December 8, 2023
AMD is coming to market with a new slate of chips optimized for artificial intelligence, including the AMD Instinct MI300 Series data center AI accelerators, ROCm 6 open software stack with new features for large language models, and Ryzen 8040 Series processors with Ryzen AI. The new offerings have received a welcome reception from customers including Microsoft, Oracle, Meta Platforms and Dell, among others that can benefit from building a strong network of suppliers of AI chips. The market is currently dominated by Nvidia, which is challenged to meet existing demand.
AMD unveiled the new products at its Advancing AI conference Wednesday in San Jose, California. Both AMD and Nvidia are headquartered in Silicon Valley and design chips that are manufactured in Taiwan by TSMC.
“AI is the future of computing” and AMD intends to supply chips “end-to-end,” from massive cloud data center installations to enterprise clusters, smart devices and PCs, AMD Chair and CEO Lisa Su said in a news announcement from the event.
“Meta, OpenAI and Microsoft said they will use AMD’s newest AI chip, the Instinct MI300X — a sign that tech companies want alternatives to the expensive Nvidia graphics processors that have been essential for artificial intelligence,” CNBC writes, adding that “if the MI300X is good enough and inexpensive enough when it starts shipping early next year, it could lower costs for developing AI models.”
“We are seeing very strong demand for our new Instinct MI300 GPUs, which are the highest-performance accelerators in the world for generative AI,” Su said. The Wall Street Journal reports that “AMD is among a number of startups and large tech companies trying to capitalize on AI demand,” Intel being another. Amazon and Google have already begun designing their own AI chips, and Microsoft in November announced it would be producing a bespoke AI chip, the Maia 100.
Su said on the company’s Q3 earnings call that AMD expects revenue of $400 million from its AI chips in Q4, and more than $2 billion in 2024, driven by AI and the demands it places on processing power.
“The market for AI chips in data centers, where most artificial-intelligence computation takes place, could grow to $400 billion by 2027, Su said,” according to WSJ, noting her estimate was far more bullish than others, including Gartner, “which in August predicted AI-chip revenue to reach around $119 billion in 2027, up from about $53 billion this year.”