Last week on the podcast, I came across some quite interesting remarks made by NVIDIA's CEO. They discussed the competitive advantage of AI chips, supply chain strategies, and sales in the Chinese market.

First, what struck me was how consistent this CEO's way of thinking is. His fundamental philosophy is simple: "Do everything necessary with full effort, and minimize unnecessary things." The reason NVIDIA is building an entire computing platform is because he believes that if they don't do it, no one else will. However, he has a different view when it comes to cloud services, thinking that this area should be left to other companies.

The discussion about the supply chain was fascinating. The NVIDIA CEO clearly explains why upstream partners (like TSMC, memory manufacturers, etc.) invest in them. It’s because NVIDIA can absorb their supply volume and has the ability to sell downstream. In other words, the larger the demand in the entire ecosystem, the more easily partners are willing to bet on upstream.

Regarding the competition with TPU, the NVIDIA CEO claims it’s “something different.” NVIDIA is not just a tensor processing unit but covers a broad range of acceleration calculations. Its strength lies in supporting all kinds of applications, from molecular dynamics and fluid dynamics to data processing.

He also talked about the value of the CUDA ecosystem. With hundreds of millions of GPUs deployed worldwide, it has become the safest choice for developers. When problems occur, it’s more likely to be a code issue rather than a hardware problem, and that trust is crucial.

However, the most debated part was about sales in the Chinese market. The NVIDIA CEO explicitly states that abandoning the market is a “loser’s mentality.” His logic is that China already has enormous computational capacity, and they will secure it through other means even without chips. Instead, he seems to believe that expanding the US’s technological stack globally is the way to maintain long-term leadership.

Regarding security concerns, he emphasizes the importance of maintaining dialogue and research exchanges. Equating AI with nuclear weapons is said to lead the issue in the wrong direction. Rather, the stance is that the US should continue to lead at the forefront and hold the most computational power to maintain its leadership.

This CEO’s perspective appears to be based on a broader strategy of how to build a global technology ecosystem, not just a simple business strategy. His statement that launching new architectures every year and being a reliable partner like a clock is NVIDIA’s greatest competitive advantage was impressive.

The architecture strategy is also interesting. The reason they don’t need to develop multiple different chip designs simultaneously is that testing everything in simulators has shown that their current direction is optimal. However, if the market structure changes significantly, they are flexible enough to change their strategy.

His answer to the question of what NVIDIA was doing before the AI era was also intriguing. The CEO is convinced that even without AI, NVIDIA would have become a very large company in the field of accelerated computing. In other words, the advent of deep learning is just a phenomenon riding on the fundamental demand for accelerated computation, and that demand itself remains unchanged.

Throughout this conversation, what I feel is the consistency of NVIDIA CEO’s thinking and the importance of a long-term perspective. He seems to judge based on a broader ecosystem view rather than short-term regulations or market pressures. As a result, the company is positioned not just as a chip manufacturer but as a core player in the global AI infrastructure.