LLM AI is exactly an architecture that makes use of multitudes of cores efficiently. Not efficiently in terms of power usage, but efficiently in terms of spreading the workload out over many, many cores. That is why they're building these huge, expensive AI compute farms.