1. Concrete and asphalt significantly attenuate radio signals, especially reinforced concrete.
2. Both concrete and asphalt expand and contract, and would likely damage the chips.
3. Asphalt isn't exactly solid at all times, compounding the issue above.
4. Reading said chips would be difficult, as they would only be in range for a few tens of milliseconds.
It's interesting that RFID developed out of, partly, a desire by freight train operators to track traincars as they moved around switching yards (moving past fixed readers). The Eurobalise system is sort of the reciprocal of that idea.
And here's a reverse system where the track identifies the passing wagons: https://turckvilant.com/success_story/running-at-250km-h/
They could instead be mounted on the lane-separation poles / walls, on the cat-eye embedded marks, and other things that are installed on the road anyway, and have a good protection and line-of-sight.
The nice thing is that you really don't need to use any intelligence when placing the markers. They could be permanently set with a random GUID at time of manufacture, and then just installed in the road. After installation, someone (e.g. Google Maps car or crowdsourced project) could drive the road and cross-reference the RFID GUIDs to their actual location.
Crazy idea: turn road stripes into magnetic tape, encode the data and read it like a tape player :P