, the prize is the site under that dead charger: it already carries the power, the finished interconnection, and the zoning for charging that you'd otherwise wait more than a year for. And an AV fleet is the tenant those sites never found: it charges every day, on a schedule, at high and steady volume. You bring the demand, the site brings the power, and both problems solve each other.

Most of this cuts against how AV operations got set up. The footprint doesn't have to be one big industrial hub run the old way. It can be built around how the cars actually behave: distributed charging near demand, batteries where the grid is slow, a rescued charging lot where one exists, and the heavy work on a low-power site you didn't have to wait on. Most operations partners weren't built for that. That's the gap we fill.

Where the fleet sleeps

There are always more cars than hub spots, even more so if you've kept your hubs small the way we just suggested, so the bulk of the fleet has to bed down somewhere else. That's the off-hail tier: when demand drops and most of the cars come off the road, they need somewhere larger and simpler than the hub to sit, and what matters there is safety and security. These are expensive, conspicuous cars, so the space has to keep them out of sight. A chain-link fence anyone can see through isn't enough; the vehicles need to be screened from the street, which rules out a lot of otherwise-fine lots. The whole fleet goes down at the same time, but it doesn't all have to sit in one place. You can spread it across a few secure sites, or pull it into one if there's a spot central to the demand area.

The obvious choice is a yard out where land is cheap. That kind of ground has a name in the real-estate world now: industrial outdoor storage, IOS, the bare fenced lots where anything too big for a building gets parked. For years nobody with a fund looked twice at them. Then logistics needed somewhere to stage trucks, investors noticed the cash flowed with almost nothing to maintain, and the overlooked truck lot became a $200 billion asset class. There are only about 1.4 million acres of it in the country, roughly one Delaware. Rents have jumped 123% since 2020, more than double what warehouses did, while supply stays frozen because cities would rather zone in housing than yards. So the obvious choice is getting more expensive even as it makes every car's morning drive longer.

The better bet is a central lot or garage you can use cheaply, and the trick is finding one whose busy hours run opposite to yours. The reflex is to assume central means expensive. But the cars sleep through the off-hail hours, which is exactly when a lot of central parking sits empty. An office garage jammed with commuter cars at 2pm is dead at 2am. Match your idle window to someone else's and you're paying for space that would otherwise earn nothing, so you skip the premium a central spot usually carries. You get the short morning run into demand and a low rate at once, because the owner had already written that capacity off. Those don't show up on a listing, which is the catch, but when you find one it beats the yard on both price and distance.

Somewhere safe to run

The last tier is the one you hope to never use. Dangerous events, a wildfire, a hailstorm, a flood, don't wait for off-hail hours; they come whenever they come, and when one does the fleet needs somewhere safe to run, all at once. The sleep yard won't do, because the same storm threatening the cars is often sitting right on top of it. So this tier is usually a lot well outside the city, somewhere that stays empty and out of harm's way, with room to absorb the whole fleet on no notice. You'd expect empty land far from town to be the cheapest of the bunch. It's often the opposite. What you pay for is a hold, the guaranteed right to move everything in at a moment's notice, the day a fire or a storm leaves you no other choice. You rarely use it, and you pay for it anyway, because the one time you need it, you need it badly.

How Mobility Places helps

Put the tiers together and the picture is a lot of different ground doing a lot of different jobs: a few hard-won powered hubs, distributed charging close to demand, sleep space matched to someone else's empty hours, and a refuge lot you pay to keep on standby. No single broker covers all of that, and no off-the-shelf listing tells you how the pieces fit your fleet, your markets, and your launch dates. Mobility Places help the teams running AV operations find the right mix across every tier. We source the off-market sites and find locations that the open market never shows. Then we hand you one platform to manage the whole fragmented footprint, every lot, lease, and partner across every part of the operation that touches parking.