{"type":"post","stable_id":"post:wafer-kimi-k3-mi355x-memory-moat","slug":"wafer-kimi-k3-mi355x-memory-moat","title":"Wafer's Kimi K3 Run Makes Prefill Memory A Hardware Economics Story","description":"Wafer describes running Kimi K3 on AMD MI355X and frames the result around prefill optimizations, throughput, performance per dollar, and memory as a potential inference moat.","retrieval_nugget":"The processor marked this reserve only because of broad Kimi prior coverage; the recovered angle is distinct: memory bandwidth, prefill economics, and AMD inference performance per dollar.","published_at":"2026-07-31","updated_at":"2026-08-05","record_date":"2026-07-31","date_kind":"published_at","status":"published","topics":["inference","hardware","open-models","memory-bandwidth","economics"],"entities":["Wafer","Kimi K3","AMD MI355X"],"source_urls":["https://www.wafer.ai/blog/kimi-k3-mi355x"],"source_title":"Is memory the moat?","source_type":"primary","origin":{"batch_id":"ada1e768-4e4e-42a3-a5fc-1ca9a880cb1b","batch_index":4,"channel":"chatgpt-batch","restored_from_reserve":true,"restored_from_supporting":false},"schema_version":"newruntime-agent-readable-v0.2","visuals":[{"role":"hero","src":"/images/drip/wafer-kimi-k3-mi355x-memory-moat/wafer-kimi-k3-mi355x-memory-moat.webp","alt":"A whiteboard infrastructure diagram showing Kimi K3 serving moving through prefill, memory bandwidth, AMD MI355X nodes, throughput, and cost-per-token economics.","caption":"New Runtime synthesis."}],"routes":{"html":"https://newruntime.com/posts/wafer-kimi-k3-mi355x-memory-moat/","markdown":"https://newruntime.com/posts/wafer-kimi-k3-mi355x-memory-moat.md","json":"https://newruntime.com/posts/wafer-kimi-k3-mi355x-memory-moat.json"}}