Nvidia, a U.S. semiconductor company, is a primary beneficiary of the AI boom. Since early last year, its stock price has surged by about 450%, and its market capitalization has exceeded $2 trillion. Nvidia has now established itself as the third highest-valued company in the U.S., following Microsoft and Apple.
Nvidia’s growth is steep. Its sales for the fourth quarter of the 2024 fiscal year (from November last year to January this year) recorded $22.1 billion, a 265% increase compared to the previous year’s period. In the industry, Nvidia, which holds more than 95% of the AI semiconductor market, is expected to continue its upward trend for some time. The Economist recently focused on why Nvidia’s chips dominate the AI semiconductor market.
The first Nvidia Graphics Processing Unit (GPU) was designed for video games. To implement graphic screens, numerous pixels had to be rendered simultaneously, a process suitable for a GPU. Nvidia’s high-performance chips currently occupy about four-fifths of the gaming GPU market.
High-performance GPUs, with thousands of cores, are convenient for processing large amounts of data simultaneously. For this reason, they are currently used for various purposes, such as AI model training, virtual asset (cryptocurrency) mining, and autonomous driving cars.
Nvidia began focusing on producing AI semiconductors after realizing that its GPUs were optimized for AI model training. Over the last decade, the data processing speed of Nvidia chips has improved nearly a thousand-fold.
In addition to product development, “networking” is another factor behind Nvidia’s growth. Data centers typically require thousands of GPUs. In response, Nvidia acquired computer networking supplier Mellanox for $7 billion in 2019. Based on Mellanox’s networking technology, Nvidia has built a high-efficiency GPU network that is difficult for competitors to follow.
Nvidia’s other strength is its software platform, CUDA. Introduced by Nvidia in 2006, CUDA is a GPU programming language widely used by developers to develop AI apps and is now established as an industry-standard platform.
The AI semiconductor market is expected to grow to an annual size of $400 billion by 2027. With Nvidia leading the way and dominating market attention, other tech companies are also jumping into development. Amazon and Alphabet are making AI chips for data centers, and U.S. semiconductor company AMD unveiled a chip in December last year that is twice as powerful as Nvidia’s.
However, it seems challenging to catch up to Nvidia with just the development of high-performance semiconductors. The Economist pointed out, “Nvidia occupies a unique position because it provides the best semiconductors, networking kits, and software. Competitors trying to replace Nvidia must win in all three areas. That will be a difficult task.”
Most Commented