AMD and Intel Partner to Unify x86 AI Development for PyTorch

AMD and Intel have announced a rare collaboration to integrate their respective CPU optimization efforts into the open-source oneDNN library, creating a unified software path for PyTorch on x86 processors. Additionally, AMD is joining the UXL Foundation as a steering committee member and bringing its ZenDNN optimizations into the shared ecosystem. This partnership addresses software fragmentation within the x86 ecosystem, making it significantly easier for developers, cloud providers, and enterprises to deploy AI workloads across both AMD and Intel hardware. By uniting their forces, traditional competitors are presenting a stronger, more cohesive front against dominant GPU and accelerator alternatives like NVIDIA and ARM. The two companies are also pushing the AI Compute Extensions (ACE), a next-generation x86 instruction set architecture designed to accelerate core matrix multiplication operations for machine learning workloads. Moving forward, these hardware-level acceleration capabilities are planned to be integrated into PyTorch via updates to oneDNN.