Spectrum licenses are not new, and have a number of problems associated with them. The winner of an auction typically gains a monopoly in a few bands in a certain geographical region. This is by design, since there needs to be regulation over the devices using a particular portion of the spectrum to ensure that they don't interfere. Since most deployments use a client-server architecture, with consumer devices receiving signals from central towers, ensuring that only one entity will be building the towers makes this relatively easy to do.
The big downside to this is that it turns the auctions into a classical prisoner's dilemma. If everybody bids for just the amount of spectrum they need, at a low price, then they will all get a good price. If, on the other hand, one company can raise the bids enough to exclude one or more of their competitors entirely. If this works, then they have a total monopoly. In practice, it rarely does work, but this doesn't stop companies from trying. In the UK, we saw this with the 3G spectrum auctions, where the winning companies ended up spending so much on the licenses that they couldn't afford to spend much deploying the infrastructure.
Some spectrum auctions try to avoid this. Rather than offering the spectrum to the bidder with the most cash on the table, the regulator sells it for a fixed price to the company willing to invest the most on infrastructure, provide the cheapest service to consumers, etc. This is not the case with the current spectrum auction in the USA.