$NBIS CBO and co-founder Roman Chernin basically just explained why inference infrastructure may become one of the most important layers of the AI stack over the next several years. Running open-source and alternative AI models at scale is not simply downloading weights and turning GPUs on. The real challenge is extracting reliability, efficiency, and economics from massive distributed compute systems powered by infrastructure tied to $NVDA, networking from $AVGO and $MRVL and hyperscaler compute from $AMZN, $MSFT, and $GOOGL. That is exactly why Nebius continues positioning Token Factory as a full-stack inference platform optimized for agentic workloads, open models, and large-scale AI deployment alongside the broader shift toward inference-driven AI infrastructure.