First in what? What's the prize?
If the prize is just throwing money down a hole on hardware that will be obsolete the moment it's deployed, then let the other guys win.
If the prize is getting accurate weather forecasts at the lowest cost, then maybe we'd be better off contracting from those countries, using spare cycles from other government agencies that are wasteful and counterproductive (cough, NSA, cough), or writing better software to run on the other guy's hardware and licensing it to them.
The second approach won't allow us to thump our chest and say, "computer that required 10 new hydro-electric dams! BOOYAH! fastest in the WORLD!!!" but it'll accomplish a meaningful goal.