Written by: Techub News Compilation

On August 9, 2024, Jensen Huang participated in an extensive oral history interview at the Computer History Museum. The most valuable aspect of this interview lies not in the repetition of the well-known "Nvidia Legend," but in the first-person narrative that weaves together how a technology company is formed, how it survives through mistakes, and how it transitions from a graphics chip company to the core of artificial intelligence infrastructure, creating a complete chain of thought.

To perceive Jensen Huang's career merely as one of "riding the wave of AI" misses his truly unique qualities. What emerges repeatedly in this oral history is not luck, but three abilities: betting on the big picture early, pivoting quickly after mistakes occur, and nurturing an ecosystem while technology and market are still taking shape. These three points almost fully explain why Nvidia was able to grow from a startup that faced failure multiple times to a company that defines a new generation of computing paradigms.

Early Character Shaped by Migration Experience

Jensen Huang was born on February 17, 1963, in Taipei, firstly living in Taiwan during his childhood before relocating to Thailand around the age of five with his family. Around 1973, due to the turmoil in Thailand, his parents decided to send him and his brother to the United States, first arriving in Tacoma, Washington, and later attending a boarding school in Kentucky before reuniting with their family in the Northwest of the United States.

This experience of migration across Taiwan, Thailand, and the United States is not packaged as an emotional story in the oral history, but it clearly had a profound impact on Jensen Huang's adaptability to environmental changes. A person who constantly moves into unfamiliar scenarios during childhood, re-establishes order, and learns how to establish oneself in different cultures is often more likely to accept the fact that "the world is inherently unstable" earlier. This capacity to bear uncertainty later became a foundational aspect of his leadership style.

He recalled that his interests during his youth were quite simple: school and sports. Swimming, soccer, tennis, and especially table tennis occupied much of his time; he even trained intensely in table tennis during his teenage years and participated in public competitions. While these pursuits seem unrelated to his later career in chips, they reveal an important characteristic: Jensen Huang became accustomed early on to pursuing excellence through repetitive training. Technical leaders are often envisioned as individuals with "eureka" moments, but his narrative aligns more closely with another model—one that relies on sustained investment to build a stable competitive advantage.

From Engineering Student to Chip Designer

Jensen Huang stated that he always performed well in science and math, largely influenced by his father, an engineer, who instilled in him a natural belief that engineering and technology were reliable paths for development. The true turning point in his life came from encountering computer equipment in high school and through his friends who were involved in the math club, science club, and computer club. It was in this environment that he decided to attend Oregon State University to study electrical engineering instead of pursuing a possible path in petroleum engineering.

During his university years, Jensen Huang not only solidified his major direction but also met his future wife, Lori. He mentioned that they became group members in the electrical engineering lab course and he pursued her under the pretext of "doing homework together on weekends." This detail is light-hearted yet reflects his distinct self-awareness: he knew that his most prominent capability was doing his homework well and executing tasks diligently.

After graduation, he worked at AMD and LSI Logic. AMD allowed him to dive into the core field of large-scale chip design, participating in CPU-related design and understanding engineering trade-offs such as performance, power consumption, and density; while LSI Logic helped him to see the deep connections between design tools, system design, and chip design. In his recollections, this phase was extremely crucial because he gradually realized that future competition would not only occur at the single-chip level but at a larger scale of "system, tool, software, and hardware collaborative design."

In other words, the "full-stack" thinking that Jensen Huang later repeatedly emphasized was not just a slogan emerging in the AI era but was taking shape back in the LSI Logic era. He witnessed how electronic design automation tools changed engineers' methods of designing complex systems, and as a result, he understood earlier that the computing industry would ultimately move towards a direction where hardware and software boundaries continuously merge.

The Birth of Nvidia: Not Just a Graphics Chip Company

When discussing the founding of Nvidia in 1993, Jensen Huang emphasized an important correction. The outside world often interprets the company's inception as simply "creating 3D graphics chips," but according to him, the three founders discussed earlier what kind of company could be created based on their accumulations in computer graphics, systems, and chip design; what business could sustain them in the short term and what type of company they could grow into in the long term.

This implies that Nvidia was not merely betting on "some kind of graphics card product" right from the start, but on a larger computing perspective: accelerating highly compute-intensive tasks. Jensen Huang clearly stated in the interview that at the time there were two routes for graphics processing: one that believed in general processors with software algorithms, and another that believed some workloads were better suited for dedicated acceleration pipelines; the three founders clearly favored the latter and believed that graphics processing was just the first realized scenario for accelerating computation.

More critically, Nvidia placed significant emphasis on software compatibility and platform continuity early on. In the oral history, he discussed the formation logic of a unified driver architecture: if each generation of hardware changes too significantly and the software investments cannot carry over, then the company cannot build genuine platform capabilities. This idea proved to be extremely important later. Many chip companies excel in creating one-time high-performance devices but struggle to establish a continuously evolving development ecosystem; Nvidia's ability to traverse multiple technology cycles can be attributed to recognizing "the iterability of hardware" and "the inheritable nature of software" as equally important considerations early on.

The First Life-and-Death Challenge: Making the Right Decision First

One passage in the oral history that is worth rereading is Jensen Huang's reflection on Nvidia's first major crisis. Early on, the company chose a 3D graphics solution that was later proven to be misaligned, employing techniques such as curves, forward texture mapping, and avoiding Z-buffering in an attempt to create appealing graphics under the extremely limited conditions of transistors and memory at the time. From an engineering perspective, this design was not without merit; the problem was that it was unfriendly to developers and did not align with the later mainstream development path in graphics processing.

Jensen Huang candidly admitted in the interview that they "almost completely went wrong." The genuinely correct direction turned out to be triangles, reverse texture mapping, and Z-buffering. This was not merely a local mistake but a directional error: the company had invested more than two years, faced numerous external competitors, and had important contracts with Sega. If they continued on the original path, there might have been less immediate financial pressure; however, in the long run, the company would have been completely misled.

From this crisis, Jensen Huang distilled a hardcore management principle: make the right decision first, then handle the consequences of that decision. He believes that many organizations end up in trouble not because they cannot see the correct direction, but because they are too afraid of "if we pivot, what will happen to contracts, what will happen to funding, what will happen to competition," and thus hesitate to act. What Nvidia ultimately chose was to first acknowledge the technical path was wrong, shift the company onto the right path, and then address contracts and financial issues one by one.

He then called Sega's CEO at the time, candidly explaining the company's situation, clarifying that if they continued with the original project, they would mislead their partner, and requested to terminate the contract while seeking funding support. Jensen Huang made it clear in the oral history that if it weren’t for their understanding and assistance, Nvidia might not have survived. But more importantly was the prior step: without first acknowledging the technical error and accomplishing the pivot, any financial support would merely delay failure.

This crisis also shaped Nvidia's later engineering culture. Due to the extreme time pressure, the team had to rebuild within about nine months, which was a decision that meant life or death, and during this process, they redefined methods for chip design, verification, and soft-hardware collaboration. He noted that the company even bought simulators that were supposed to be scrapped from a soon-to-be-defunct company, which helped establish a more mature soft-hardware development process. It can be said that Nvidia later became known for its execution not because it started with perfect order but because it was forced to forge its execution system on the line of life and death.

From Graphics to CUDA: A Longer Route

Many people today discuss Nvidia and view CUDA and AI as the turning point in the company’s fate. However, from Jensen Huang's narrative, this route did not appear suddenly but gradually evolved from "making the virtual world more real." Initially, Nvidia entered physical simulation from graphical rendering because merely having beautiful images remained static; what truly "brings the virtual world to life" are the abilities to simulate physics, particles, fluids, and a broader range of simulations.

As GPUs gradually became programmable, Nvidia began exploring how to leverage these parallel engines for tasks beyond graphics. The company first developed CG, which was closer to a C-language expression, and then advanced all the way to CUDA. The core behind this was not just the introduction of a programming tool but the continual effort to make GPUs "sufficiently programmable" without losing the acceleration advantages relative to general-purpose CPUs.

What truly mattered during this phase wasn't the technical terms but Nvidia's ongoing pursuit of "expressive power." Jensen Huang repeatedly emphasized in the interview that the company was not aiming to turn GPUs into another type of CPU, but was carefully navigating that boundary: enabling developers to express more complex computing tasks while maintaining high acceleration and energy efficiency. If this boundary is well managed, GPUs will become a new computing platform; if poorly managed, they either become cumbersome dedicated devices or lose their acceleration value.

Ecosystem Precedes Market: How Nvidia "Built the Road"

One aspect that Jensen Huang is most proud of regarding Nvidia is not the leading performance of any single generation of products, but the company's proficiency in simultaneously creating both technology and market. The early 3D PC gaming market had not truly formed, Windows 95 had not yet arrived, and consumer-level 3D games could almost be said to start from scratch; Nvidia not only had to make chips but also promote development tools, support developers, foster the software ecosystem, and cultivate player awareness, essentially growing alongside the entire market.

Later, the entrance of GPUs into scientific computing followed the same logic. In the interview, it was mentioned that seismic data processing in oil exploration was one of the early applications that validated GPU potential; Schlumberger's team saw the significant acceleration GPUs could bring to related calculations, which boosted Nvidia's confidence in the direction of scientific computing. Subsequently, Oak Ridge National Laboratory adopted Nvidia GPUs to build a new generation of supercomputers, further unlocking a wide range of research scenarios from molecular dynamics to climate science.

What is most difficult for outsiders to understand during this process is that "the market isn't naturally there waiting for you to occupy." In the field of accelerated computing, without appropriate algorithms, toolchains, developer education, and industry models, the market itself will not emerge automatically. Jensen Huang described this as simultaneously addressing the "chicken and egg" problem between "technology" and "market," and the reason Nvidia could later occupy a unique position in the AI era is precisely that it learned this set of tactics earlier than most hardware companies.

AI is Not a Trend but a Paradigm Shift in Computing

In the latter part of the interview, Jensen Huang described AI as an "extremely fundamental" transformation in computing. He believes this fundamental change is expressed at least in three layers: the way of computing has changed, the way software is developed has changed, and ultimately the applications that can be built have also changed. In his understanding, a large amount of computation in the future will be completed by parallel systems such as GPUs; at the same time, software will no longer consist solely of human-written rules but increasingly gain capabilities through data-driven approaches and model training.

This judgment, rather than merely catering to the AI hype, is more an organic extension of Nvidia's long-term trajectory. If the core work of the company over the past three decades has been to establish a platform capable of supporting accelerated computing, then as deep learning pushes "massive parallel linear algebra" and "training-based software generation" to the industry's forefront, Nvidia happens to be in the most advantageous position.

Moreover, it is noteworthy that Jensen Huang did not frame AI merely as an optimistic narrative. He explicitly mentioned that the more powerful the AI capabilities become, the more important issues such as safety, governance, data screening, synthetic data, and guardrail systems become, and these problems themselves also require AI for solutions. In other words, from his perspective, AI is not a singular product but a whole set of continuously evolving infrastructure systems that include model capabilities, training data, toolchains, governance mechanisms, and real-world application scenarios.

Leadership Style: Switching Between Future and Present

If the technological path explains "what Nvidia does," then Jensen Huang's description of the CEO role during the interview explains how he leads the team to accomplish these things. He believes that the CEO has an irreplaceable responsibility for the company’s strategy: not all ideas must be generated by the CEO, but ultimately the CEO must be responsible for establishing direction, organizing resources, and driving execution.

Jensen Huang summarizes his management style as: working simultaneously in both the future and the present, and being highly personally involved. He does not spend much time in daily operational meetings; rather, he is more like a central figure who can "float" between different teams in research, engineering, and marketing, going wherever connection and promotion are needed most. This approach is fundamentally not easy, as it requires leaders to see long-term visions while also understanding short-term obstacles and continuously adjust the organization to adapt strategic plans to external changes.

He even mentioned that Nvidia does not operate on fixed annual plans and five-year plans, but the company operates more like a continuously adjusting strategy under the premise of an unchanged long-term vision. This aligns very well with the realities of high-tech industries: when industry changes are fast enough, rigid plans may not be reliable, and what truly matters is maintaining a clear direction and a continuous ability to make corrections.

The "Moderate Ignorance" of Entrepreneurs

During the entire interview, one of Jensen Huang's most shareable judgments might be his description of the "superpower" of entrepreneurs: a psychological state characterized by ignorance, flippancy, and excessive confidence. It sounds like a joke, yet it is very real. For if entrepreneurs completely understood how difficult the problems were from the onset, many companies might never have come into existence.

He repeatedly mentioned that his way of thinking is centered around "how difficult can this be." Whether starting a company, entering deep learning, working on robots, or venturing into autonomous driving, he always took an almost reckless approach initially, and then relied on willpower, the team, and adaptability to push through as reality struck back. This assertion is not advocating blind optimism, but rather reminding later generations that significant innovations often cannot begin when all information is fully available and all risks are controllable.

Ultimately, what brings the company to completion is not merely that initial "ignorance-fueled courage," but the capability to endure setbacks, correct paths, organize teams, and continue moving forward after the harsh realities become thoroughly apparent. From this perspective, Jensen Huang's oral history is not a victorious legend but more akin to a practical note on how technological entrepreneurship navigates high uncertainty.

Conclusion: What are Nvidia's Underlying Capabilities?

It can be observed that what Jensen Huang truly wants to convey is not "a specific product succeeded," but that Nvidia has developed a few deeper capabilities: first, the ability to judge long-term technological directions; second, the ability to pivot swiftly in case of directional error; third, the capability to organize hardware, software, tools, and developer ecosystems together; and fourth, the ability to lay down infrastructure before the market fully matures.

Therefore, Nvidia's rise cannot be simply summed up as "hitting the AI jackpot." More accurately, it has built platforms around accelerated computing, accumulated ecosystems, and trained organizations over the long term, ultimately leading to a concentrated realization in the AI era. Jensen Huang’s personal traits do not only include "foresight," but also integrate foresight, strong execution, extreme honesty, and long-term patience for complex systems.

The most important insight may lie here: great technology companies do not win the future through a single correct prediction, but rather through repeatedly making more correct decisions amid chaos and being willing to bear the consequences of those decisions.