Uber is easy to replicate
Write an app, offer a bigger share of the revenue to the drivers (or wait for Uber to inevitably start screwing over their drivers with various bad policies or taking a bigger cut) and take a share of their market. Or operate somewhere they currently don't, or somewhere they got kicked out of for thinking they're above local laws.
This is not like building a competing search engine, where you need specialized algorithms, huge server farms, and have advertisers uninterested in bothering with you until you reach critical mass so you better have money to operate and grow for a few years with little or no revenue. It isn't even like building a competing Android phone, where you need to have supplier relationships with dozens of suppliers to get all the parts you need, engineers to do at least basic physical design, a contract manufacturer to make it, programmers to at least paste your logo onto Android, etc.
If the market is anywhere near big enough to justify Uber being worth $40B, there will be a flood of entrants. They don't need to take on the world, I might start one in my town, you start one in yours, and we all provide the thousand cuts that bleed Uber's market valuation down to commodity levels.
One may argue that Uber has some network effects due to its size, but as competing services come, someone will write a meta app that checks all the services available where you are and lets you choose between the cheapest price or quickest arrival. Even the tiniest service can compete once that's in place, and Uber can do nothing to fight it.