On November 18, 2026, a very ordinary thing happened in driveways and parking lots across the country: people plugged their phones into their cars and set up electric vehicle routing on CarPlay and Android Auto. It sounds like a chore. A settings menu. A checkbox you tick once and forget.
It isn't. It is one of the clearest examples of artificial intelligence quietly moving from a demo you show someone at a party into a system that makes a real decision for you, one with real consequences. An EV router decides where you stop, when you stop, how long you sit there, and whether you roll into your destination with a comfortable buffer of charge or a flashing warning light and a sweaty forehead.
That is a big deal. And it tells us a lot about where AI is actually heading: out of the chat window, into the hardware, and into the boring-but-critical infrastructure of everyday life.
Standard navigation answers one question: what is the fastest way to get there? For a gasoline car, that is basically the whole problem. Gas stations are everywhere, a fill-up takes five minutes, and the math rarely breaks.
Electric vehicles break the math in several places at once. A planner has to know:
None of that is a lookup table. It is a prediction problem. And prediction problems, at scale, are what AI is good at.
When you set up EV routing and let your phone plan a long trip, several models run at once, most of them invisible.
The system builds a personalized estimate of how much energy you use per mile. It learns from your actual driving history: how hard you accelerate, how fast you cruise, whether you crank the climate control. Two drivers in identical cars get different predictions, because they drive differently.
This is the counterintuitive one. Electric cars do not charge at a constant speed. Charging is fast when the battery is low and slows down dramatically as it fills. A naive planner tells you to charge to one hundred percent. A smart planner tells you to stop twice for shorter periods instead of once for a long one, and gets you to your destination sooner.
That is not a rule a human wrote. It is an optimization that has to be solved fresh for every trip, every car, every temperature.
Chargers break. They get busy. They get blocked by a car that is parked and not even plugged in. The best routing systems factor in which stations are likely to work and likely to be available when you arrive. That is a live data problem that looks a lot like the recommendation systems used in retail.
Traffic changes. So does weather. A route that made perfect sense an hour ago may not make sense now. Good EV routing recalculates quietly in the background rather than making you press a button.
Stack these four layers together and you have something genuinely useful. Not a chatbot. Not a generated image. A small, narrow, deeply practical AI that saves you forty minutes and a bad mood.
The most interesting strategic point is where this intelligence lives. It does not live in the car. It lives in the phone that projects onto the car's screen through CarPlay and Android Auto.
That is a structural shift, and it has three causes:
The car is becoming a screen, a battery sensor, and a set of wheels. The thinking is happening elsewhere. For automakers, that is uncomfortable. It means the feature customers care most about on a long drive may not be one the automaker built or controls, unless it exposes battery data cleanly and partners well.
The setup itself is simple, but each step matters because the AI is only as good as the inputs you give it.
Those six steps are the difference between a system that works and one you stop trusting after one bad trip.
Being in the routing graph now matters more than a highway billboard. If your station is not in the data layer that planners read, you effectively do not exist for a huge share of drivers. Reliability becomes a marketing asset, because the models that predict uptime will reward it.
The competitive pressure is to expose clean, standardized battery and vehicle data so third-party routing works well. A car that plans badly, or refuses to share, will feel broken next to a competitor that does not.
Route optimization for electric vans and trucks compounds fast. Shaving charging time off every vehicle every day is a direct cost saving, and it is the same underlying technology as consumer routing.
Charging stops create predictable dwell time, often twenty to forty minutes. Knowing where drivers will stop, and roughly when, is a location-intelligence goldmine for coffee shops, restaurants, and rest stops.
Aggregate routing data signals where demand will land and when. That has real value for grid planning, even when individual data stays private.
Privacy. A routing system knows where you live, where you work, when you leave, how you drive, and when your battery is low. That is a rich profile. Permission design and on-device processing are not nice-to-haves here.
Over-trust. AI predictions are probabilities, not promises. A charger that the model thought was open may be broken. Drivers need to keep a human buffer, and good systems should communicate uncertainty rather than hide it.
Data deserts. Routing quality depends on trip data. Areas with fewer EV drivers generate weaker models, which makes those areas less attractive to drive in. That is a feedback loop worth watching.
Standardization. If every automaker exposes battery data differently, only the largest platforms can integrate well, and smaller players get locked out.
The most important AI systems of the next few years will probably not be the ones that write essays or generate video. They will be the ones that make a thousand small decisions you never notice, which charger to use, how long to stay, whether to reroute now or in ten minutes.
EV routing on CarPlay and Android Auto is a preview of that world. It is unglamorous. It is also the kind of AI that people will quietly refuse to live without, the way they refuse to live without a maps app today. The setup takes a few minutes. The shift it represents will take a decade.