Like rival Nvidia, AMD spent much of CES 2026 focused on artificial intelligence (AI) for enterprise. Although the technology is usually associated with massive server systems, the "Red Team" did not forget everyday users, making several major AI announcements for personal devices. The good news for AMD graphics card owners is that they will get access to one of the company's newest and most useful tools as soon as next week, promising to make this cutting-edge technology considerably easier to try.
Adrenalin Software Update and a New AI Bundle
AMD has officially confirmed that the next version of its Adrenalin driver software, due on January 21, will introduce new features that streamline installing local generative AI development tools. The announcement follows the multi-front AI push the company previewed at CES last week. It is a well-received move that gives mainstream users easier access to complex technology without needing deep programming knowledge.
Coming January 21st: AMD Software: Adrenalin Edition™ AI Bundle
AI Made Simple.
AI Bundle is a streamlined installer that equips your AMD-powered system with the essential tools for AI development and creative workflows.
- AMD Radeon (@AMDRadeon) January 14, 2026
AMD has not yet shared specifics on what is packed into the "AI Bundle," but the main draw is an optional unified installer. It gathers the software needed to run local AI models in one place, so users can reach image generators, local large language models (LLMs), and other AI development software far more easily. Support for PyTorch on Windows has also been expanded, which should help developers and enthusiasts.
The AI bundle appears to be designed to get the most out of the machine learning capabilities of the integrated graphics processors (iGPUs) and neural processing units (NPUs) in AMD's recent Ryzen AI chips. Alongside it, AMD announced a major update to its ROCm software, the foundation that supports AI workloads on Ryzen AI processors. ROCm 7.2 now supports installation through ComfyUI, improves compatibility with both Windows and Linux, and simplifies access to new PyTorch builds on Windows.
New Ryzen AI Hardware and What Comes Next
To support the new software ecosystem, AMD expanded its Ryzen AI lineup at CES with the Ryzen AI 400 series. Laptops built on these new APUs, a refresh of the Zen 5 CPU architecture and the RDNA 3.5 iGPU architecture, will reach the market from many manufacturers this quarter. Desktop versions are expected to arrive in Q2 2026. The standout in the seven-chip lineup is the flagship Ryzen AI 9 HX 475, which packs impressive power: 12 cores, 24 threads, 16 GPU compute units, boost clocks up to 5.2 GHz, 36 MB of cache, and up to 60 TOPs of AI compute.
AMD did not stop there. It also introduced Ryzen AI Halo, a workstation mini PC built on the Ryzen AI Max+ processor. The device is designed to run AI models with up to 200 billion parameters without relying on cloud servers. With up to 128 GB of unified memory, RDNA 3.5 GPU performance of up to 60 teraflops, and support for either Windows or Linux, this hardware platform has much in common with Nvidia's DGX Spark system.
Despite growing negative sentiment around a generative AI bubble, AMD CEO Lisa Su remains optimistic about the technology. At CES, she dismissed speculation that the industry is forming a dangerous bubble. Instead, she predicted that consumers will adopt AI widely and quickly, affirming her confidence in the company's long-term strategy.
