Core Insights - Advanced Micro Devices (AMD) has developed a significant AI accelerator business, but it remains small compared to Nvidia, which generated over $39 billion in revenue from its data center segment in the last quarter, while AMD's total AI accelerator revenue for 2024 is only $5 billion [1] - AMD's latest GPUs, the Instinct MI350X and MI355X, deliver four times the AI compute performance and 35 times the AI inferencing performance compared to previous models, but they do not surpass Nvidia's latest offerings [2] - AMD plans to release the MI400 series in 2026, claiming the MI400X will be ten times more powerful than the MI300X, featuring up to 432 GB of HBM4 memory and 19.6 TB/s memory bandwidth [4] - Nvidia is expected to release its Vera Rubin chips in late 2026, which will significantly enhance performance over its current generation [5] - AMD's MI400X is anticipated to outperform Nvidia's current products, but Nvidia's next-gen Rubin chips may ultimately prevail [6] - AMD is also set to launch a new rack-scale AI solution called Helios in 2026, which will include up to 72 MI400 GPUs and advanced server CPUs [7] - Despite AMD's advancements, Nvidia is likely to maintain its market leadership due to its powerful hardware and established CUDA software ecosystem [8] - The AI chip market is projected to reach $500 billion by 2028, and AMD's growth in AI chip revenue does not necessarily require it to surpass Nvidia [9] - With the upcoming MI400 series and Helios, AMD is positioned for substantial growth in AI-related revenue, although it is expected to remain in second place behind Nvidia [10]
Could AMD Finally Challenge Nvidia With Its MI400 AI Chips?