CES 2026: AMD Keynote: AI Everywhere, For Everyone

Dr. Lisa Su, chair and CEO of AMD, shared her vision for delivering future AI solutions at the CES 2026 opening keynote session on Monday evening, alongside innovators and executives spanning the healthcare, space, science, agentic and physical AI, gaming, and PC industries.

Su’s keynote centered on AMD’s motto of “AI Everywhere, For Everyone,” the idea that artificial intelligence is just beginning its transformation of computing and society, and that AMD’s technology is poised to power that future.

“This moment in tech not only feels different, AI is different,” Su enthused. “AI is the most powerful tech that has ever been created, and it can be everywhere for everyone.”
Additionally, Su underscored AI as the defining force in technology today, likening its adoption to the rapid rise of the internet and mobile phones. She projected up to five billion active users within five years as AI becomes indispensable across cloud services, enterprise workflows, personal computing, science, healthcare, and more, citing how AI usage grew from a million users to more than a billion in just a few years.

At the heart of AMD’s presentation was the immediate need for an increase in compute capacity—the raw performance required to train, run, and scale modern AI workloads. Su highlighted that global compute demand has already jumped from about 1 zetaflop in 2022 to 100 zetaflops by 2025, and that meeting the future demands of increasingly smart and agent-based AI will require expanding that capacity another 100 times toward the yottaflop era.

To enable that vision, AMD introduced Helios, a next-generation AI infrastructure rack-scale platform designed for hyperscale data centers. Built around the newly unveiled Instinct MI455 accelerators and next-gen EPYC “Venice” CPUs, Helios integrates high-speed networking and modular rack design to deliver up to 2.9 exaflops per rack. Su emphasized that this tightly engineered system, which uses advanced 2 nm/3 nm process tech and HBM4 memory to create a scalable backbone for frontier models and real-time inference at massive scale.

AMD also showcased a broader portfolio that reflects its all-layer AI strategy: from edge hardware to PCs and cloud infrastructure. On the client side, AMD highlighted Ryzen AI 400 Series processors and the AI Halo developer platform, aimed at bringing personalized AI performance to PCs. And in the data center, AMD continues to extend its footprint: Instinct accelerators power major cloud providers and advanced AI systems, while newer MI400-series variants like the MI440X target enterprise AI and sovereign AI applications.

Part of AMD’s emphasis during the keynote was also on the company’s ecosystem focus. Su reiterated that open software, anchored by AMD’s RockM stack with day-zero support for major frameworks, is essential for developers to build and scale AI efficiently.

Beyond silicon, the CES showcase spotlighted real-world impact scenarios: AI accelerating scientific research, healthcare diagnostics, and autonomous systems, from models that combine multimodal data to partnerships expanding AI vision for automotive autonomy.

“We’re entering this era of Yotta-scale computing, where the deployment of more powerful models everywhere will require a massive increase in the amount of compute in the world,” said Su. “Meeting that demand will take a broad portfolio of solutions from the largest systems in the cloud to AI PCs to embedded computing. And just as important, it takes an open ecosystem to build the industry standards. All of us are working together to bring AI everywhere for everyone.”

See also: CES 2026: ‘All-In’ Podcast At CES On AI Fundamentally Reshaping Investment Strategy, M&A, And Hiring