AMD has announced a series of new computing solutions that are poised to revolutionize data center operations in the AI era.
AMD expects revenue of approximately $6.7 billion for the third quarter of 2024. AMD’s second quarter revenue was $5.8 billion. AMD’s data center segment revenue in Q2 2024 was $2.8 billion driven by demand for AMD Instinct GPU and 4th Gen AMD EPYC CPU sales.
Key product releases include the 5th Gen AMD EPYC server CPUs, AMD Instinct MI325X accelerators, AMD Pensando Salina DPUs, AMD Pensando Pollara 400 NICs, and AMD Ryzen AI PRO 300 series processors designed for enterprise AI PCs. These innovations aim to meet the growing demands for data center performance and AI infrastructure, which are critical to the advancement of artificial intelligence and enterprise computing.
“The data center and AI represent significant growth opportunities for AMD,” AMD Chair and CEO, Dr. Lisa Su, said. “AMD’s EPYC CPUs, AMD Instinct processors, and Pensando DPUs are designed to handle demanding workloads.”
Lisa Su highlighted the company’s commitment to lead the global AI accelerator market, projected to reach $500 billion by 2028.
Transforming Data Center Solutions in the AI Era
AMD’s data center products are aimed at addressing a range of AI, enterprise, and cloud workloads, with solutions that enhance both compute power and system efficiency.
5th Gen AMD EPYC 9005 Series Processors: Built on the Zen 5 architecture, these processors offer up to 192 cores and deliver optimized performance for diverse data center needs. They are expected to be deployed across various platforms from major OEMs starting today.
AMD Instinct MI325X Accelerators: The latest AI accelerator from AMD provides superior performance and memory capabilities for demanding AI applications. AMD also previewed its upcoming MI350 series accelerators, expected in 2025, and made advancements in the development of MI400 series accelerators, planned for release in 2026.
Expanded AI Software and Networking: AMD has enhanced its ROCm open-source AI software stack, improving performance on the Instinct MI300X accelerator. Additionally, its networking solutions, such as the AMD Pensando Salina DPU and Pollara 400 NIC, help optimize data center operations, ensuring high scalability and performance.
Meeting the Future Needs of AI and Data Centers
AMD’s data center solutions are rapidly being deployed by industry-leading companies to power AI, cloud infrastructure, and enterprise computing, cementing AMD’s position as a main player in the AI and data center sectors.
Key partners detailed how they are leveraging AMD’s EPYC CPUs, Instinct accelerators, and Pensando DPUs.
Generative AI and Cloud Infrastructure Powered by AMD
AMD Instinct MI300X Accelerators: Since their launch in December 2023, these accelerators have been deployed at scale by top cloud and OEM partners, serving millions of daily users on popular AI platforms, including OpenAI’s ChatGPT, Meta’s LLaMA, and more than a million open-source models on Hugging Face. This deployment highlights AMD’s role in accelerating generative AI workloads across industries.
Google Cloud: Google underscored how its AI Hypercomputer, powered by AMD EPYC processors, maximizes AI return on investment (ROI). AMD’s EPYC processors are utilized across multiple Google Cloud instances for AI, high-performance computing, and confidential workloads. Google also announced new virtual machines (VMs) based on the EPYC 9005 series, expected to be available by early 2025.
Oracle Cloud Infrastructure (OCI): OCI detailed its use of AMD EPYC CPUs, Instinct accelerators, and Pensando DPUs to deliver high-speed, energy-efficient compute and networking services for customers such as Uber, PayPal, Red Bull Powertrains, and Fireworks AI. OCI also unveiled its new E6 compute platform, powered by the 9005 series of AMD EPYC processors.
Databricks: Databricks reported that testing with AMD Instinct MI300X GPUs resulted in a performance boost of over 50 percent for LLaMA and proprietary models. These GPUs, paired with AMD’s ROCm software stack, allow for seamless execution of models, benefiting from the MI300X’s large memory capacity and compute capabilities.
Collaboration with Leading Enterprises
Microsoft: Satya Nadella, CEO of Microsoft, highlighted the company’s close collaboration with AMD across its infrastructure and AI product offerings. He emphasized the strong performance of AMD Instinct MI300X accelerators on Azure and GPT workloads. Microsoft is already planning to leverage future generations of AMD Instinct, including the MI350 series, to drive AI performance while improving energy efficiency.
Meta: Meta detailed its use of AMD EPYC CPUs and MI300X accelerators to handle live traffic on its LLaMA 405B model, underscoring the scalability and performance of AMD solutions for its AI infrastructure. Meta is collaborating with AMD to optimize performance across the entire stack, from silicon to networking and software.
OEM Expansion and Data Center Modernization
Leading OEMs such as Dell, HPE, Lenovo, and Supermicro are expanding their EPYC processor-based platforms, which deliver energy-efficient, high-performance computing tailored for the AI-driven data center. These new platforms are designed to meet the growing demands of the AI era, modernizing enterprise data centers and supporting a wide array of workloads.
Baburajan Kizhakedath
