Isn't that a solved problem since like the early 70s? Even computing power is not a limiting factor now, only those pesky humans bickering about "central planning" and "democratization".
Put in the time windows, put the capacity limits, put the pickup and delivery (this isn't same as just delivery).
It's far from solved, and far from efficient. Field isn't even mature as theoretical computer science, not even rigorous enough :D
You have papers talking about their newest hybrid genetic memetic evolutionary deep simulated annealing bullcrap algorithm getting hundreds of citations.
You have services like Routific, Routyn, Viamente, WorkWave, and others struggling with it. No one in this world has the technology powerful enough to optimize at scale.
Example. WorkWave is talking about exploiting 45Billion dollar market of optimizing thousands of technicians and their routes, but they can't even scale that on a daily basis for large number of delivery points, given how slow their optimization engine is. Same goes for every service mentioned above. Oh yeah, WorkWave bought Viamente for $4M, that's how much their tech was worth. Seems a little bit low for something that could attack the 45Billion market.
I have not stumbled upon a single one in the last 10 years that has any potential. All are leeching or would like to leech the big companies that can afford a 1 day waiting time for optimization. UPS seems to be all happy about its routing engine, but they too are being weird about it, if it's good then show it.
Someone who solves this for Uber will definitely get its first billion.
This would be a huge engineering effort since the problem is NP-hard :D