NVDA/long
NVIDIA's B300/B200 GPUs dominate AMD MI355X on LLM inference throughput, cementing AI hardware leadership
NVIDIA mogs AMD. Also great to see B300 faster than B200.
Peak+1.6%
SemiAnalysis benchmark shows NVIDIA B300 at ~5,600 tok/s/gpu vs AMD MI355X at ~3,100 — a near-2x gap on Kimi K2.5 LLM inference. Two trades: NVDA long (hardware dominance confirmed) and AMD short (MI355X uncompetitive for AI workloads).
originalNVIDIA mogs AMD. Also great to see B300 faster than B200.
NVIDIA mogs AMD.