Source
AMD Tracks Ahead of Rack-Scale AI Energy-Efficiency Goal
AMD has achieved an estimated 4x increase in AI energy efficiency as of mid-2026, surpassing its original 3x target. This is part of a strategic goal to reach a 20x increase by 2030.
Why it matters
Improved efficiency allows data centers to increase AI performance without raising power consumption or operational costs. It also helps users overcome physical infrastructure limits regarding cooling and electricity.
The details
- Two 2030 racks may deliver the same compute as 570 2024 racks.
- AMD uses ROCm software to help developers deploy AI workloads more efficiently.
- The company focuses on co-optimizing silicon, memory bandwidth, interconnects, and system design.
Show entities and relationshipsHide entities and relationships
In this article
Key connections
AMD develops and maintains the ROCm software stack.
Sam Naffziger works at AMD
Sam Naffziger is Senior Vice President and Corporate Fellow at AMD.
AMD uses High Bandwidth Memory
AMD integrates high-bandwidth memory into its compute architectures to increase bandwidth and energy efficiency.
AMD ROCm is related to Artificial Intelligence
AMD ROCm software stack enables deployment and optimization of AI workloads.
Related events
AMD Tracks Ahead of 2030 Rack-Scale AI Energy-Efficiency Target
Get the weekly recap
The stories like this one, picked and explained — once a week, straight to your inbox.