The costs should be lower by now as a matter of fact due to an unfortunate side effect in the China semiconductor industry. There is starting to be too many too many fabs and ventures in China, and even the GaAs is going into overproduction.
Even with the prices down, nonetheless its futile to pursue this course because GaAs MMIC is becoming obsolete thanks to GaN MMIC. Again, there is now a gold rush into this material around the world even in China becomes it promises a whole new era of telecommunications where you get less stations but each covering much wider ranges with much more power. Currently no use in pursuing GaAs AESA designs when second generation GaN AESA designs promise to be much more powerful. To put it this way, a GaAs MMIC generates up to 4 watts, so a thousand element array will generate 4000 watts. But a GaN element can generate up to 50 watts (Japanese research), so let's say conservatively, you got 20 watts per element, an array with a thousand elements can generate up to 20,000 watts. When you get 20 kilowatts of radar into a tight beam, that's so brute force you can't even stealth it. At 50 watts per element, the entire array can become a death ray machine.
I shudder at the thought of MMICs this powerful in your celphone being held up close to your brain.