AMD Helios Rack Scale AI Platform Accelerates Towards 2026 Mass Shipments with Support from Taiwanese Manufacturing Partners and Major Hyperscale Cloud Providers
AMD has sped up the transition to its new generation Helios rack scale AI platform, as it aims to initiate mass shipments in the second half of 2026. Based on CTEE report, its major Taiwanese manufacturing partners are preparing their production capacities to meet the increasing demand and support the substantial growth of the local hardware supply chain. This initiative would significantly contribute to diversifying the high performance computing (HPC) infrastructure of major cloud service providers.
Three major Taiwanese electronics manufacturers will synchronise their production volumes according to AMD's new hardware rollout. Wistron has obtained massive orders for front end motherboard configurations of the MI350 platform and plans to use its newly acquired 2nd plant in Chupei to handle manufacturing. Wiwynn will gradually ship out its Helios rack systems starting in Q3 of 2026, whilst Inventec is also ramping up production and aiming for large scale shipments in Q4 of 2026 to cater to major cloud companies' requirements.
The fast pace of the supply chain mobilisation is also in line with the TrendForce findings that the top 5 North American hyperscalers are increasing their adoption of complete rack scale hardware configurations, moving away from a single provider. Nvidia systems are still heavily utilized, but the industry has been actively integrating both AMD MI300 and MI350 silicon families, allowing cloud providers to have a more resilient and balanced computing infrastructure.
Microsoft is currently leading the adoption of AMD's hardware architecture and considers it its second largest accelerator option. It has deployed over 100 racks of legacy MI200 hardware and has been consistently integrating MI300 and MI350 models. Microsoft is one of the main beneficiaries of the 2026 second half delivery of Helios rack scale systems, has completed its plan to deploy MI450 rack configurations by end of 2026, and intends to boost its network capabilities by adopting the upcoming MI500 series in 2027.
Meta is also integrating AMD hardware into its ongoing Meta Compute initiatives and is currently deploying a massive number of MI300 and MI350 processors in its facilities as a supplement to Nvidia systems and its custom ACICs. Elon Musk's xAI has committed to utilizing the Helios rack systems starting second half of 2026 and will use them to speed up deep learning development. Neocloud is expected to adopt AMD systems alongside its existing Nvidia clusters as part of its alliance strategy to provide its client base with a diversified offering of computing services.
