At PyTorch Conference 2026, held October 20-21 in San Jose, AMD showcased how it is advancing the open AI ecosystem across training and inference through continued contributions to the ROCm software, performance optimization, and emerging AI workloads. AMD demonstrated broad PyTorch support across cloud, data center, workstation, and client platforms, including native PyTorch integration across the ROCm software stack, TorchInductor with the MLIR native FlyDSL backend, reinforcement learning post training and rollout, and PyTorch inference on AMD XDNA NPU architecture. Performance topics included PerfModel for validation driven Triton kernel optimization, Grouped GEMM on the ROCm software, TorchAO kernels and MXFP8, FlexAttention and KV cache optimization, and inference performance analysis using TraceLens, along with agentic workload optimization with Hyperloom and efficient masked diffusion language model inference. AMD highlighted scaling progress including MXFP8 pretraining across more than 1,000 AMD Instinct MI355X GPUs, distributed training with TorchTitan, rocSHMEM symmetric memory, and RCCL GPU initiated networking, plus MoRI and vLLM for expert parallel and disaggregated Mixture of Experts serving, large scale reinforcement learning training and rollout, and resilient multimodal inference routing. AMD engineers contributed through breakout sessions, lightning talks, and posters, with featured sessions from Mahdi Ghodsi, Jiahui Cao, Prachi Gupta, She