[Chipformation] Nvidia Realizes $1 Trillion Market Cap Dream... World's No.1 GPU Manufacturer

Market Cap $1 Trillion 'Numbers' Are Scary, But
The Real Power Lies in Insight, Price Competitiveness, and Organizational Strength
AI Ecosystem's Success in 'Vertical Integration'

The United States-based Nvidia, the world's number one graphics processing unit (GPU) company, has emerged as the leading company in the artificial intelligence (AI) era. On the 30th of last month (local time), the company surpassed a market capitalization of $1 trillion (approximately 1,315 trillion KRW) for the first time in 30 years since its founding, becoming the first semiconductor company to do so. Jensen Huang, Nvidia's founder and CEO, predicted that "AI will be applied to all manufacturing industries."

Jensen Huang, Founder and CEO of NVIDIA.
Photo by Reuters Yonhap News

Nvidia's AI semiconductor market share stands at 92%. According to Article 6 of the Fair Trade Act, a business operator with a market share of over 50% is called a 'market-dominant business operator.' With 92%, it is no exaggeration to say that Nvidia monopolizes the global GPU market.

Its annual revenue was $27 billion (approximately 35.7 trillion KRW) as of last year, and its market capitalization was about $982.319 billion (approximately 1,288 trillion KRW) as of the closing price on the 1st (local time). Its stock price rose 103% over the past year. The market capitalization of 1,288 trillion KRW is 2.1 times that of TSMC (about 612 trillion KRW) and three times that of Samsung Electronics (about 423 trillion KRW).

Nvidia's presence cannot be fully explained by 'numbers' alone. Nvidia's strength lies in ▲ the 'insight' that GPUs would become the dominant AI semiconductor ▲ the 'price competitiveness' of providing affordable components specialized for large language models (LLMs) used in generative AI like ChatGPT ▲ and the 'organizational capability' encompassing data centers, software, and platforms. A GPU refers to a device in computers that quickly processes graphics and video to output to a monitor.

Nvidia released a product called 'GeForce 256' in 1999. At that time, as game graphics quality increased, it was judged that the existing CPU (central processing unit) alone was insufficient to handle graphics, so Nvidia preempted the market. The term GPU was used for the first time with the release of GeForce 256. GeForce was mocked as a component only for high-end gaming consoles.

Nvidia knew that GPUs were suitable for AI deep learning. Unlike CPUs, which process information sequentially, GPUs are designed for parallel processing, handling multiple pieces of information simultaneously, making them better suited for AI deep learning.

GPUs also offer high cost-effectiveness (performance per price). Although the unit price per chip is higher for GPUs than CPUs, their efficiency is superior. Nvidia's latest GPU, the 'H100,' costs around $30,000 (approximately 39.16 million KRW) each. This is more expensive than Intel's latest CPU product, the Xeon 4th Generation Scalable Processor (60 cores), which costs $17,000 (approximately 22.2 million KRW). At Computex 2023, a computer and IT expo held in Taipei, Taiwan, on the 29th of last month, Jensen Huang said, "With a $10 million investment, you can train one LLM with CPUs, but 44 with GPUs."

What is even more formidable is that Nvidia has vertically integrated the entire AI supply chain beyond just supplying affordable and high-quality components.

Nvidia is already the world's top GPU product supplier. It delivers high-end products such as the supercomputer 'DGX GH200' to big tech companies like Google, Amazon, and Meta. In this sense, it is a 'super secondary supplier.' Not only does it supply products, but it also entrusts the production of AI semiconductor products like the A100 and H100 to foundries such as TSMC and Samsung Electronics. In other words, it is both a super secondary supplier and a 'super primary supplier.'

Nvidia also dominates AI software and platform businesses. The GPU programming language 'CUDA,' created in 2006, operates only on Nvidia products. Based on powerful proprietary software technology, Nvidia has diversified its business into AI, autonomous driving, robotics, and more, increasing its market share to 92%. It is known that about 40,000 companies use Nvidia AI programs.