AMD has unveiled its next-generation AI infrastructure portfolio at Advancing AI 2026, introducing the AMD Helios rackscale solution alongside new processors, GPUs, and embedded AI platforms designed for the growing era of agentic AI.
The announcements include the launch of 6th Gen AMD EPYC processors, AMD Instinct MI400 Series GPUs, AMD Helios AI rackscale solutions, AMD Ryzen AI Embedded X100 processors, and the AMD Kria AI SOM and Robotics Developer Platform.
AMD targets the next wave of AI computing
As AI workloads shift beyond model training toward inference, autonomous agents, and physical AI, AMD said demand for computing infrastructure continues to accelerate.
“The next phase of AI will span frontier models, agents and physical AI, creating new opportunities to bring intelligence everywhere,” said Dr. Lisa Su, Chair and CEO of AMD.
“Realizing that potential will take the entire industry working together. AMD is partnering across the ecosystem to deliver leadership compute and open platforms that give customers the performance, flexibility and choice to scale AI from the data center to the edge.”
AMD also projects its total addressable market for AI silicon—including data center, PC, edge, and embedded processors—to reach approximately $2 trillion by 2030.
Helios leads AMD’s rack-scale AI portfolio
At the center of Advanced Micro Devices’s latest announcements is AMD Helios, the company’s first rack-scale AI solution built for frontier AI workloads.
Each Helios rack combines 72 AMD Instinct MI455X GPUs, 18 6th Gen AMD EPYC “Venice” CPUs, AMD Pensando networking, and the AMD ROCm open software platform into a single AI infrastructure.
According to AMD, Helios delivers up to 30% more inference tokens per dollar than competing rack-scale AI systems.
The company added that Helios is already in production and will be deployed at gigawatt scale by leading AI companies.
Major AI companies adopt AMD Helios
Several leading AI developers and cloud providers have committed to deploying Advanced Micro Devices’s latest AI infrastructure, including OpenAI, Anthropic, Meta Platforms, Microsoft, Oracle, HUMAIN, TensorWave, Vultr, and Cirrascale Cloud Services.
AMD also said the platform will be available through OEM partners, including Hewlett Packard Enterprise, Lenovo, Supermicro, Bull, Sanmina, and Wiwynn.
Expanding AI partnerships
AMD also highlighted new collaborations announced during Advancing AI 2026.
Anthropic will deploy up to 2 gigawatts of AMD Instinct MI455X GPUs through Helios while collaborating with AMD to optimize ROCm software using Claude AI.
OpenAI is working with AMD to optimize GPT-class workloads through the Triton framework and ROCm software, with Helios deployments expected to begin in the fourth quarter of 2026 before expanding throughout 2027.
Meanwhile, Meta is validating AMD’s latest EPYC processors and Helios racks for future large-scale AI deployments, while Cerebras is integrating its inference platform with AMD Helios to improve low-latency AI inference.
