Advanced Micro Devices (AMD) unveiled its latest artificial intelligence processors on Monday and outlined its AI chip development strategy for the next two years, aiming to challenge industry leader Nvidia.
“AI is our top priority as a company, and we’ve focused all our development efforts to support that,” Su told reporters. “The annual release cycle is essential to meet the market’s demand for new products and capabilities. Each year, we aim to have the most competitive portfolio.”
AMD also announced the upcoming MI350 series, expected in 2025, featuring new chip architecture. The MI350 is anticipated to deliver 35 times better performance in inference—computing generative AI responses—compared to the current MI300 series. Additionally, AMD introduced the MI400 series, set for a 2026 release, based on an architecture called “Next.”
Investors, who have invested billions in AI-related technologies, are seeking long-term updates from chip manufacturers to gauge the sustainability of the booming AI market. AMD’s shares have more than doubled since the start of 2023, although this growth is modest compared to Nvidia’s seven-fold increase over the same period.
AMD aims to maintain a one-year product cycle for its AI chips, aligning with Nvidia’s strategy of releasing new AI chip families annually. Su stated in April that AMD expects AI chip sales to reach approximately $4 billion in 2024, an increase of $500 million from the previous estimate.
At the Computex event, AMD also announced that its latest generation of central processor units (CPUs) will likely be available in the second half of 2024. While businesses typically prioritize AI chips for data centers, some of AMD’s CPUs are used alongside graphics processor units (GPUs), though GPUs remain the primary focus.
AMD detailed the architecture for its new neural processing units (NPUs), designed for handling on-device AI tasks in AI PCs.
