>>725172107
cost will come down, they haven't even optimized most of it yet and they're in the midst of making their own AI chips and speed ups and power effiency from software emulation to purpose fit sillicon for operations in hardware is a lot but you first need to know requirements