I feel like we're still stumbling about a bit and don't know all the answers, which is fine. But NVIDIA frames AI agents as the next computing paradigm, but most of what's described here still looks like orchestration + retrieval + tool use on top of LLMs.
What actually has to change at the systems level (data layout, memory, scheduling, storage, networking, etc.) for agents to become a first-class workload rather than just another application pattern on existing infrastructure?
What actually has to change at the systems level (data layout, memory, scheduling, storage, networking, etc.) for agents to become a first-class workload rather than just another application pattern on existing infrastructure?