AMD Cements Data Center Leadership at Financial Analyst Day 2025

News Snapshot

• AMD reiterated its annual accelerator cadence and previewed “Helios” rack-level AI solution targeted for Q3’26 availability
• AMD ROCm adoption is accelerating, with downloads up ~10x YoY; a strong signal of developer momentum
• AMD EPYC processors continue their strong adoption rate, with a ~40% revenue market share¹

AI is transforming every layer of the data center, creating a new industrial revolution of IT. The modern AI factory requires harmonized tools engineered to work together: accelerated clusters for LLMs, general‑purpose compute and storage to serve them, and advanced networking to secure and accelerate data movement.

At our Financial Analyst Day, we laid out how AMD is building the end‑to‑end platform with EPYC™ CPUs for compute, AMD Instinct™ GPUs for training and inference, Pensando™ networking to offload and protect the data path, and ROCm™ to unify the software stack. From chiplets to clusters, we’re engineering the performance and efficiency that will allow customers to run AI at a global scale.

Connecting Billions of Users Every Day

Data-intensive companies rely on AMD EPYC processors to run mission-critical workloads. From hyperscalers and cloud-native innovators like Netflix and Uber to enterprises modernizing their data centers, AMD continues to gain momentum.

• 60 percent of the Fortune^® 100 companies have deployed AMD EPYC processors².
• 5^th Gen AMD EPYC™ processors can accelerate time to results by 1.8x for industry workloads³ and can enable approximately 80% lower TCO for core IT work⁴.

"Venice," the next generation of EPYC CPUs, is expected to extend leadership in performance, density and energy efficiency for AI and general-purpose workloads. 

Advancing AI Leadership Across AMD Instinct GPUs and ROCm

We are accelerating AI innovation with an open-source software stack that scales from AI inference to training. Seven of the world’s top ten AI companies now deploy AMD Instinct accelerators at scale, underscoring growing industry trust in AMD hardware, software, and systems.

At Financial Analyst Day, we also revealed more about the next generation AMD Instinct™ MI400 portfolio, which is designed to advance AI and scientific computing with leadership performance and efficiency. Building on this momentum, AMD detailed the following products within the MI400 family:

• The AMD Instinct MI455X GPU. The most advanced AI accelerator; they are co-designed for next generation AI workloads and will utilize the AMD leadership chiplet architecture, 3.5D packaging and 12x HBM4 stacks of memory to drive leadership performance and scale for AI training and Inference workloads. AMD Instinct MI455X GPUs power the AMD “Helios” rack-scale platform, which provides the foundation to deliver the open, scalable infrastructure that will power the world’s growing AI demands.
• The AMD Instinct MI430X GPU. Optimized for a blend of Sovereign AI and HPC compute, these GPUs will deliver hardware based FP64 performance for HPC workloads and high-capacity HBM4 memory for AI capabilities. The AMD Instinct MI430X GPUs are at the heart of the newly announced Discovery supercomputer at Oak Ridge National Labs. 

At the foundation of this innovation at scale is the AMD ROCm open software stack. Delivering a unified software stack and consistent performance for everything from a single GPU to entire clusters, ROCm puts open source and developers first and its adoption is rapidly accelerating, with a 10x year-over-year increase in downloads and 2 million models supported on Hugging Face.

AMD also highlighted its relentless progress and focus on developers with ROCm including:

• Building on software in leading open-source AI projects, with support for frameworks and tools such as PyTorch, TensorFlow, JAX, Triton, Hugging Face Transformers, vLLM, SGLang, Ollama, ComfyUI and Unsloth.
• Building out higher levels of abstraction to make code portable across platforms
• Expanding AI capabilities to help developers optimize code generation and automate GPU Programming.

With ROCm, AMD continues to shorten the path from idea to deployment. Through faster time to releases, deeper framework integration, and a developer-first approach, AMD is giving teams the tools to push performance boundaries and bring new AI innovations to market faster.

Networking that Scales with AI

AMD advanced networking technologies provide the fabric that stitches AI solutions together, creating an open, scalable and adaptive platform for front-end, scale-up, scale-out and scale-across networking - accelerating intensive workloads and enabling innovation at the speed of AI.

• Front-End connects users, storage, and applications by accelerating AI and cloud workloads, offloading networking and security tasks from CPUs and GPUs, and helping secure data in motion with line-rate encryption and stateful firewall protection. AMD programmable P4-based engines power our Pensando^™ NICs and allow incredible flexibility in offloading concurrent services – SDN, storage, security – while keeping full line-rate performance.
• Scale up tightly couples hundreds of GPUs in a cluster. AMD Instinct™ MI450 Series GPUs can deliver up to 3.6 TB/s of bandwidth per GPU, and using the UALink™ protocol, AMD enables coherent GPU-to-GPU communication across the scale-up fabric.
•  Scale out links hundreds of thousands of GPUs to work as one. AMD Pensando AI NICs are Ultra Ethernet Consortium (UEC) ready, delivering the performance and reliability needed for large-scale AI workloads. These NICs feature programmable architectures that evolve rapidly to support to reduce switch costs, advanced reliability mechanisms, and accelerator offloads that enhance workload performance.
• Scale across federates data centers for giga-scale performance, with intelligent traffic management, dynamic load balancing, and an open systems approach that maintains line-rate communication between distributed environments.

From Exascale to Enterprise AI: A Unified Compute Platform

Stitching it all together, we showcased the “Helios” rack-scale platform that brings together AMD EPYC CPUs, AMD Instinct GPUs, AMD Pensando advanced networking and AMD ROCm software to deliver performance, efficiency and scalability for large-scale AI infrastructure. “Helios” is expected to be available in Q3 2026. We are delivering solutions on an annual cadence, with our next generation AI rack-scale platform scheduled to launch in 2027, powered by AMD EPYC “Verano” CPUs, AMD Instinct MI500 Series GPUs and AMD Pensando “Vulcano” networking. 

Advancing the AI Factory of the Future

AMD continues to drive the technologies that make the world’s most advanced computing possible. By unifying compute, acceleration, networking, and software into a cohesive platform, AMD is redefining performance and efficiency for the data centers powering the AI revolution.

With proven execution, deep ecosystem support and a clear product roadmap, AMD is positioned to lead the next era of the data center and AI infrastructure.

Cautionary Statement

This blog contains forward-looking statements concerning Advanced Micro Devices, Inc. (AMD) such as the features, functionality, performance, availability, timing and expected benefits of AMD products, including “Helios” rack-scale platform; AMD Instinct™ MI400 series accelerators and MI500 series accelerators; and AMD being positioned to lead the next era of data center and AI infrastructure, which are made pursuant to the Safe Harbor provisions of the Private Securities Litigation Reform Act of 1995. Forward-looking statements are commonly identified by words such as "would," "may," "expects," "believes," "plans," "intends," "projects" and other terms with similar meaning. Investors are cautioned that the forward-looking statements in this presentation are based on current beliefs, assumptions and expectations, speak only as of the date of this presentation and involve risks and uncertainties that could cause actual results to differ materially from current expectations. Such statements are subject to certain known and unknown risks and uncertainties, many of which are difficult to predict and generally beyond AMD's control, that could cause actual results and other future events to differ materially from those expressed in, or implied or projected by, the forward-looking information and statements. Investors are urged to review in detail the risks and uncertainties in AMD’s Securities and Exchange Commission filings, including but not limited to AMD’s most recent reports on Forms 10-K and 10-Q. 

AMD does not assume, and hereby disclaims, any obligation to update forward-looking statements made in this presentation, except as may be required by law.