AMD has launched its "AI full-scale expansion" strategy at CES 2026, unveiling a massive lineup of new products spanning consumer PCs, data centers, and embedded systems.

AMD Dr. Lisa Su Lisa Su, CEO of Intel, unveiled new products across client, graphics, embedded, and commercial segments in her opening keynote address at CES 2026 on the 5th.

This announcement is seen as AMD's declaration of intent to lead the next-generation computing paradigm through a platform strategy that integrates CPU, GPU, and NPU, amidst the intensifying competition in the AI PC market.

■ 'Fastest Gaming CPU' 9850X3D Unveiled

AMD has once again taken aim at the high-performance market with the unveiling of the Ryzen 7 9850X3D in the gaming segment.

This product is based on the 2nd generation 3D V-Cache technology and 'Zen 5' architecture, and provides a boost clock of up to 5.6GHz and 104MB cache, emphasizing that it is the 'highest performing gaming CPU available'.

Additionally, it showcased a number of next-generation rendering technologies, including frame generation and ray tracing optimization, through FSR 'Redstone', which includes machine learning-based upscaling technology.

■ Ryzen AI 400 series with 60 TOPS NPU

The field that received the most attention was, by far, AI PC.

AMD delivers NPUs with up to 60 TOPS of performance with the Ryzen AI 400 series, strengthening the foundation for generative AI and on-device LLM execution.

In particular, the top model, Ryzen AI 9 HX 475, integrates a 12-core, 24-thread configuration and Radeon 800M graphics to achieve both high performance and low power efficiency.

Also, the Ryzen AI Max+ series, which is aimed at ultra-thin notebooks and mini PCs, and supports up to 128GB of integrated memory, which is 200B processors.Ryzen AI Halo, a developer platform capable of running locally up to the metric model, was also unveiled.

■ PRO 400 Series Enhanced Security and Manageability

The Ryzen AI PRO 400 series, aimed at the enterprise market, enhances security and management features based on AMD PRO technology and delivers AI performance of up to 60 TOPS.

AMD has announced its strategy to secure competitiveness in the corporate AI PC market through this.

■ ROCm 7.2·AI Bundle Ecosystem Expansion

AMD is accelerating the expansion of its software ecosystem as well as its hardware.

ROCm 7.2 improves AI performance by up to 5x over the past year and expands support across Windows and Linux.

We also improved developer accessibility by providing PyTorch builds and ComfyUI integrated downloads.

Here, we are introducing the Adrenalin Edition AI Bundle, which allows you to configure a local AI environment with one click, and we emphasize that this has significantly reduced the complexity of building an AI development environment.

■ MI400 Series and Next-Generation MI500 Roadmap Revealed

In the data center space, the Instinct MI400 series was key.


In particular, the Helios system delivers up to 3 AI exaflops per rack, laying the groundwork for “yottascale computing” for training ultra-large models.

In addition, it targets AI infrastructure needs at the corporate and national level with various configurations such as the MI440X based on 8 GPUs and the MI430X for sovereign AI and HPC.

AMD has garnered attention from the industry by announcing that the MI500 series, scheduled for release in 2027, aims to improve AI performance by more than 1,000 times compared to the MI300.

■ Targeting the 'Physical AI' era

AMD also announced the Ryzen AI Embedded series in the embedded field, targeting the 'physical AI' market, including automotive, medical, industrial automation, and robotics.

In particular, it is designed to simultaneously process real-time graphics and AI inference by integrating the RDNA 3.5 GPU and XDNA 2 NPU into a single chip.

Along with this, Italian IIT's robotics startup 'Generative Bionics' attracted great attention at the scene by unveiling the next-generation humanoid robot GENE1.0 based on AMD.

■ A full-stack strategy encompassing all areas in the AI era

At CES 2026, AMD clearly presented its AI-centric, full-stack strategy, spanning from consumer PCs to data centers and embedded systems.

In particular, it has an architecture that integrates CPU, GPU, and NPU, an enhanced software ecosystem, and a system that encompasses developers, companies, and even countries.The product portfolio is expected to significantly increase AMD's presence in the 2026 market, when AI competition intensifies.