As the biggest microprocessing companies lean heavily into artificial intelligence (AI), AMD is joining in with a series of new chips made for AI applications, it announced. The move comes as competitor NVIDIA pivots from a focus on consumer graphics cards to AI cards that are designed for data centers. AMD is bringing these AI chips to market as both mobile processors and data center applications.
The new Ryzen 8040 line will be the next generation of AMD laptop processors and have an upgraded neural processing unit (NPU) that is designed using AMD’s XDNA architecture. This makes it particularly improved for AI uses, which is the heart of AMD’s announcement. There were nine laptop processors announced in the 8040 series, with the highest configuration sporting eight cores and 16 threads. That chip can be boosted to up to 5.2 GHz, and has a TDP of 45W. In total, seven of the nine chip configurations now have an NPU.
On the industry side, AMD is debuting its AMD Instinct MI300 series accelerators. These new chips have increased memory bandwidth and efficiency gains, making them great for use with AI. Large AI models require a lot of memory bandwidth, and the AMD Instinct MI300 platform offers 1.5x memory capacity compared to last-generation chips as a result. AMD claims this is the “highest performing accelerator in the world,” and says it performs comparably to Nvidia’s H100 platform.
“LLMs continue to increase in size and complexity, requiring massive amounts of memory and compute,” AMD CEO Lisa Su said, as reported by The Verge. “And we know the availability of GPUs is the single most important driver of AI adoption.”
Microsoft is already putting AMD’s AI chips in VMs
AMD also announced a few partnerships that will see the AMD Instinct MI300 series debut in AI applications. Specifically, the MI300A APU is intended for data center deployment, which Meta has already committed to doing. Additionally, Microsoft will use the MI300X chip in its Azure virtual machines. AMD has also worked with OEMs, like Dell and HP, who have made servers using the MI300 platform.
Large chipmakers are finding that there is a lot of financial incentive to develop chips made for AI. The technology is developing faster than some systems and manufacturers can keep up, so demand is high. Plus, from big companies to end users, plenty of people are interested in AI. That’s why AMD is releasing chips for laptops and data centers with AI capabilities, instead of just focusing on enterprise deployment.
However, AMD’s Ryzen 8040 series looks very similar to last-generation chips outside of the new NPU. Something to watch will be how much better the Ryzen 8040 will be for uses other than AI.