Actually both product websites are **highly misleading**.
In both cases they appear to be using standard Atheros chipset cards running in Turbo modes, 40MHz channels giving raw bit rates of 108Mbps using OFDM QAM64 mode.
I don't understand the "QAM256" claim on airtegrity's website, that does not correspond either to the chipset (QAM64 max) or the quoted performance of 2x108Mbps from 2 cards aggregated.
which shows non-turbo modes
for the current Atheros5414 as used in R52 etc.
Simply aggregating channels from adjacent cards is not new, MT themselves have the record with 4 turbo channels in parallel, results are on the wiki. And yes, with the right hardware you can do that with a single high speed x86 platform running RouterOS.
Such platforms are available, my co. demonstrated 200Mbps P2MP net throughput on own-brand 1GHz passively-cooled platforms running RouterOS.
However, there are practical issues in finding enough clear RF channels and potentially multiple dual-polar antennas required to implement bonded P2P links.
Compared to current-generation Atheros chipsets, stepping into the world of licensed microwave it is possible to run higher modulation schemes such as 128QAM to give 311Mbps throughput and so-called XPIC (cross-polar interference cancellation) implementing 128QAM simultaneously on H and V polarisations for 622Mbps FDX throughput etc.
However, that only works in low-noise no-interference environments, **not** in unlicensed bands (except perhaps in remote rural areas).
That being said, I gather MT is working on 802.11n chipset implementations which do have higher spectral density/throughput than current Atheros cards, and we all look forward to seeing that.