Advanced Micro Devices (AMD) announced new details on Thursday about its next-generation AI chips, the Instinct MI400 series, set to ship next year. The company also introduced a full server rack system called Helios, which allows thousands of these chips to work together as one unified “rack-scale” system.
AMD CEO Lisa Su revealed at a launch event in San Jose that the Helios rack is architected as a single, massive compute engine. This innovative design allows the chips to be combined at scale, creating a powerful AI computing resource. Su contrasted Helios with Nvidia’s upcoming Vera Rubin racks, expected next year, emphasizing AMD’s integrated approach.
OpenAI CEO Sam Altman joined Su on stage and confirmed that his company plans to use the MI400 chips. Altman expressed enthusiasm, calling the specs “totally crazy” and anticipating it to be “an amazing thing.”
Competing Against Nvidia’s Dominance
AMD’s rack-scale technology enables its MI400 chips to compete directly with Nvidia’s Blackwell chips, which currently feature configurations with up to 72 GPUs linked together. Nvidia remains AMD’s primary competitor in the AI data center GPU market.
AMD is positioning its MI400 and MI355X chips to challenge Nvidia not only on performance but also on price and energy efficiency. A company executive emphasized that AMD’s chips will cost less to operate due to lower power consumption, with “aggressive” pricing strategies to undercut Nvidia.
Despite Nvidia’s stronghold supported by proprietary CUDA software, AMD’s MI355X chip has demonstrated superior performance in some areas. CEO Lisa Su highlighted that open software frameworks have improved significantly, allowing AMD’s hardware to compete effectively.
AMD shares have remained flat so far in 2025, indicating Wall Street’s cautious view of AMD as a major threat to Nvidia’s dominance.
Andrew Dieckmann, AMD’s general manager for data center GPUs, pointed out that the company offers meaningful cost savings on both acquisition and operation. He cited double-digit percentage advantages layered on top of AMD’s competitive performance.
The AI chip market is rapidly growing, with major cloud companies and countries planning to invest hundreds of billions in data center GPU clusters to support AI development. Analysts expect the AI chip market to exceed $500 billion by 2028, though Nvidia currently controls over 90% of this space.
AMD has acquired or invested in 25 AI-related companies in the past year, including ZT Systems, a server maker key to AMD’s rack-sized system technology. Lisa Su stressed the complexity of AI systems and the critical need for full-stack solutions.
Current Offerings and Customer Adoption
The most advanced AMD AI chip currently in production is the Instinct MI355X, which began shipping last month and will be available from cloud providers starting in the third quarter.
AMD’s chips have been adopted by seven of the top ten AI customers, including OpenAI, Tesla, xAI, and Cohere. Oracle plans to offer clusters with over 131,000 MI355X chips, while Meta is already using AMD clusters for running inference on its Llama model. Microsoft also uses AMD chips to power its Copilot AI features.
AMD does not disclose chip prices, as they are sold primarily through hardware partners like Dell and Super Micro Computer. The MI400 series will be paired with AMD’s CPUs and networking chips, integrated using the open-source UALink technology. This contrasts with Nvidia’s proprietary NVLink.
The MI355X chip delivers approximately 40% more AI output tokens per dollar compared to Nvidia’s offerings, thanks to lower power consumption.
Author’s Opinion
AMD’s latest AI chip launch and Helios rack system show a strategic commitment to disrupting Nvidia’s near-monopoly in the data center GPU market. By focusing on full-stack integration, cost efficiency, and open-source networking, AMD is appealing to customers frustrated with Nvidia’s dominance and proprietary systems. The involvement of OpenAI and other top AI players signals growing confidence in AMD’s technology. However, breaking Nvidia’s grip will require sustained innovation and marketing, especially as Nvidia continues to expand aggressively. AMD’s progress is encouraging, but the battle for AI hardware supremacy is just heating up.
Featured image credit: Fortune Brainstorm Tech via Flickr
For more stories like it, click the +Follow button at the top of this page to follow us.