Intel and AMD Introduce New AI Chips to Compete with Nvidia's H100

Technology companies Intel and AMD are making significant investments in the field of artificial intelligence (AI) by unveiling their latest AI processors. The two companies aim to challenge the dominance of Nvidia's H100, which currently holds the largest share of the expanding AI market. The new processors are designed to accelerate the complex computational tasks required in AI applications, such as machine learning and natural language processing. With these advancements, Intel and AMD hope to capture a larger portion of the lucrative AI market and meet the growing demand for AI technology across various industries.

Intel has recently unveiled its latest AI chip, called Gaudi3, designed to cater to the needs of generative AI software. The chip is expected to be launched later this year and will compete with other leading AI chips, including Nvidia's H100 flagship. Although specific details about Gaudi3 are still scarce, it is believed that the chip will offer improved performance and efficiency, and will be capable of handling complex AI workloads with ease. This makes it a promising addition to Intel's growing portfolio of AI solutions, which are increasingly becoming a key part of the company's strategic focus.

Furthermore, Intel introduced new Central Processing Units (CPUs) equipped with a Neural Processing Unit (NPU) designed for "low-power AI acceleration and off-loading of CPU/GPU tasks." The NPU is intended to handle specific AI-powered functions such as the local execution of tasks like background blur, eye tracking, and picture framing.

AMD has introduced the MI300X chip, branding it as "the most advanced AI accelerator in the industry." The company anticipates that the MI300 chips will become its fastest product to reach a sales milestone of USD 1 billion, to achieve this feat by approximately mid-2024.

“On raw specs, MI300X dominates H100 with 30% more FP8 FLOPS, 60% more memory bandwidth, and more than 2x the memory capacity,” wrote Dylan Patel and Daniel Nishball at semiconductor consulting firm SemiAnalysis. Patel and Nishball expect AMD’s performance to keep growing against the H100.

Nvidia has released a modified version of its advanced gaming chip, called the "GeForce RTX 4090 D," in response to US export controls specifically aimed at China. The new chip is now available for purchase in China.

In October last year, the US government released new export rules that prevented Nvidia, a leading AI chip manufacturer, from selling its modified AI chips that were designed to comply with the previous export rules, namely A800 and H800. These new rules affected the sale of Nvidia's top-of-the-line gaming chip, the RTX 4090, which is popular among gamers.

"The GeForce RTX 4090 D has been designed to fully comply with US government export controls. While developing this product, we extensively engaged with the U.S. government,” says a Nvidia spokesperson.

According to a recent report by Reuters, Nvidia, the American multinational technology company, currently dominates more than 90% of China's AI chip market, which is estimated to be worth around $7 billion. However, due to the aggressive curbs imposed by the US on AI chips, Chinese companies like Huawei are now looking to invest in designing and manufacturing their own AI chips, thereby reducing their dependence on American companies. This move by Chinese companies is expected to have a significant impact on the AI chip industry in the coming years.