Jensen Huang: Nvidia and TSMC Never Signed a Contract, Greatest Moat is CUDA Universal Programming GPU Installed Base

NVIDIA CEO Jensen Huang recently shared in an interview the secrets to the long-term success of TSMC (Taiwan Semiconductor Manufacturing Company), NVIDIA’s key partner, and the deep bond between the two companies. He pointed out that TSMC’s technology is only the surface visible to most people; the true foundation of TSMC’s position lies in quality and trust. Huang also emphasized that NVIDIA’s greatest moat comes from its GPU installed base. CUDA, a universal programming platform, is widely embraced by developers worldwide. Millions of developers have confidence in NVIDIA’s technology, making it the preferred platform for software development, which solidifies NVIDIA’s market position.

TSMC’s Leading Position Stems from Quality and Customer Trust

TSMC’s success in the semiconductor industry is often attributed to advanced process technology, but Huang believes there is a misconception about this. He stated that the real difference lies in TSMC’s ability to coordinate across global operations—handling the changing needs of hundreds of customers, balancing production volume, yield, and cost control. TSMC values its commitments to customers, giving them confidence to entrust their orders for manufacturing.

NVIDIA Has Never Signed a Contract with TSMC

Huang emphasized that TSMC not only possesses cutting-edge technologies like transistors, packaging, and silicon photonics but also has a corporate culture that “takes commitments seriously.” This culture ensures customers believe wafers will be delivered on time, stabilizing the company’s operations. Huang highlighted that over the 30 years of collaboration, many billion-dollar deals between NVIDIA and TSMC have been completed without formal contracts. This longstanding trust is what makes TSMC irreplaceable.

CUDA Universal Programming and the GPU Installed Base as NVIDIA’s Biggest Moat

When host Lex Fridman asked what NVIDIA’s greatest moat is, Huang responded that the most important asset is not just chip technology but the GPU installed base on the compute platform.

Millions of developers worldwide use CUDA for general-purpose programming, creating a powerful cycle of trust: developers believe NVIDIA will continue to maintain the platform, so they prioritize releasing software on CUDA to reach the broadest cloud and enterprise users. The large user base, combined with NVIDIA’s annual updates to complex systems, forms a moat that is difficult for competitors to cross.

Another advantage for NVIDIA is its vertical integration capability. Huang explained that NVIDIA has achieved vertical integration of complex systems and extended its architecture across various industries. From cloud giants like Google Cloud and Amazon AWS to biotech R&D supercomputers like Eli Lilly, and even to automotive, robotics, satellites, and edge computing with base stations—NVIDIA’s architecture is everywhere. This extensive coverage means that general-purpose computing architectures can solve problems across different industries, further strengthening its market position. For developers, supporting CUDA means their applications can automatically adapt to computers and industrial equipment worldwide. The universal scale of CUDA is a key pillar of NVIDIA’s continued leadership.

Huang stated that with the evolution of artificial intelligence, computing units have advanced from GPUs to entire computers, clusters, and even gigawatt-scale infrastructure. Launching an AI factory requires thousands of network and hardware engineers working together, involving complex power grid connections and cooling systems. He looks forward to NVIDIA building planetary-scale computing capabilities in the future.

This article, “Jensen Huang: NVIDIA and TSMC Have Never Signed a Contract, and the Largest Moat is CUDA’s GPU Installed Base,” first appeared on Chain News ABMedia.