AI infrastructure
AMD officially launches Helios rack-scale AI system
AMD launched Helios at Advancing AI 2026, combining 72 Instinct MI455X accelerators with EPYC Venice processors, Pensando networking and ROCm software in an open, liquid-cooled rack architecture.
AS1 News
AMD
NVDAAMD officially launched Helios, its rack-scale artificial intelligence system, at Advancing AI 2026 on July 23, 2026.
The system integrates 72 AMD Instinct MI455X accelerators, EPYC Venice processors, Pensando networking and the ROCm software stack. These components are packaged in an open, liquid-cooled rack architecture.
Helios matters because it represents AMD’s principal system-level challenge to Nvidia’s rack-scale AI infrastructure. Rather than offering only individual processors or accelerators, AMD is presenting an integrated platform for hyperscalers building large AI training and inference clusters. The launch also gives those operators another potential infrastructure supplier.
The official launch and the core system configuration are confirmed by the supplied event package. However, the package does not establish pricing, shipment timing, customer deployments or comparative performance. Those details therefore remain uncertain.
Helios strengthens AMD’s position in AI infrastructure by extending its offering to a complete rack-scale system and providing hyperscalers with an alternative supplier for large training and inference deployments.