Helion is an open rack-scale design built on an Open Compute Project specification. A CPU-only rack using 6th Gen EPYC processors supports up to 32,768 cores, aimed at hyperscale customers, while a combined CPU and GPU version targets enterprises. Customers can tune core density, performance and power for agent workloads. The blueprint includes an integrated token router that directs an agent's model calls across frontier and open-source models. Configurations are being developed with ecosystem partners, and AMD plans to contribute the specification to the Open Compute Project for community development. The launch came at the OCP Global Summit alongside Nvidia's Vera Rubin rack contribution and Intel's inference GPU, as vendors position open rack designs for agent workloads. For infrastructure teams running agent platforms, AMD is pitching CPU capacity as a separate line item from GPU capacity. Rack designs that put routing in hardware also hint at where model-routing logic could end up, though availability and pricing are still unannounced.