my prediction is that nvidia will release inference-optimized hardware in the future release cycles. these chips might have a hard time to trickle down to lower-tier, but money printer go brrr.