AMD projects sustained growth
AMD CEO Lisa Su projected roughly 35% annual growth to 2030 and framed a $1.6 trillion market opportunity as cloud and custom silicon efforts scale. Commentary in the same social posts noted that Amazon’s Graviton and Trainium offerings are maturing as alternatives to Nvidia and Intel at scale. (x.com) (x.com)
Advanced Micro Devices is telling investors its growth story now runs through data centers, not just personal computers. At its November 11, 2025 analyst day, Chief Executive Lisa Su said AMD is targeting more than 35 percent annual revenue growth over the next three to five years. (ir.amd.com) AMD tied that forecast to a market it says could reach $1 trillion by 2030, with a goal of more than $100 billion in data center revenue and more than $20 in non-Generally Accepted Accounting Principles earnings per share by 2030. The company presented that plan in New York as part of its Financial Analyst Day webcast. (ir.amd.com) That pitch starts from a simple shift in the chip business: cloud providers and artificial intelligence developers are buying far more server processors and accelerators than consumer laptop chips. AMD reported $16.6 billion in 2025 data center revenue, up 32 percent from 2024, out of $34.6 billion in total company revenue. (ir.amd.com) AMD’s data center products span two jobs inside a server. Epyc processors handle general computing tasks, while Instinct accelerators are built for the heavy matrix math used to train and run large artificial intelligence models. (ir.amd.com) Su’s case to investors is that customers want more than one supplier after years of tight supply and Nvidia dominance in artificial intelligence chips. CNBC reported that AMD said its artificial intelligence data center business could grow about 80 percent a year over the same three-to-five-year period and reach tens of billions of dollars in sales by 2027. (cnbc.com) The pressure on AMD does not come only from Nvidia. Amazon Web Services has spent years building its own chips, with Graviton for central processing unit workloads and Trainium for artificial intelligence training and inference inside Amazon’s cloud. (aws.amazon.com 1) (aws.amazon.com 2) Amazon says Graviton-based instances now account for more than half of new central processing unit capacity added to Amazon Web Services for a third straight year, and 98 percent of the top 1,000 Amazon Elastic Compute Cloud customers use Graviton. In December 2025, Amazon also introduced Graviton5-based M9g instances in preview with up to 25 percent higher compute performance than the prior generation. (aboutamazon.com) (aws.amazon.com) Amazon is making a similar argument in artificial intelligence hardware. Amazon says Trainium chips are designed for large-scale model training and inference with lower cost and scalable clusters, giving cloud customers an in-house alternative to renting Nvidia systems. (aws.amazon.com 1) (aws.amazon.com 2) That leaves AMD trying to win two fights at once: take share from Intel in server central processors and from Nvidia in artificial intelligence accelerators, while also persuading cloud giants not to replace merchant chips with their own designs. The next test comes on April 28, 2026, when AMD is scheduled to report first-quarter 2026 results after laying out its long-range targets five months earlier. (ir.amd.com)
