- AMD Ryzen AI Halo delivers 16 CPU cores and 32 threads for AI workloads
- Integrates an NPU alongside Radeon GPU cores for AI tasks
- Ryzen AI Halo offers full ROCm support across Windows and Linux platforms
AMD has confirmed it will launch its first PC in 2026, named Ryzen AI Halo, a system built around its Ryzen AI Max+ 395 processor with up to 16 Zen 5 CPU cores and 32 threads.
The company describes the device as a reference platform for local AI development, designed to run Windows and Linux with full support for AMD ROCm and day one AI model compatibility.
With up to 128GB of unified memory and an integrated NPU, the device can handle large generative AI models, although actual performance will depend on workload intensity.
Processing architecture and core specifications
The Ryzen AI Halo processor uses 4nm process technology with boost clocks up to 5.1GHz.
Cache specifications include 16MB of L2 and 64MB of L3, while configurable TDP ranges from 45 to 120W depending on system tuning.
The processor pairs with Radeon 8060S graphics featuring 40 cores and a maximum frequency of 2900MHz.
The GPU supports multiple display resolutions, including up to 7680×4320 at 60Hz, and includes DisplayPort 2.1 with adaptive sync, an HDMI 2.1 interface, and HDR metadata.
With this configuration, the device supports up to four displays simultaneously.
The system supports up to 128GB of LPDDR5x 8000 memory across a 256-bit interface, along with NVMe boot and RAID storage options supporting RAID0 and RAID1.
Connectivity options include two USB4 ports at 40Gbps, three USB 3.2 Gen 2 ports, three USB 2.0 ports, sixteen usable PCIe 4.0 lanes, and wireless features.
The system supports advanced AI capabilities rated at up to 126 TOPS overall, including 50 TOPS from the integrated NPU.
Security features include AMD Enhanced Virus Protection with the NX bit.
Ryzen AI Halo is designed to compete directly with Nvidia’s DGX Spark mini PC, a compact AI system targeting similar local AI workloads.
Both systems offer high performance computing in small form factors and support extensive memory capacity for large AI models.
DGX Spark relies on Nvidia’s HGX architecture and CUDA optimized frameworks, while Ryzen AI Halo integrates an NPU and Radeon GPU cores with full ROCm support.
This approach allows developers to run AI models locally without relying on cloud managed infrastructure.
Halo’s 128GB LPDDR5x memory capacity and 126 TOPS overall AI compute rating aim to match or exceed DGX Spark mini PC performance, offering an alternative for institutions seeking local AI experimentation.
However, Nvidia’s DGX Spark mini PC already has established benchmarks and a mature ecosystem, while Ryzen AI Halo will need to demonstrate comparable or superior performance in real world AI workflows.
TechRadar will be extensively covering this year’s CES, and will bring you all of the big announcements as they happen. Head over to our CES 2026 news page for the latest stories and our hands-on verdicts on everything from wireless TVs and foldable displays to new phones, laptops, smart home gadgets, and the latest in AI. You can also ask us a question about the show in our CES 2026 live Q&A and we’ll do our best to answer it.
And don’t forget to follow us on TikTok and WhatsApp for the latest from the CES show floor!
https://cdn.mos.cms.futurecdn.net/GH6EA6aCeGqAAnJvpByzAE-1920-80.png
Source link
