Founded in 1993 by brothers Tom and David Gardner, The Motley Fool aims to make the world smarter, happier, and richer. We assist millions of people in achieving financial freedom through our website, podcasts, books, newspaper column, radio show, and premium investing services. Nvidia's high-performance chips are driving the AI industry, but their expensive price and equally costly stock may raise concerns. On the other hand, AMD, often overlooked, is also making strides in AI with its powerful chips and the game-changing Xilinx merger. However, this stock is not exactly inexpensive either. While the future of AI-specific semiconductors looks promising, the overheated stock prices may need to cool down. Please note that this article presents opinions that may differ from The Motley Fool's Premium Investing Services. To access our top analyst recommendations, in-depth research, investing resources, and more, become a Motley Fool member today. Learn More. Nvidia (NVDA 1. 85%) is widely recognized as a leading provider of chips specifically designed for AI systems. Its chief rival in the computing sector, Advanced Micro Devices (AMD 1. 67%), receives less attention in the AI field. However, within the AI industry, how close is this two-horse race?It is widely known that OpenAI relied on Nvidia A100 Tensor Core chips for training its GPT-3 and GPT-4 AI engines. The GPT-3. 5 model, which powers the entry-level ChatGPT tool, was trained on the Selene supercomputer equipped with 4, 480 A100 chips. These A100 GPUs are high-performance and not easily found on store shelves. Online platforms offer used A100 cards at prices ranging from $10, 000 to $18, 000 per unit. The newer and faster H100 chip, which will train OpenAI's next-generation GPT system, is even scarcer and over twice as expensive. Therefore, the hardware used in a single AI-training system can have a significant impact on Nvidia's income statement. However, this is not the entire story. Supercomputers like Selene also rely on Nvidia's InfiniBand networking solutions, which came with the Mellanox acquisition in 2019, for extremely high-performance network connections. Additionally, specialized high-performance computing cards require support from more traditional server CPUs. For Selene, 8, 680 AMD EPYC processors fulfill this role. Thus, Nvidia's biggest AI success to date also involved thousands of AMD processors. This collaboration demonstrates how bitter rivals often work together to create systems with world-class performance for specific workloads. Depending on the system design, other Nvidia-powered supercomputers may work with Intel chips instead. Interestingly, two out of the three fastest supercomputers in the world currently use AMD Instinct MI250X chips rather than Nvidia A100 or H100 chips.

Similar to Nvidia, AMD has been producing AI-specific accelerator chips for several years. Furthermore, their $60 billion Xilinx acquisition has made them proficient in creating custom and reprogrammable chip designs to meet the unique requirements of high-end customers. AMD's AI system experts, including engineers from the Xilinx team, are actively pursuing large contracts for training and operating GPT-style language models. Hence, the competition for AI chip contracts may be more intense than it initially appears. While Nvidia has been in the spotlight, AMD is also attacking this opportunity with enthusiasm and expertise. Moreover, these companies sometimes collaborate as partners rather than rivals in the AI field. However, Nvidia has yet to fully capitalize on this business opportunity. In the first quarter, both Nvidia and AMD saw a year-over-year decline in their total revenues. The upcoming second-quarter reports may present a different story. Until then, the booming AI sales remain more of a future growth prospect than a proven profitable idea for these two companies. Considering the current lofty prices, I am not rushing to buy Nvidia shares. Even AMD's lower valuation seems overinflated at the moment. Other more affordable and equally impressive investment options are available in the AI market. I recommend exploring those alternatives first. Anders Bylund holds positions in Intel and Nvidia. The Motley Fool has positions in and recommends Advanced Micro Devices and Nvidia. The Motley Fool recommends Intel and suggests long January 2023 $57. 50 calls on Intel and long January 2025 $45 calls on Intel. The Motley Fool has a disclosure policy. *Average returns of all recommendations since inception. Cost basis and return based on previous market day close. Top Economist Ed Yardeni believes generative AI will unleash a "Roaring '20s" for the stock market. Here's how investors can profit. Should we be fearful of AI stocks?Check out these four excellent growth stocks in the wake of the Nasdaq Bear Market Dip. Invest better with The Motley Fool. Get stock recommendations, portfolio guidance, and more from The Motley Fool's premium services.