Want to Put Your Self-Driving Taxis on Our Roads? Fix Them First.

There’s a pothole on my road that’s been there for months. It started small. It’s now about 40cm across and deep enough to make you wince every time you drive over it. The drains next to it have been clogged with mud for years, so when it rains the water just runs straight across the surface and carries on destroying the road structure underneath. Nobody is coming to fix any of it. The council is skint, and the tendering process means contracts go to whoever quotes the lowest number, which usually means a patch job that lasts eighteen months before the whole thing opens up again.

Meanwhile, companies like Waymo are making noise about expanding their autonomous vehicle operations globally, and governments are more or less treating it as something inevitable they need to accommodate. A thing that is just going to happen.

I’d like to suggest a different framing.

These are companies with enormous balance sheets. Google could put a billion pounds into road infrastructure and its shareholders would barely notice. They already have the UK mapped in extraordinary detail through Google Maps and they know the road network better than most councils do. Their entire business model depends on having smooth, reliable, well-maintained roads for their sensors and software to operate on. A potholed, waterlogged surface isn’t just an inconvenience to their fleet, it’s a technical problem.

So rather than just handing over access, what if permission to operate was tied to a straightforward condition: you want to use our roads commercially, you help pay to bring them up to a proper standard first. Not as a token gesture. Actual infrastructure funding, covering resurfacing and drainage.

The obvious worry is that this just invites corporate capture of public infrastructure. You don’t want Waymo paving a road in some proprietary way that only plays nicely with their own vehicles. That would be a disaster, a private monopoly on what is supposed to be shared public space.

The answer to that is a consortium model, similar in principle to how landing slot access works at major airports. Rather than one company funding and therefore controlling an upgrade, you’d have a collective body made up of the various tech and automotive companies with a stake in autonomous driving, pooling contributions into a standardised programme. No single company dictates the spec. The result is open-access infrastructure that any compliant operator can use, which is better for competition and better for the public.

There’s a reasonable counter-argument that AV companies could just say no and take their pilots somewhere more accommodating. That’s true, but it only holds if one country tries this alone. With coordinated requirements at a national or European level, the negotiating position looks rather different. These companies want access to large, mature markets and they don’t have unlimited alternatives.

None of this should be a blank cheque for any operator to put unsafe software on public roads just because their consortium dues are paid up. Safety standards need to be rigorous and independently verified before anyone gets a licence. And if something does go wrong, if a company has been sloppy and one of its vehicles causes an incident, the response shouldn’t be to shut down the whole sector. A targeted suspension on that specific operator, while everyone else who has been doing things properly carries on, seems like the right approach.

The roads aren’t going to get better on their own. Vehicles are getting heavier, not just because of SUVs but because even ordinary electric cars carry substantial battery weight, and maintenance budgets are being stretched in the wrong direction.

If autonomous taxi companies want to use our streets as a commercial platform, it seems entirely reasonable to ask them to contribute to making those streets worth using. The pothole outside my house isn’t going to fix itself.