Comment by mechagodzilla

You all seem like one of the only companies targeting low-latency inference rather than focusing on throughput (and thus $/inference) - what do you see as your primary market?