Key Takeaways from AMD at CES 2026
AMD highlighted three overarching trends that will dominate gaming and AI in 2026: the rise of on‑device AI workloads, increasing memory costs, and a shift toward modular component upgrades.
- On‑device AI is moving from the cloud to notebooks, consoles, and high‑end workstations.
- DRAM pricing volatility is affecting CPUs, GPUs, and system builds.
- Consumers are favoring targeted upgrades (CPU, GPU, motherboard) over full‑system replacements.
Radeon AI Pro R9700 and ROCm 7.2
The Radeon AI Pro R9700, a 32‑GB Navi 4‑based GPU, comes in four configurations: dual‑slot active‑cooled workstation cards, dual‑slot passive server cards, and single‑slot passive server cards. It is designed to pair with the upcoming ROCm 7.2 driver suite, which adds out‑of‑the‑box support for:
- Ryzen AI and AI Pro 400 series
- Ryzen AI Max accelerators
- Radeon 9000 and 9070 series GPUs
- Both Windows and Linux ecosystems
This integration simplifies AI development across a wide range of platforms.
Adrenaline AI Software Bundle
AMD’s Adrenaline stack now includes an “AI software bundle” that consolidates the tools and frameworks needed to run AI workloads on any AMD GPU, whether it’s an integrated notebook chip, a Strix Halo, or a discrete graphics card. The goal is an “easy button” that lowers the entry barrier for creators and developers.
Local AI Model Migration
One of the most exciting trends discussed was the migration of large AI models from the cloud to local devices. This enables:
- Real‑time inference on notebooks and thin clients
- Secure processing of proprietary data on‑premise
- Scalable workloads that can span from a single R9700 to multi‑GPU rigs
AMD expects this shift to accelerate throughout 2026, driving demand for both consumer‑grade and enterprise‑grade GPUs.
Memory Pricing and Channel Configurations
AMD acknowledges that DRAM costs are rising across the board. While dual‑channel memory still offers performance advantages, many popular titles show limited sensitivity to single‑ versus dual‑channel setups. As a result, AMD advises builders to evaluate memory configurations on a per‑title basis to maximize value.
Component‑by‑Component Upgrades vs Full System Refresh
Consumers are increasingly opting for incremental upgrades—swapping out CPUs, GPUs, or motherboards—to extend the life of existing rigs. This trend is fueled by:
- Rising component costs
- Improved compatibility across AMD’s 600‑series and 700‑series platforms
- The desire to avoid waste from full‑system replacements
AMD’s roadmap aims to support this modular approach with stable socket and BIOS updates.
Pre‑built PCs vs DIY in a Shifting Market
Higher memory prices and supply chain volatility are nudging some users toward pre‑built systems, which can lock in component pricing and offer optimized configurations. However, AMD stresses that the NPU (Neural Processing Unit) and AI‑focused GPUs remain niche tools that benefit power users who need persistent, low‑power AI workloads.
The Role of NPUs in Future Gaming Systems
NPUs are positioned as highly efficient engines for always‑on AI tasks, delivering high TOPS per watt while preserving battery life in laptops. Their programmability is more limited than GPUs, so they complement rather than replace traditional graphics processors.
Looking Ahead
AMD’s CES 2026 outlook points to a future where AI, memory economics, and flexible upgrade paths intersect. With Radeon AI Pro hardware, ROCm 7.2, and the Adrenaline AI bundle, AMD aims to give developers the tools to harness on‑device AI while keeping performance‑per‑dollar competitive.