AMD & Nutanix Partner for Agentic AI

This partnership goes beyond a simple press release, involving a significant financial commitment from AMD. AMD is making a strategic investment of $150 million in Nutanix stock and will also fund up to $100 million for joint engineering and go-to-market initiatives. This collaboration is a direct challenge to more vertically integrated AI solutions, emphasizing an open ecosystem. The collaboration will tightly integrate AMD's hardware with Nutanix's software. Specifically, Nutanix's Cloud and Kubernetes Platforms will be optimized to run on AMD EPYC CPUs and Instinct GPUs. The joint effort aims to create a new class of open AI infrastructure to support enterprise AI agents and multi-modal inference services. A key component of this partnership is AMD's ROCm open software stack, which will be integrated into Nutanix's full-stack AI solutions. ROCm is designed for GPU computing and serves as an alternative to NVIDIA's CUDA ecosystem, supporting frameworks like PyTorch and TensorFlow. This integration is crucial for developers and ML engineers looking for open and flexible software to build upon. The hardware centerpiece of this collaboration is the AMD Instinct MI300X accelerator. This GPU is designed for large-scale AI and HPC workloads, featuring 192GB of HBM3 memory and 5.3 TB/s of memory bandwidth. These specifications are particularly relevant for training and deploying large language models, a core task for ML engineers. Nutanix brings its "GPT-in-a-Box 2.0" platform to the partnership, which is designed to simplify enterprise AI adoption. This full-stack solution provides a unified interface for managing foundation models and creating API endpoints. It also includes integrations with Hugging Face's library of large language models, offering a broad range of open-source models for enterprise use. While this partnership with AMD is significant, Nutanix will continue to support Nvidia-based solutions. Nutanix's AI platform also integrates with Nvidia NIM microservices, providing customers with a choice of underlying hardware. This multi-vendor approach allows enterprises to avoid vendor lock-in for their AI infrastructure. The first jointly developed platform from AMD and Nutanix is expected to be available in late 2026. This timeline reflects the engineering effort required to optimize Nutanix's software for AMD's hardware and validate its performance for enterprise AI workloads.