Rapid adoption of artificial intelligence (AI) continues to drive digital transformation of businesses around the globe, and the demands on data centers – and on the world’s energy supply – are skyrocketing.
As we collectively observe this unprecedented data center boom, the sheer demand for compute power continues to redefine infrastructure on a global scale.
At AMD, our vision for the future of data centers is clear: open standards are the foundation for innovation, scalability, and long-term success. But our commitment to open compute is not just open for open’s sake – it is open for the purpose of innovating towards leadership performance, energy efficiency and security. These are the three imperatives for AI-era data center modernization. Each is crucial to ensuring that data centers can handle continued massive increases in workloads, power, and security challenges.
I had the honor of presenting on these topics during the Open Compute Project Foundation’s 2024 OCP Global Summit. You can watch a recording of the session here.
Then and today, I want to emphasize how AMD – together with the server ecosystem and industry standards bodies like OCP – are not only acting on these imperatives, but paving the way to create open, scalable solutions to redefine what’s possible in data center architecture.
1) The Imperative for Leadership Performance: In the AI era, data centers must be equipped to manage large-scale, compute-heavy tasks efficiently. AMD’s EPYC™ CPUs have played a vital role in supporting these growing workloads. When AMD introduced EPYC in 2017, we held less than 1% of the server market. Fast forward through five generations of AMD processors to today – and that percentage has grown to 34% of market share. We have grown because our customers increasingly recognize the performance, efficiency and value they get from AMD EPYC-based servers. Our newly announced 5th-generation EPYC CPUs are fundamental, critical building blocks for tomorrow’s modern data center.
2) The Imperative for Energy Efficiency: Energy consumption in data centers has become a critical challenge as data and AI workloads have grown exponentially. Today, many of the world’s data centers have a large share of servers that are at least 3- to 5-years old. If an organization with 1,000 top-of-the-stack CPUs from four years ago replaced that hardware today, they would need only 131 of AMD’s newly released 5th generation EPYC processors. This technology upgrade would result in 68% less power usage, up to 87% fewer servers, and a 67% reduction in total cost of ownership over three years.[i]
Looking at these benefits on a larger scale, if 1M of these aging, inefficient servers around the globe were replaced with 130K EPYC 9965 servers, this could save 2.6 million tons of CO2 emissions annually – the equivalent of removing 614,000 cars from the road or planting 3.1 million acres of forest[ii].
3) The Imperative for Confidential Computing: Security concerns are mounting as AI becomes more embedded in business processes. With vast amounts of data being processed, data centers are increasingly vulnerable to sophisticated attacks. Ensuring that sensitive data remains encrypted throughout its lifecycle is vital for maintaining trust in our AI-driven world. AMD has been at the forefront of confidential computing since 2017, when we introduced Secure Encrypted Virtualization (SEV). This technology, built into EPYC processors, ensures that all data remains encrypted, even when in use, providing a higher level of security. AMD’s 5th generation processors take this a step further, extending the secure boundary beyond the CPU to include all connected devices such as memory, accelerators, and data processing units (DPUs). This holistic approach ensures that every component within the data center operates in a trusted, encrypted environment.
To further promote transparency, AMD opened the source code for SEV to the public in 2023, reinforcing our pledge to open standards. This open-source approach allows the broader community to innovate without compromising on security, laying a foundation for a secure and scalable future.
AMD’s Commitment to Open Standards
A cornerstone of AMD’s vision for the future is our devotion to open standards. We participate in hundreds of open standards initiatives across the industry, including OCP, and these are essential for building scalable, flexible, and future-proof, AI-ready infrastructure. AMD plays a very critical part in the OCP Advisory Board and in the Server Project Steering Committee.
Most recently, in collaboration with Intel, we announced a collaboration to establish an x86 Ecosystem Advisory Group, a first-of-its-kind partnership aimed at bolstering and standardizing x86 architecture. The advisory group will develop a unified, consistent x86 platform, supporting both companies’ x86-based products across diverse computing environments, including data centers, cloud infrastructure, and edge devices.
By consolidating technical standards and providing a platform for developers to contribute insights, AMD and Intel aim to streamline compatibility, boost interoperability, and ease software development across x86 platforms. The advisory group, including major industry players like Google, Microsoft, and Meta, will work to introduce features to meet new demands in AI and customized chip technologies.
Driving Open Standards through Industry Collaboration
Our collaborations also reflect our commitment to open standards. For example, as memory becomes an integral part of scaling compute, AMD remains a critical part of JEDEC, and particularly the DDR5 MRDIMM Standard.
Another example is AMD’s recent work with Meta to develop the Sentinel Dome, Meta’s next-generation single-socket server, built on our 5th generation EPYC CPUs. Sentinel Dome will be integrated into Meta’s Yosemite V4 platform, and the entire specification is being donated to the OCP community.
This isn’t just about developing a cutting-edge server; it’s about making our innovations available to the broader community to drive real impact.
Together, we advance.
__________________________________________________________________________________________________________________________________________________________
[i] 9xx5TCO-002A: This scenario contains many assumptions and estimates and, while based on AMD internal research and best approximations, should be considered an example for information purposes only, and not used as a basis for decision making over actual testing. The AMD Server & Greenhouse Gas Emissions TCO (total cost of ownership) Estimator Tool - version 1.12, compares the selected AMD EPYC™ and Intel® Xeon® CPU based server solutions required to deliver a TOTAL_PERFORMANCE of 391000 units of SPECrate2017_int_base performance as of October 10, 2024. This estimation compares a legacy 2P Intel Xeon 28 core Platinum_8280 based server with a score of 391 versus 2P EPYC 9965 (192C) powered server with a score of 3000 (https://www.spec.org/cpu2017/results/res2024q4/cpu2017-20240923-44837.pdf) along with a comparison upgrade to a 2P Intel Xeon Platinum 8592+ (64C) based server with a score of 1130 (https://spec.org/cpu2017/results/res2024q3/cpu2017-20240701-43948.pdf). Actual SPECrate®2017_int_base score for 2P EPYC 9965 will vary based on OEM publications.
Environmental impact estimates made leveraging this data, using the Country / Region specific electricity factors from the 2024 International Country Specific Electricity Factors 10 – July 2024, and the United States Environmental Protection Agency 'Greenhouse Gas Equivalencies Calculator'.
For additional details, see https://www.amd.com/en/legal/claims/epyc.html#q=epyc4#SP9xxTCO-002A.
[ii] 9xx5TCO-003A: This scenario contains many assumptions and estimates and, while based on AMD internal research and best approximations, should be considered an example for information purposes only, and not used as a basis for decision making over actual testing. The AMD Server & Greenhouse Gas Emissions TCO (total cost of ownership) Estimator Tool - version 1.12, compares the selected AMD EPYC™ and Intel® Xeon® CPU based server solutions required to deliver a TOTAL_PERFORMANCE of 391000000 units of SPECrate®2017_int_base performance as of October 10, 2024. This estimation compares a legacy 2P Intel Xeon 28 core Platinum_8280 based server with a score of 391 versus 2P EPYC 9965 (192C) powered server with an score of 3000 (https://www.spec.org/cpu2017/results/res2024q4/cpu2017-20240923-44837.pdf) along with a comparison upgrade to a 2P Intel Xeon Platinum 8592+ (64C) based server with a score of 1130 (
https://spec.org/cpu2017/results/res2024q3/cpu2017-20240701-43948.pdf). Actual SPECrate®2017_int_base score for 2P EPYC 9965 will vary based on OEM publications.
Environmental impact estimates made leveraging this data, using the Country / Region specific electricity factors from the 2024 International Country Specific Electricity Factors 10 – July 2024', and the United States Environmental Protection Agency 'Greenhouse Gas Equivalencies Calculator'.
For additional details, see https://www.amd.com/en/legal/claims/epyc.html#q=epyc4#SP9xxTCO-003A
