“CoreWeave has developed a proprietary caching solution called "Lotta Cache" designed to maximize data throughput to GPUs for AI workloads.”