The article is very misleading, but the real gist is that all of these AI dipshits figured out what FPGA was, and are now flooding the zone to use it. This tool has been an absolute gift to companies looking to move to FPGA for things like ML workflows, NL processing, and semantic flows, WITHOUT needing the dumb shit GPU pricing. Better results, cheaper bills.
AMD is two steps ahead of the game in these arenas versus Nvidia, and people just figured it out, so of course they are going to start charging for it. It uses resources on their end, so they need to bill for it to make it make sense. The same way every dumbass startup gives shit away to get you hooked is the way this works.
FPGA is going to usurp the reliance on CUDA in a massive way in the next few years, and they want to get paid for the upfront work they put in to make this possible.
Not shocked at all.