Based on engagements I have priced, a full kernel-bypass stack with FPGA integration has run 1Mto1M to 2M all-in —…