At CES 2026 in Las Vegas, AMD chair and CEO Lisa Su used the show’s opening keynote to argue that the company’s growing portfolio of artificial intelligence products and cross industry collaborations is turning the promise of artificial intelligence into real world impact. Partners including OpenAI, Luma Artificial Intelligence, Liquid Artificial Intelligence, World Labs, Blue Origin, Generative Bionics, AstraZeneca, Absci and Illumina joined the presentation to describe how AMD hardware and software are powering advances from data center scale training to edge deployments.
Su framed the talk around what she called the blueprint for yotta scale compute, noting that compute infrastructure is the foundation of artificial intelligence and that accelerating adoption is pushing global capacity from today’s 100 zettaflops of compute capacity to a projected 10+ yottaflops in the next five years. AMD positioned its new “Helios” rack scale platform as an answer to that demand, describing it as delivering up to 3 artificial intelligence exaflops of performance in a single rack through an open, modular design that combines AMD Instinct MI455X accelerators, AMD Epyc “Venice” CPUs and AMD Pensando “Vulcano” NICs on the ROCm software stack. The company provided an early look at Helios while for the first time unveiling the full AMD Instinct MI400 Series portfolio and previewing its next generation MI500 Series GPUs, which are planned to launch in 2027 and are on track to deliver up to a 1,000x increase in artificial intelligence performance compared to the AMD Instinct MI300X GPUs introduced in 2023.
Within the MI400 lineup, AMD highlighted the new Instinct MI440X GPU for on premises enterprise artificial intelligence, designed for scalable training, fine tuning and inference in an eight GPU form factor, and positioned it as building on the previously announced Instinct MI430X GPUs that will power artificial intelligence factory supercomputers such as Discovery at Oak Ridge National Laboratory and the Alice Recoque exascale system in France. Extending its story to the edge, AMD introduced the Ryzen Artificial Intelligence Embedded P100 and X100 Series processors aimed at automotive, healthcare and autonomous systems that require efficient artificial intelligence compute in constrained environments. Su was joined on stage by Michael Kratsios, director of the White House Office of Science and Technology Policy, to discuss AMD’s role in the US government’s Genesis Mission, which includes the Lux and Discovery artificial intelligence supercomputers at Oak Ridge, and to announce AMD’s commitment of $150 million to expand access to artificial intelligence education as part of a broader national pledge that has already seen more than 15,000 students participate in the AMD Artificial Intelligence Robotics Hackathon with Hack Club.
