As AI expands from training to inference and agentic workloads, compute demand is accelerating rapidly. AMD delivers an open, full-stack AI platform that gives customers the flexibility to deploy the right compute for every workload.<\/p>\n
\u201cThe next phase of AI will span frontier models, agents and physical AI, creating new opportunities to bring intelligence everywhere,\u201d said Dr. Lisa Su, chair and CEO, AMD. \u201cRealizing that potential will take the entire industry working together. AMD is partnering across the ecosystem to deliver leadership compute and open platforms that give customers the performance, flexibility and choice to scale AI from the data center to the edge.\u201d<\/p>\n
AMD Helios: The Highest Performance Rack-Scale AI Solution Leading AI labs and cloud providers are choosing AMD Helios for its open, full-stack performance. They include OpenAI, Anthropic, Meta, Microsoft, Oracle, HUMAIN, Tensorwave, Vultr, Cirrascale and others. Systems will be available from leading OEMs, including Bull, HPE, Lenovo and Supermicro, as well as infrastructure partners Sanmina and Wiwynn.<\/p>\n At Advancing AI, AMD partners detailed how they deploy AMD AI infrastructure at scale for frontier training and inference:<\/p>\n Anthropic and AMD further outlined Wednesday\u2019s strategic partnership announcement to deploy up to 2 gigawatts of AMD Instinct MI455X GPUs in AMD Helios rackscale solutions. The companies are launching a multiyear engineering collaboration to use Claude to accelerate AMD software development. Specifically, the teams will use Claude to optimize workloads for AMD Instinct GPUs and accelerate ROCm software development. AMD will also broadly adopt Claude across its engineering and product development teams.OpenAI and AMD are partnering to optimize the full AI stack, from silicon to software. Leveraging OpenAI\u2019s Triton framework with AMD ROCm software, the companies are optimizing GPT-class workloads on AMD Instinct MI455X GPUs and AMD Helios racks. OpenAI expects to bring Helios online beginning in the fourth quarter of 2026, with deployments accelerating throughout 2027.Meta and AMD are co-designing for gigawatt-scale deployments, optimizing AMD\u2019s full AI compute stack for Meta workloads. Meta is now validating 6th Gen EPYC CPU platforms in its labs and has begun testing and validating workloads on AMD Helios racks as they prepare to deploy at scale.Cerebras and AMD are collaborating<\/a> to deliver a combined solution of Cerebras ultra-low-latency AI compute and AMD Helios high-throughput rack-scale infrastructure to help improve inference efficiency, scalability and economics for ultra-low-latency inference serving. <\/p>\n Delivering the Highest Performance Data Center CPUs and GPUs With AMD Instinct\u2122 MI400 Series GPUs<\/a>, AMD delivers powerful performance for cloud, enterprise and HPC workloads. AMD Instinct MI455X GPUs deliver 34x higher token throughput compared to MI355X GPUs7. For high-precision workloads, the AMD Instinct\u2122 MI430X accelerator is the most advanced for HPC and sovereign AI with up to 288 TFLOPS of hardware-based FP64 performance for scientific computing. Instinct MI430X accelerators are powering the next wave of exascale-class supercomputers across the U.S. and Europe.<\/p>\n AMD also launched the Instinct MI350P GPU, bringing seamless AI acceleration to existing infrastructure with leadership token economics. MI350P GPUs deliver up to 4.2x more tokens per second per dollar than the competition8.<\/p>\n Advancing the Open Software Ecosystem ROCm.ai is accelerating software enablement for AMD Instinct MI455X GPUs while optimizing performance. Leading open-source frameworks including PyTorch, Hugging Face, vLLM and SGLang are already enabled on MI455X and seeing great results.<\/p>\n Accelerating Next-Generation AI Infrastructure Next-generation EPYC server CPUs based on the \u201cZen 7\u201d architecture are coming in 2028. The \u201cFlorence,\u201d \u201cFerrara\u201d and \u201cFidenza\u201d CPUs are expected to extend AMD\u2019s leadership in density, performance, performance-per-system dollar and performance-per-watt.\u201cRavenna\u201d CPUs based on the \u201cZen 8\u201d architecture are coming in 2030, designed to continue AMD server CPU leadership.Next-generation AMD Instinct MI500 Series GPUs are coming in 2027, with next-generation compute, memory and interconnect technologies for leadership performance.AMD Instinct MI600 Series GPUs are coming in 2028.The next-generation AMD Helios 500 rackscale solution will be powered by AMD Instinct MI500 Series GPUs and AMD EPYC \u201cVerano\u201d CPUs, with next-gen Pensando \u201cComo\u201d and \u201cMonza\u201d networking. The AMD Helios 600 rackscale solution will follow, powered by AMD Instinct MI600 Series GPUs, EPYC \u201cFerrara\u201d CPUs and Pensando \u201cPalma\u201d and \u201cLevanzo\u201d networking. <\/p>\n Scaling AI Across Enterprise At Advancing AI, AT&T illustrated how it is deploying flexible enterprise AI using AMD technology across cloud, on-premises and air-gapped environments. AT&T is also using AMD Instinct GPUs and ROCm software to power its OTel 2.0 model, an open-source model trained specifically for telecoms.<\/p>\n With the AMD Ryzen\u2122 AI Halo developer platform, AMD delivers performance, efficiency and simplicity that makes local AI development accessible. More AMD Ryzen AI Halo platforms, powered by Ryzen\u2122 AI Max PRO 400 Series processors, will be available later this year from AMD and OEM partners.<\/p>\n Cisco and AMD are collaborating to combine AMD high-performance inference engines, including AMD Ryzen AI Halo systems, with Cisco networking, observability and security capabilities, so enterprises can deploy, govern and manage hybrid and local agentic AI at scale.\u00a0\u00a0<\/p>\n Advancing the Next Frontier of Physical AI The portfolio includes new AMD Kria AI system-on-modules<\/a> (SOMs), powered by the new AMD Ryzen AI Embedded X100 Series processors<\/a>, and the AMD Kria AI Robotics Developer Platform<\/a>, the first open, turnkey integrated platform for autonomous robotics combining CPU, GPU, NPU and FPGA compute. Together with an expanded open software ecosystem, AMD Kria AI solutions remove vendor lock-in and help developers and customers accelerate the path from prototype to production for next-generation physical AI systems.<\/p>\n Supporting Resources<\/p>\n About AMD CAUTIONARY STATEMENT <\/p>\n This press release contains forward-looking statements concerning Advanced Micro Devices, Inc. (AMD) such as AI accelerating demand for the full range of AMD silicon, driving AMD\u2019s total addressable market to ~$2 trillion in 2030; compute demand accelerating; the next phase of AI; the features, functionality, performance, availability, timing and expected benefits of AMD products, including, but not limited to, 6th Gen AMD EPYC\u2122 CPUs, AMD Instinct\u2122 MI400 Series GPUs, AMD Helios\u2122 AI rackscale solutions, AMD Ryzen\u2122 AI Embedded X100 processors and the AMD Kria\u2122 AI SOM and Robotics Developer Platform; expected plans, benefits, scalability and deployments with AMD partners including Anthropic, OpenAI, Meta and Cerebras; AMD\u2019s AI infrastructure product roadmaps through 2030; and AMD\u2019s collaboration with AT&T and Cisco, 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 press release are based on current beliefs, assumptions and expectations, speak only as of the date of this press release 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 are 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. Material factors that could cause actual results to differ materially from current expectations include, without limitation, the following: impact of government actions and regulations such as export regulations, import tariffs, trade protection measures, and licensing requirements; competitive markets in which AMD\u2019s products are sold; the cyclical nature of the semiconductor industry; market conditions of the industries in which AMD products are sold; AMD\u2019s ability to introduce products on a timely basis with expected features and performance levels; loss of a significant customer; economic and market uncertainty; quarterly and seasonal sales patterns; AMD’s ability to adequately protect its technology or other intellectual property; unfavorable currency exchange rate fluctuations; ability of third party manufacturers to manufacture AMD’s products on a timely basis in sufficient quantities and using competitive technologies; availability of essential equipment, materials, components (such as memory supply), substrates or manufacturing processes; ability to achieve expected manufacturing yields for AMD\u2019s products; AMD’s ability to generate revenue from its semi-custom SoC products; potential security vulnerabilities; potential security incidents including IT outages, data loss, data breaches and cyberattacks; uncertainties involving the ordering and shipment of AMD\u2019s products; AMD\u2019s reliance on third-party intellectual property to design and introduce new products; AMD’s reliance on third-party companies for design, manufacture and supply of motherboards, software, memory and other computer platform components; AMD’s reliance on Microsoft and other software vendors’ support to design and develop software to run on AMD\u2019s products; AMD\u2019s reliance on third-party distributors and add-in-board partners; impact of modification or interruption of AMD\u2019s internal business processes and information systems; compatibility of AMD\u2019s products with some or all industry-standard software and hardware; costs related to defective products; failure to maintain an efficient supply chain as customer demand changes; AMD’s ability to rely on third party supply-chain logistics functions; AMD\u2019s ability to effectively control sales of its products on the gray market; impact of climate change on AMD\u2019s business; AMD\u2019s ability to realize its deferred tax assets; potential tax liabilities; current and future claims and litigation; impact of environmental laws, conflict minerals related provisions and other laws or regulations; evolving expectations from governments, investors, customers and other stakeholders regarding corporate responsibility matters; issues related to the responsible use of AI; restrictions imposed by agreements governing AMD\u2019s notes, the guarantees of Xilinx\u2019s notes and the revolving credit agreement; AMD\u2019s ability to satisfy financial obligations under guarantees, leases and other commercial commitments; impact of acquisitions, joint ventures and\/or investments on AMD\u2019s business and AMD\u2019s ability to integrate acquired businesses; impact of any impairment of the combined company\u2019s assets; political, legal and economic risks and natural disasters; future impairments of technology license purchases; AMD\u2019s ability to attract and retain key employees; and AMD\u2019s stock price volatility. Investors are urged to review in detail the risks and uncertainties in AMD\u2019s Securities and Exchange Commission filings, including but not limited to AMD\u2019s most recent reports on Forms 10-K and 10-Q.<\/p>\n _________________________<\/p>\n 1Based on AMD Performance Labs estimates as of July 2026, tokens-per-dollar performance was calculated using the Kimi K2 Thinking workload (32K input \/ 8K output) on an AMD Helios rackscale solution compared to an NVIDIA Vera Rubin NVL72 rack. Results reflect estimated aggregate throughput across low, medium, and high-interactivity operating points and hourly pricing projection of system GPUs based on market conditions. System configurations may vary by manufacturer and may produce different results. MI400-025 3 EPYC-025D: As of July 2026, 6th Gen EPYC 9996 has 256 cores and 512 threads with SMT enabled which is higher than any other publicly disclosed 1P CPU 5 9xx6-013: Comparison based on published Top-of-stack core counts and 1Ku pricing for estimated highest Agents \/ CPU $ and threads \/ CPU $ AMD EPYC\u2122 9006 (512 threads), AMD EPYC\u2122 9005 (384 threads), and Intel\u00ae Xeon\u00ae 6 (256 threads) SKUs as of 7\/22\/2026. Threads derived as 2 threads per core (SMT). Intel and Xeon are trademarks of Intel Corporation or its subsidiaries. Source: https:\/\/www.amd.com\/content\/dam\/amd\/en\/documents\/solutions\/ai\/methodology-description.pdf Agent counts are estimates derived from available CPU thread resources used as a proxy under a consistent theoretical workload. Actual agent capacity and throughput will vary based on workload, model, memory, software, orchestration, and system configuration.<\/p>\n 6 9xx6-014: Comparison based on published Top-of-stack core counts and CPU W, Default CPU Power for 6th Gen EPYC, and TDPs for 5th Gen EPYC, Intel\u00ae Xeon\u00ae, and Nvidia Vera for estimated Highest Agents \/ CPU W across AMD EPYC\u2122 9956 (400W Default CPU Power), AMD EPYC\u2122 9965 (500W TDP), Nvidia Vera (450W TDP), ARM AGI (300W TDP), and Intel\u00ae Xeon\u00ae 6980P (500W TDP) powered servers as of 7\/22\/2026. 2 threads per core (SMT). 1 thread per core for ARM AGI. 7 MI400-020: Based on measurements and calculations by AMD Performance Labs in July 2026, for the AMD Instinct\u2122 MI455X GPU to determine measured token throughput at high, medium and low interactivity points run on Deepseek V4 Flash with FP4 serving compared to AMD Instinct\u2122 MI355X GPU. System manufacturers may vary configurations, yielding different results.<\/p>\n 8 MI350P-007: Based on AMD internal testing (July 2026), on a (1x) AMD Instinct MI350P GPU vs (1x) NVIDIA H200 NVL GPU on the Llama 3.3 70B Instruct (FP8) online serving output-throughput per dollar (tok\/s\/USD) comparison at ISL\/OSL 1024\/1024 across concurrency levels 1, 4, 8, 16, 32, 64, 128, 256, 512; median of 3 runs per point. MI350P based server internal AMD estimated pricing as $327,238.40 USD. RTX_PRO_6000 based server public list price reported on OEM website as $265,928.24 USD as of 7\/16\/2026. Stated results are the peak per-concurrency ratios: MI350P served via AIMS silogenai\/aim-instinct-meta-llama-llama-3-3-70b-instruct:0.12.0-rc6; H200 NVL via NVIDIA NIM nvcr.io\/nim\/meta\/llama-3.3-70b-instruct:2.0.6; RTX PRO 6000 via NVIDIA NIM nvcr.io\/nim\/meta\/llama-3.3-70b-instruct:2.0.6. Configuration: 8x AMD Instinct MI350P PCIe Card (CDNA4, gfx950, 128 CUs, 144 GB HBM3E, SPX compute \/ NPS1),\u202fvBIOS\u202f113-350P-01-1K1-000A, GPU driver 6.19.13-2353916.24.04,\u202fROCm\u202f7.14.0 (AMD-SMI 26.5.0); host 2P AMD EPYC 9455 (48-core), Dell PowerEdge XE7745, BIOS 1.7.6, microcode 0xb002162, SMT Enabled, Ubuntu 24.04.4 LTS, Linux 6.8.0-124-generic || NVIDIA RTX PRO 6000: 8x NVIDIA RTX PRO 6000 Blackwell Server Edition,\u202fvBIOS\u202f98.02.8D.00.01, GPU driver 595.45.04, CUDA 13.2; host 2P AMD EPYC 9455 (48-core), Dell PowerEdge XE7745, BIOS 1.6.4, microcode 0xb00215a, SMT Enabled, Ubuntu 24.04.4 LTS, Linux 6.8.0-124-generic. Sever manufacturers may vary configurations, yielding different results. Results may vary due to factors including system configurations, software\u202fversions\u202fand BIOS settings.<\/p>\n Contact:\u00a0 Liz Stine
Delivering frontier AI requires a fully integrated rack architecture, with every part of the stack pushing the boundaries of performance. AMD Helios rackscale solutions<\/a> are built for this, with co-optimized silicon spanning 72 high-performance AMD Instinct\u2122 MI455X GPUs and 18 powerful 6th Gen AMD EPYC\u2122 \u201cVenice\u201d CPUs, connected by AMD Pensando\u2122 front-end, scale-up and scale-out networking, and accelerated by AMD ROCm\u2122 open software. AMD Helios combines leadership compute performance, memory capacity and networking bandwidth to deliver up to 30% more tokens per dollar than the leading competitive solution1.<\/p>\n
6th Gen EPYC processors deliver the broadest server CPU portfolio for agentic AI,2 spanning cloud, enterprise, general-purpose and high-performance computing (HPC) workloads. With leadership per-core performance and the highest thread density3, they enable the most agents per watt, per dollar and per rack.4,5,6 For AI host nodes, 6th Gen EPYC CPUs<\/a> deliver the speed and memory bandwidth to keep accelerators fully fed. And for general-purpose servers, they bring leadership performance and energy efficiency to run business critical applications and AI support tasks.<\/p>\n
For developers, AMD ROCm is the open software platform with the performance, flexibility and ecosystem support needed to build and deploy AI on AMD hardware. Building on that foundation, AMD is introducing ROCm.ai<\/a>, an AI-driven development platform that helps developers build, optimize and deploy GPU software faster across AMD platforms. ROCm.ai brings AI-assisted GPU programming to developers by enabling popular coding agents such as Claude, Codex and Cursor to understand AMD platforms and ROCm natively.<\/p>\n
AMD is extending its annual cadence of CPU, GPU, networking and rack-scale innovation through 2030. The company shared new details on its roadmaps, including:<\/p>\n
Leading enterprises run on AMD infrastructure, from cloud, hybrid and on-prem data centers to AI-enabled PC fleets. AMD technologies are helping customers scale quickly and accelerate enterprise transformation.<\/p>\n
As AI expands across cloud, enterprise and local systems, the next frontier is bringing intelligence into machines that perceive, reason and act in the physical world. Building on a long legacy in robotics with AMD FPGAs and adaptive SoCs, AMD introduced AMD Kria\u2122 AI solutions, extending the company\u2019s robotics capabilities from the robot body to the robot brain. AMD uniquely brings AI perception, reasoning and agentic decision-making and control together on a single platform to deliver the performance required for demanding real-world robotic systems.<\/p>\n
AMD (NASDAQ: AMD) drives innovation in high-performance and AI computing to solve the world\u2019s most important challenges. Today, AMD technology powers billions of experiences across cloud and AI infrastructure, embedded systems, AI PCs and gaming. With a broad portfolio of AI-optimized CPUs, GPUs, networking and software, AMD delivers full-stack AI solutions that provide the performance and scalability needed for a new era of intelligent computing. Learn more at\u00a0www.amd.com<\/a>.<\/p>\n
2 EPYC-068 – The AMD EPYC server CPU portfolio spans the industry\u2019s broadest ranges of data center deployments, from general-purpose enterprise, cloud, telecom, SMB, and HPC systems to emerging AI environments including sandboxed agentic AI deployments and GPU head node servers. AMD EPYC 6th Generation platforms extend this breadth by uniquely combining high core and thread density of up to 512 threads, advanced memory bandwidth of up to 16 channels of 12.8 GT\/s MRDIMM support, next-generation PCIe\u00ae Gen 6 connectivity, and select SKUs with boost frequencies up to 5 GHz.<\/p>\n
4 9xx6-012: Based on estimated performance data for Nvidia, Intel\u00ae, and AMD EPYC\u2122 Server Processors for Agentic AI, the AMD EPYC9996 provides the most\u00a0agents\u00a0per rack at a 100Kw power envelope per rack.\u00a0
\u00a0Compared to the cores per rack of Nvidia Vera (88c) powered server racks:\u00a0
\u00a0– The AMD EPYC 9996 (256C) provides 2.08x the cores (and threads with SMT) per rack\u00a0
– The AMD EPYC 9965 (192C) provides 1.86x\u00a0the cores\u00a0(and threads with SMT) per rack\u00a0
– The Intel Xeon 6980P (128C) provides 1.24x\u00a0the cores\u00a0(and threads with SMT) per rack\u00a0
Source: https:\/\/www.amd.com\/content\/dam\/amd\/en\/documents\/solutions\/ai\/methodology-description.pdf\u00a0\u00a0
Agent counts are estimates derived from available CPU thread resources used as a proxy under a consistent theoretical workload. Actual agent capacity and throughput will vary based on workload, model, memory, software, orchestration, and system configuration.<\/p>\n
Starting with the 6th Gen AMD EPYC\u2122 server processor family, AMD uses Default CPU Power to describe processor power consumption, succeeding AMD’s historical TDP reference. Default CPU Power reflects total power consumed across the processor’s compute and I\/O dies for the stated performance target. Default CPU Power and TDP may both serve as processor power references for product comparison, platform planning, and performance-per-watt analysis. Agent counts are estimates derived from available CPU thread resources used as a proxy under a consistent theoretical workload. Actual agent capacity and throughput will vary based on workload, model, memory, software, orchestration, and system configuration. Intel Xeon TDP from ark.intel.com. Nvidia Vera TDP from https:\/\/developer.nvidia.com\/blog\/nvidia-vera-cpu-sets-a-new-standard-for-agentic-workloads-in-ai-factories\/. ARM AGI Specifications from https:\/\/www.arm.com\/products\/cloud-datacenter\/arm-agi-cpu#Specifications<\/a><\/p>\n
Brandi Martina\u00a0
\u00a0AMD Communications\u00a0
(512) 705-1720\u00a0
Brandi.martina@amd.com\u00a0<\/p>\n
AMD Investor Relations
(720) 652-3965
liz.stine@amd.com<\/p>\n![\"\"](\"https:\/\/www.europesays.com\/ai\/wp-content\/uploads\/2026\/07\/1784844638_843_Advanced-Micro-Devices-Inc-.png\")
<\/p>\n","protected":false},"excerpt":{"rendered":"News Highlights At Advancing AI 2026, AMD launches 6th Gen AMD EPYC\u2122 CPUs, AMD Instinct\u2122 MI400 Series GPUs,…\n","protected":false},"author":2,"featured_media":112706,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[6],"tags":[2208,179,7493,24,3125,43471,43312,829,3887,59507,57813],"class_list":["post-117079","post","type-post","status-publish","format-standard","has-post-thumbnail","category-agentic-ai","tag-advanced-micro-devices","tag-agentic-ai","tag-agentic-artificial-intelligence","tag-ai","tag-amd","tag-epyc","tag-helios","tag-inc","tag-inference","tag-instinct","tag-nasdaqamd"],"_links":{"self":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts\/117079","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/comments?post=117079"}],"version-history":[{"count":0,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts\/117079\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/media\/112706"}],"wp:attachment":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/media?parent=117079"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/categories?post=117079"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/tags?post=117079"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}