AMD presents the solutions with which it intends to scale computing capacity up to yottaflops for AI

AMD has presented the products with which it intends Unleash the true power of artificial intelligence (AI), a technology that requires scaling computing capacity from zettaflops to yottaflops, and which seeks to achieve this with the AMD Helios rack-scale platform and AMD Instinct accelerators.

Artificial intelligence has been the protagonist of the conference that AMD CEO Lisa Su held at CES 2026, in which she assured that “although the pace and speed of AI innovation have been incredible over the last few years, you haven’t seen anything yet.”

The company has shared the news with which it intends to reach the true power of AI, something that it will do together with its partners, and under the vision “AI Everywhere, for Everyone.”

To do this, AMD is working on the computing foundation that drives AI, on an infrastructure that allows global computing capacity to scale from current zettaflops to exceed 10 yottaflops in the next five yearsthanks to a modular and open rack-scale system design.

As Su explained, a yottaflop is a unit of computing speed that is equivalent to a one followed by 24 zeros, being 10 yottaflops. a computing capacity one hundred times greater than what was available in 2022.

To enable it, trust the AMD Helios rack-scale platformdesigned to train models with billions of parameters, capable of offering up to 3 exaflops of AI per rack and based on Instinct MI455X accelerators, EPYC Venice CPU, Pensando Vulcano networks and the ROCm software ecosystem.

Likewise, it has expanded the portfolio of AMD Instinct acceleratorsincluding the introduction of the AMD Instinct MI440X GPU for enterprise AI deployments, and has previewed the next generation AMD Instinct MI500 Series, scheduled for release in 2027 and expected to deliver a thousand-fold increase in AI performance over the MI300X GPUs launched in 2023.

AI EXPERIENCES FOR EVERYONE

Continuing with its vision, for the automotive sector and industry, it has announced Ryzen AI Embeddednew x86 processors with Zen 5 architecture, available in the P100 and X100 series, delivering high performance and efficient AI computing in embedded systems.

Likewise, it has unveiled processors las series Ryzen AI 400 y Ryzen AI Max+, with which it seeks to lead the range of computers with native AI, a product category that puts a personal assistant and a trainer in the hands of users who learn from them through everyday interactions.

By Editor

Leave a Reply