Hey, Lucas from Lilac here.
Wrote a short piece on GPU waste. We've been talking with many enterprises and neoclouds for the past 9 months to get a good understanding of utilization rates (and they are shockingly low).
Most clusters sit with less than half of their GPUs utilized, translating to billions in lost compute time. Our goal is to help solve this problem at Lilac. Let me know what your thoughts are!
