AMD's CEO, Lisa Su, has declared that the world of computing is shifting to a new performance metric, as we now enter the 'YottaScale' era, driven by massive demand for compute power.
AMD's Lisa Su Says That the AI Compute Demands Are Evolving At a Rapid Pace, Creating Room For TAM Growth
We are entering the next era of the AI infrastructure race, as recent announcements at CES from NVIDIA and AMD clearly indicate that the technology has significant room to grow, whether it is physical AI, edge AI, or agentic AI. While discussing the dramatic evolution of the computing market, AMD's CEO, Lisa Su, presented an astonishing estimate of the increase in compute power requirements over the next five years, claiming that YottaScale computing will become the new norm. And, more shockingly, the world would need up to 10+ YottaFLOPS by the end of the decade.
Interestingly, the need for power isn't driven only by the cloud; AI has managed to make its way across every mainstream workload, whether it is in edge computing, personal computing, healthcare, space, or several other areas. And, since the advancements being made within these individual segments are unprecedented, they have created a demand for computing power that no one could've anticipated a few years ago. Lisa Su predicts that in the YottaScale era, compute capacity requirements will be approximately 10,000 times higher than they were in 2022.
- 1 YottaFLOPS of AI compute = 344,828 Helios AI racks
When you look at such estimates, you'll realize that the TAM for the accelerator and rack-scale market is gigantic, and not limited to a single entity. AMD's CES keynote was a testament to how the company is moving to capitalize on the future demand from cloud and personal computing segments. The unveiling of AMD's Instinct MI455X AI GPUs, EPYC Venice CPUs, and Helios AI-rack scale solutions points to the company being an integral part of the next AI infrastructure race. Similarly, with Gorgon Point APUs, Team Red is also aiming to dominate the PC segment.
Compute power estimates are defintely optimistic, as they indicate that AI still has a long way to go. However, at the same time, they also raise concerns about whether this growth is sustainable when factoring in energy and land concerns.
About the author: Muhammad Zuhair is a hardware and technology reporter for Wccftech, specializing in the semiconductor industry and the complex interplay between technology, manufacturing, and geopolitics. His coverage focuses on the corporate strategies and technological roadmaps of industry giants like TSMC, NVIDIA, Samsung, and Intel. Zuhair's expertise lies in deconstructing complex topics such as fabrication nodes (e.g., 2nm process), the economic impact of policies like the CHIPS Act, and the strategic development of AI infrastructure from NVIDIA, AMD and Intel.
Follow Wccftech on Google to get more of our news coverage in your feeds.
