The car became a consumer electronic — notes from a week in Shanghai

I spent an intensive week in Shanghai recently, supporting our local team and meeting current and future customers. I've been coming to China for more than ten years now and it still rearranges my assumptions every time — in some ways it feels like a trip into the future and the past simultaneously.

The thing that stuck with me this time wasn't in a customer meeting. It was in the back of a Didi.

Several of the taxis I rode in had a transparent chassis system — 透明底盘, the composited under-vehicle view that lets a driver effectively see through the floor when parking or navigating a tight space. In Europe I had filed that away as a premium feature, something you find near the top of an options list. In Shanghai it was in ordinary working taxis with serious mileage on them.

Feature parity arrived quietly

The numbers behind that are the part worth sitting with. More than 70% of new cars sold in China now ship with L2 or higher ADAS. Cars priced around ¥196,000 — roughly €25,000 — come with systems I would have called premium two years ago.

The easy read is that features got cheap. I think that misses it. What actually happened is that the car changed category.

The replacement cycle tells the real story

Chinese industry data puts the average age of a used new-energy vehicle at around 3.4 years, against roughly 8.6 years for a used petrol car. Somewhere near 90% of NEVs get replaced inside five years, while about 70% of petrol cars stay with their owner longer than that.

Those aren't the numbers of a durable asset. They're the numbers of a device. People are buying cars on something much closer to a phone cycle, and everything downstream of that assumption changes — how the thing is sold, how it's serviced, and above all what the software is expected to do after the sale.

The retail side made the same point visibly for a while. NIO opened its Shanghai flagship inside the HKRI Taikoo Hui mall in 2017, complete with a library and an art gallery, and for several years the mall showroom was the signature of the new Chinese brands. That particular experiment is now receding — EV brands have been pulling out of mall space, with robotics and sportswear brands moving into the units. The storefront was a phase. The buying behaviour it was designed around stayed.

What that does to software expectations

This is where it gets directly relevant to anything I work on. In the customer conversations, the expectation that came up repeatedly was constant updates. Not "does the software work" — that's assumed. The bar is whether the experience still feels new and relevant a year in. If the car is a device on a three-year cycle, the UX has to keep earning its place for those three years, and a release cadence measured in model years doesn't survive contact with that expectation.

The regulatory posture follows the same logic. Local OEMs have adopted the Western frameworks — ASPICE, AUTOSAR — selectively rather than wholesale, optimising for speed over ceremony. It's a genuine trade rather than an oversight, and it carries genuine risk; the interesting question is not whether it's correct but how long the gap between the two approaches stays survivable for the slower side.

The uncomfortable part

If your development assumptions are built around a fifteen-year vehicle life and a multi-year release cadence, you may be building very carefully for a product category your competitor has already left.

What I can't tell you is whether a three-year replacement cycle is a market that stays that way, or an artefact of how people treat any new technology before it settles down. Ask me again after the next trip.

Close ↑