Intel Optane could have eased AI memory crunch

Intel's Optane memory, discontinued in 2022, offered sub-10-microsecond latency and extreme write endurance ideal for AI inference KV caching. The technology's absence now compounds DRAM price pressure as GPU memory fills with model weights and chat sessions.