Intel D270 positions memory-rich inference for agentic AI

D270: up to 480GB LPDDR5X, 350W air-cooled, 32 Xe3P cores, 256 XMX engines, PCIe 5.0, vLLM/PyTorch ecosystem. Positioned for long-context prefill, high-concurrency agent fleets, RAG, and memory-bound inference without HBM.