The invisible bottleneck of Artificial Intelligence
By Javier Surasky
.png)
AI is a field marked by constant power struggles. This series introduces its main players.
A growing share of the debate around artificial intelligenceis about hardware. And Nvidia sits at the center of that discussion. Nvidia is
often described as a successful chip company, but its power goes much further:
its GPUs have become the de facto standard for training and running advanced AI
models.
What makes it especially hard to replace is not just its
chips, but the ecosystem around them, especially CUDA, its software platform,
which over the years has created a base of developers, tools, and technical
dependencies that make switching away enormously costly.
That position also makes Nvidia decisive in shaping national
strategies: when countries talk about developing their own AI capabilities,
they know they still depend on highly concentrated technology chains, and
Nvidia holds a key place in the critical chain that keeps the entire AI
computing system running.
But Nvidia is not autonomous. Its power also depends on an
extremely sensitive supply chain, with TSMC as its critical partner, shaped by
export restrictions, technological rivalry, and tensions among international
actors.
That is why Nvidia should be understood not only as a
dominant company but as a point of convergence between computing capacity,
critical supply chains, and global competition for technological leadership.
To understand Nvidia is to understand the material backbone
that supports artificial intelligence.
Basic Facts
NVIDIA is a U.S. technology company founded in California on April 5, 1993 by Jensen Huang, Chris Malachowsky and Curtis Priem, Its headquarters currently are in Santa Clara (California), but it have offices, research centers, and industrial partnerships in numerous countries, among them Taiwan, China, India, Israel, Japan, Singapore, Germany, the United Kingdom, France, and Canada.
Jensen Huang has held the CEO position since the company was founded in 1993, which is really unusual in Silicon Valley, and he is also the President of the company, leading approximately 42,000 employees (fiscal year 2026). He was born in Taiwan in 1963 and emigrated to the United States, where he became an electrical engineer. He is considered one of the central architects of the current AI boom.
- NVIDIA's main industry areas are AI chips, GPUs, data centers, video games, scientific computing, and autonomous vehicles.
- In 1999, NVIDIA introduced the modern concept of the GPU with the GeForce 256, and in 2006, it launched CUDA, a platform that made it possible to use GPUs for advanced computing and, later, for AI, a key to understanding the contemporary rise of deep learning.
- Its H100, A100, and Blackwell chips have become critical infrastructure for companies such as OpenAI, Microsoft, Meta, Amazon, and Google.
In #2, the map turns to another key player in this story: TSMC.