Advanced Micro Devices (AMD) unveiled several of its latest artificial intelligence processors at CES in Las Vegas on Monday, as CEO Lisa Su demonstrated a portfolio encompassing both data center and enterprise-focused chips.
Key among the announcements were the advanced MI455 AI accelerators, which form part of server rack solutions targeted at major clients such as OpenAI, the maker of ChatGPT.
Su also introduced the MI440X—an enterprise variant of the MI400 series designed for organizations with existing IT infrastructure not tailored specifically for AI clusters. The MI440X traces its origins to a previous chip slated for use in an upcoming U.S. supercomputer, according to AMD.
While AMD continues to position itself as a leading competitor to Nvidia, its ability to match Nvidia’s scale remains a challenge. Analysts note that despite a confidence-boosting deal struck with OpenAI in October—which is expected to add billions of dollars to AMD’s annual revenue and see the MI400 series rolled out this year—Nvidia’s AI chips still dominate the rapidly expanding market.
Greg Brockman, President of OpenAI, joined Su during the presentation, emphasizing the crucial role ongoing chip innovation will play in meeting the soaring computational demands of companies like OpenAI.
Looking further ahead, Su previewed the forthcoming MI500 series, claiming a 1,000-fold performance improvement over earlier versions. AMD expects to launch the new MI500 chips in 2027, aiming to capture future enterprise and hyperscale demand.
Meanwhile, AMD used the CES showcase to launch its Ryzen AI 400 Series for AI-powered PCs, as well as Ryzen AI Max+ for enhanced local inference and gaming workloads. Intel, in parallel, held an unveiling for its Panther Lake chips—set to become available for orders beginning Tuesday.
Earlier the same day, Nvidia introduced its next-generation Vera Rubin AI platform, comprising six distinct chips. CEO Jensen Huang confirmed production is fully underway with expectations for market launch within the year.
