AMD has officially unveiled the Threadripper Halo Station, a new ultra-high-performance workstation that the company describes as a "personal super computer." The device was introduced at the Internationale Funkausstellung tech event in Germany by Jack Huynh, AMD’s Senior Vice President.
Designed to address the growing global demand for AI processing, the workstation is engineered to deliver data-center-level performance and capacity in a desktop form factor. According to Huynh, the machine is the "most powerful workstation in the world," boasting the capability to run AI models featuring more than a trillion parameters. During a demonstration at the event, the system showcased its potential by generating a full 3D world and a flight simulator from a single prompt.
The technical specifications of the Threadripper Halo Station include up to 96 cores of Threadripper Pro, two MI350P data center accelerators, and a liquid cooling system. The device supports up to two terabytes of system memory and up to 576 gigabytes of high bandwidth memory (HBM3E). By utilizing standardized components to achieve this level of performance, AMD aims to lower the barrier to entry for AI and personal computing sectors.
AMD plans to release the supercomputer early next year. While the company has not yet disclosed pricing, it is expected that the cost could exceed €100,000, positioning the product for corporate and institutional budgets rather than the traditional consumer market. The platform is intended to allow users to deploy large-language models and agentic AI systems locally, reducing reliance on cloud computing services.
The workstation serves as a direct competitor to the Nvidia DGX Station. While both systems are capable of running large AI models in a desktop environment, they utilize different strategies. AMD’s offering distinguishes itself with significantly higher memory capacity and bandwidth—offering more than three times the memory of its Nvidia counterpart—which facilitates the local execution of trillion-parameter AI models. Conversely, Nvidia’s system is already available at a retail price of approximately €100,000 and offers more seamless integration when transitioning code from local workstations to the cloud. Industry analysts suggest that AMD’s announcement could serve as a significant catalyst for reshaping the competitive landscape of the local AI computing market.
Therefore, AMD's approach of delivering data-centre-class performance using standardised components to lower the barrier to entry could be a major turning point in the expansion of the AI and personal computing sectors.
