The Big Problem with GPUs and How To Solve It | Randolph and Kandan, MantisGrid AI

Podcast: The Neon Show · Siddhartha Ahluwalia

Opublikowano: 2026-09-15

We don't want the industry to fall apart, that's why we started this company. 6 months ago, everyone was talking about model deployment. Now everyone has moved on to agentic AI deployment and autonomous agent swarms. But all of this is running on an infrastructure that itself is not automated. And this underlying infrastructure has massive GPU cluster systems. If even one GPU fails the whole training workload fails. You can't have 100,000 employees managing 100,000 GPUs. But LLMs are not the right solution for managing the infrastructure. Kandan and Randolph have spent decades operating large-scale...