That's partly because LTE is so new, nobody wants to risk building an entire run of highest-of-high-end Android phones that might end up in a landfill... at least, not until they've had time to get a few real municipal areas lit up and see how LTE works in real American cities. Simulations and models can only get you so far. That's why they're only doing USB dongles first -- they're cheap. It would suck for Verizon to have to junk 100,000 USB LTE modems, but it would suck even MORE to have to junk the equivalent of 100,000 "Droid XL" phones.
That's the real reason why Sprint is moving ahead with WiMax, regardless of its long-term future. It's been around long enough that even before the first Evo came out of Taiwan, they were quite confident it would work as expected. Thanks to Clearwire, they even had a few medium-sized cities around the US where they could provision a pre-production phone and try it out. LTE is *nowhere* near that point yet in the US.
Yes, LTE exists in two cities in Europe... but as everyone knows, European cities are different from American cities in both big and small ways. OK, it's not quite the gulf some envision, but there are plenty of things that could still break fundamental assumptions (steel roof trusses vs wood? aluminum siding vs vinyl? brick vs waferboard stapled to 2x4s? home power wiring? you get the idea). Nobody at Verizon (or, more likely Motorola) is crazy enough to risk his career over a production run of expensive phones that work perfectly in Scandinavia as predicted, but end up not *quite* working right in the US for some reason that hasn't even occurred to anyone yet.
The nice thing about LTE and WiMax is that they can coexist within the same spectrum, the same way that 1XRTT can coexist with EV-DO (and UNLIKE the way GSM/EDGE needs a totally separate chunk of spectrum from UMTS). That's a very, very big deal, because it means there's never going to be a day of reckoning when Sprint HAS to shut down one to recover spectrum for the other, because modulation-wise, they can both exist side by side without interfering. If an Evo owner buys a new phone that does LTE, he's going to use the exact same spectrum around his local tower that he was using the day before -- it's just that at the tower, the LTE subsystem will start paying attention to his signal instead of the WiMax subsystem. Or more likely, the signal will be processed by the LTE algorithm instead of by the WiMax algorithm. As radio signals, LTE and WiMax are basically indistinguishable from one another.
The real WiMax shutdown (if/when it happens) will be the day the WiMax radio (or dual-mode radio) at a tower dies, and Sprint decides that there aren't enough WiMax users left using that particular tower to justify the cost of replacing the WiMax radio (or the cost of a more expensive dual-mode radio, vs one that only does LTE).
It's also important to keep in mind that functionally, Verizon LTE is almost identical to Sprint WiMax. Both are for data only, with voice calls continuing to be handled by CDMA2000 infrastructure. Verizon is playing up the "International" angle, but the truth is, it's going to be a really, really long time before anyone goes to Europe with a Verizon phone, powers it up after the plane lands in London, Paris, Amsterdam, or even Helsinki, gets a viable roaming connection, and makes a voice call with his rooted phone using Skype... and even longer before there's any hope of making that same voice call over LTE *officially*.
Put another way, two years from now, the difference between WiMax and LTE to most Sprint and Verizon customers will be academic. WiMax will be the reason why unlocked Sprint phones can't to "4G" on Verizon, and LTE will be the reason why an unlocked Verizon phone wouldn't be able to do "4G" on Sprint if Sprint allowed you to use it in the first place.
