Advanced Micro Devices NASDAQ: AMD executives used a question-and-answer session at the company’s Advancing AI 2026 conference to expand on its AI data center roadmap, customer engagements and expectations for growth in CPUs, GPUs and full rack-scale systems.
Chair and CEO Dr. Lisa Su said AMD is “tremendously excited” about the opportunity in AI and highlighted the company’s Venice CPU launch and Helios rack-scale systems as key parts of its strategy. Matt Ramsay, who leads financial strategy and investor relations at AMD, told participants that management would not discuss near-term financial results ahead of the company’s upcoming second-quarter earnings report.
AMD Sees Expanding CPU Opportunity From Agentic AI
Executives said AI workloads are increasing demand for CPUs, particularly as agentic AI requires more orchestration around end-to-end workloads. Su said AMD remains focused on capturing more than 50% of the CPU market, citing progress in recent quarters and rising customer interest in Venice.
Dan McNamara, who runs AMD’s server business, said the company’s estimate of the CPU server market was based on customer discussions and analysis of AMD’s own workloads. In the “outer years,” he said agentic AI applications could represent “probably like 50%” of the CPU server opportunity.
Su added that CPU-to-GPU ratios could change meaningfully as AI systems evolve. While some head-node configurations today may use four GPUs per CPU, she said agentic AI could eventually push the ratio above one CPU per GPU, potentially reaching two CPUs for one GPU in some scenarios.
Helios Shipments to Begin in September
Su clarified that AMD expects first shipments of Helios systems to begin in the third quarter, specifically in September. She said the ramp will continue into the fourth quarter and the first half of next year.
“We’ve actually built the ramp this way because it is a complex system,” Su said, adding that AMD wants original design manufacturers to tune the manufacturing process and align shipments with customer data center buildouts.
On customer deployments, Su said Anthropic, OpenAI and Meta moving into Helios is “a big deal for AMD.” Regarding Anthropic, she said AMD will start shipments for the first gigawatt in the first half of 2027 and aims to get as much of that first gigawatt into 2027 as possible, depending on data center readiness. AMD previously announced up to 2 gigawatts for the MI450 engagement with Anthropic.
Customer Deals Seen as Multi-Generation Relationships
Asked about the Anthropic agreement, Su said each large customer engagement is structured differently, but she emphasized that customers generally do not choose an accelerator for only one generation because of the engineering effort involved.
“We are actively talking with every one of our largest customers, including Anthropic, about what’s beyond MI450,” Su said. She said there is “a lot of excitement” around MI500 and discussions about future workloads beginning with MI600.
Vamsi Kompella, who runs AMD’s AI business, said AMD is also working with Anthropic to tune and extend Claude’s capabilities for high-performance optimization on AMD platforms. He said AMD’s open approach to instruction sets, compilers and tool chains helps AI systems become productive on the platform more quickly.
Kompella also discussed ROCm.ai, calling it AMD’s biggest software leap since the early days of its strategy. He said collaborations with OpenAI on Codex and Anthropic on Claude are expected to improve developer access to AMD platforms over the coming months.
Manufacturing, Power and Deployment Are Key Focus Areas
Su said AMD’s market projections consider not only demand but also power availability, supplier capacity and customer capital. She said AMD has planned capacity for “significant growth” in 2027 and 2028, while longer-term growth in 2029 and 2030 would require the broader ecosystem to build at a similar pace.
Forrest Norrod, who leads AMD’s data center business, said AMD is working closely with OEM and ODM partners, including Sanmina and Wiwynn, to ensure capacity to build, integrate, test and validate rack-scale systems. He also said AMD retained a large services arm from its ZT acquisition, which is being used for internal deployments and to help customers deploy MI350 and MI455 systems.
Su added that AMD now works with customers much earlier in the data center planning process, saying the company has “easily 12 to 18 months of visibility” into power planning and corresponding GPU and Helios system needs.
Roadmap Includes MI500 Networking and Memory Flexibility
Norrod said the MI500 generation is expected to begin a transition from purely electrical scale-up networking toward optical networking, though he emphasized it will not be an immediate shift. He said AMD is working with ecosystem partners and expects optical technologies to play a larger role over future generations.
On scale-up protocols, Norrod said MI450 supports UALink transported over Ethernet and that ESUN is a set of Ethernet extensions that can help with that approach. He said AMD expects UALink over Ethernet to continue into MI500, while adding that the company will provide more detail closer to the MI500 timeframe.
Asked about HBM memory, Kompella said AMD studies workload characteristics and separates bandwidth and capacity considerations. He said AMD’s chiplet architecture gives it flexibility to optimize memory capacity while preserving bandwidth constraints. Su added that memory capacity remains valuable to customers, including for inferencing performance, but said AMD will work to ensure memory is used efficiently because it is a significant part of total cost of ownership.
Su closed the session by saying AMD views AI as “a complete compute picture,” spanning CPUs, GPUs, Helios systems and other compute elements. She said AMD believes it can differentiate through an end-to-end approach across AI infrastructure.
About Advanced Micro Devices (NASDAQ:AMD)
Advanced Micro Devices, Inc NASDAQ: AMD is a global semiconductor company that designs and sells microprocessors, graphics processors, chipsets and adaptive computing solutions for a broad set of markets. The company's product portfolio includes consumer and commercial CPUs under the Ryzen and Threadripper brands, data center processors under the EPYC brand, and Radeon graphics processing units for gaming and professional visualization. AMD also offers semi-custom system-on-chip (SoC) products for gaming consoles and other specialized applications, and provides supporting software and platform technologies for OEMs, cloud service providers and end users.
Founded in 1969, AMD has evolved from a supplier of logic chips into a diversified, fabless semiconductor designer.
This instant news alert was generated by narrative science technology and financial data from MarketBeat in order to provide readers with the fastest reporting and unbiased coverage. Please send any questions or comments about this story to contact@marketbeat.com.
Before you consider Advanced Micro Devices, you'll want to hear this.
MarketBeat keeps track of Wall Street's top-rated and best performing research analysts and the stocks they recommend to their clients on a daily basis. MarketBeat has identified the five stocks that top analysts are quietly whispering to their clients to buy now before the broader market catches on... and Advanced Micro Devices wasn't on the list.
While Advanced Micro Devices currently has a Moderate Buy rating among analysts, top-rated analysts believe these five stocks are better buys.
View The Five Stocks Here
The AI boom is creating opportunities across semiconductors, cloud computing, enterprise software, infrastructure, cybersecurity, and automation.
Inside this report, you’ll find 10 companies positioned to benefit as artificial intelligence moves from hype to real-world deployment and becomes a core growth driver for corporate America.
Get This Free Report
