Yaaay! The world sucks slightly less than expected! :D
Of course we're still going to have a problem in the long run if the AR9xxx chipsets get fazed out and we're stuck with only 802.11ac that require binary blobs.
In the long term (a few years down the road) we might want to start looking at using at SDR-based solution. I talked to the Lime SDR folks at HOPE and they were sharing table with a guy from Fairwaves who was using their radio frontend chip to make mini pci-e SDR boards. The cost for those boards will be about $200 (but that was including a GPS chip and in low quantities). These SDRs have a 62.5 MHz bandwidth, are capable of 256 QAM and support 2x2 MIMO. They're not currently able to reach the 5 GHz spectrum but they're launching a board that will increase their max frequency to 12 GHz soon. These SDRs, coupled with cheap high-performance single board computers, have the potential to become alternatives to the TP-Link and Ubiquiti gear we're using now. We obviously would need the price to come down for this to be a viable replacement for the < $100 gear we're using now, but perhaps it makes sense to start working on this now?
I haven't done the math but as compute power per $ increases shouldn't we be able to get rid of the FPGA completely and simply create a single board computer with a DAC/ADC and a programmable radio frontend where all processing happens on the CPU? When will we see an 802.11ac stack implemented in pure JS? :p