Nearly all wind and solar farms inject energy into the grid just like any other generator. In an open energy market like ERCOT (where this wind farm is located) there's a market where energy is traded.
If they inject "X" units of energy into the grid and pull out the same "X" units all is good. If not (the usual case since generation and demand need to equal 100% at all times) there are rules to govern the buying and selling of energy all generators ("market participants") follow.
Suffice it to say, grid management in an open energy market is complicated. If you really want to know more about the rules you can read more here: http://ercot.com/mktrules
The US has three grids: East, West, and Texas.
Folks ask why Texas has its own grid and we generally say "because Texas", but the real reason is just the way the grid grew in the early 20th century. Texas was isolated from the big industrial centers of the coasts and had abundant internal energy sources (coal and oil) so it just grew into its own grid and stayed that way.
Inside each grid the frequency is internally regulated. Each grid is out of sync with each other so power transfer between grids is limited. A tie between grids requires converting energy to DC and back into AC synchronized to the receiving grid. ERCOT has 5 of these ties and their capacity is a small percentage of the demand. Additionally, there's a few generators at the border of the grid than can operate on two different grids (one grid at a time). They choose where to operate depending on pricing.
Another problem to deal with is transmission capacity. If you have too much generation away from demand transmission gets congested. To deal with this we have nodal markets (http://www.ercot.com/content/wcm/training_courses/14/nodal10...).
Note that this wind farm is located right on the border between ERCOT and SPP (Eastern grid member). Depending on transmission capacity buildout I could imagine them being switchable.
If they were together there could be a tech angle - like how this was going to insulate their data centres from problems with the grid, or how they were going to shed load when the wind wasn't blowing, or something like that.
Otherwise it sounds like AWS and Amazon Wind Power could basically be two unrelated companies.
I'm wondering if wind power isn't an effort to decrease the price they pay on the open market for peak power, or to at least recoup some of those costs.
Anyway, if anyone knows the answer, I'd like to know as well. And to be clear: I have no opinion about how things should be in this case, I just want to know how they are.