AMD Unveils Ryzen AI Halo, Pushing Personal AI Beyond the Cloud

AMD introduced Ryzen AI Halo , a high-memory platform with up to 192 GB of unified memory that can run AI models of up to 300 billion parameters locally, targeting more private and cost-efficient “personal AI” workloads.

AMD highlighted new devices and partnerships, including Lenovo’s ThinkCentre X, an HP ZBook-based laptop concept, Microsoft’s Project Zenith developer configuration, and SUSE tools designed to help move AI projects from local systems into enterprise and cloud environments.

The company also unveiled the Threadripper Halo Station , a workstation aimed at models exceeding 1 trillion parameters, featuring a 96-core processor, up to four MI350P accelerators, 2 TB of system memory, and up to 576 GB of HBM3E memory.

Advanced Micro Devices NASDAQ: AMD used its IFA 2026 presentation in Berlin to outline its vision for “personal AI,” introducing new high-memory Ryzen AI systems, a workstation concept and collaborations with Microsoft and SUSE intended to bring more AI workloads from cloud infrastructure to local devices.

Jack Huynh, AMD’s senior vice president and general manager of the Computing and Graphics Group, said AI is changing computing from a tool that waits for instructions into technology that can work alongside users. He argued that agentic AI systems—which can plan, use tools and continue work across tasks—will substantially increase demand for computing capacity.

Huynh cited AMD’s estimate that monthly AI token processing rose from about 0.7 quadrillion to 1.7 quadrillion in one year, and said estimates point to 120 quadrillion tokens being processed monthly by 2030. He said that growth is creating pressure on AI budgets and makes efficient deployment increasingly important. “The future of AI isn’t just about more compute, it’s about better compute,” Huynh said.

AMD framed local processing as a way to address both cost and privacy concerns. Huynh said personal AI requires three elements: local compute, privacy and control over data, and personal context that enables AI systems to tailor responses to a user’s goals and preferences.

The company highlighted its Ryzen AI platform and its Strix Halo architecture, which Huynh said has 120 GB of unified memory and can run AI models with as many as 200 billion parameters locally. AMD also introduced Ryzen AI Halo, which Huynh said increases unified memory to 192 GB and expands local model capacity to 300 billion parameters.

According to Huynh, Lenovo introduced the ThinkCentre X in Berlin powered by Ryzen AI Halo. He also unveiled a laptop code-named Sundance, described as a next-generation HP ZBook powered by Ryzen AI Halo and equipped with 190 GB of unified memory.

Huynh demonstrated what he described as the local creation of a 3D world from a single prompt and said the system could also run Microsoft Flight Simulator.

AMD said it plans to make its Ryzen AI Halo “box assist” system for developers available in Europe soon.

Pavan Davuluri, executive vice president of Windows and Devices at Microsoft, joined Huynh on stage to discuss the companies’ long-standing partnership. Davuluri said the Windows PC is evolving from a device that executes commands into a system that can understand workflows and help users complete tasks. “With your permission, agents can work across your apps, your services, your files to help you get things done,” Davuluri said, adding that trust, security and user control wi...