Speech by Huang Renxun at the National Taiwan University graduation ceremony: How did NVIDIA, after experiencing multiple failures, grow into a trillion-dollar AI giant?

Huang Renxun's speech at National Taiwan University's graduation ceremony focused on NVIDIA's success story despite multiple setbacks, leading it to become a trillion-dollar AI company.

Original: “NVIDIA unveils super chip, AI has reshaped computer industry” by Huang Renxun

Translated by: Katie

Edited by: Wendao

On May 29, NVIDIA released its latest product, the “NVIDIA DGX™ Supercomputer,” which provides hardware support for generative AI language applications, recommendation systems, and data analysis workloads for giant models.

• Three major product matrices, exploring X2E in multiple scenarios, Salad Ventures builds Web3 economic flywheel effect.
• BKEX: Suspends withdrawals due to cooperating with the police to investigate funds suspected of money laundering.
• Analysis of a whale who spent $385,000 buying new tokens within 72 hours

The core component of this computer is the NVIDIA Grace Hopper super chip, which has 20 billion transistors and integrates 72-core Crace CPUs, Hopper GPUs, and other components in the same package. This chip is said to significantly reduce the cost and energy consumption of training large language model (LLM).

The new product was unveiled three days after NVIDIA’s quarterly earnings report, which reached $7.19 billion in Q1 2023. After this “epic” financial report was released, NVIDIA’s market value skyrocketed by $200 billion in one day, and its total market value of $963.185 billion almost made it a member of the “trillion-dollar club.” Once it breaks through, it will become the seventh company after Meta, Amazon, Microsoft, Alphabet, Apple, and Tesla to achieve this milestone.

Undoubtedly, this chip giant, which provides “brain” to computers, is the most indispensable part of the AI era, and it is seen as the “biggest winner” by the outside world.

NVIDIA, which seems to have caught the windfall, has been solving the problem of computing speed since its inception in the 1990s by focusing on the PC gaming market. After becoming the leader in programmable graphics processing technology worldwide, NVIDIA saw the potential of deep learning technology in 2014, providing a Cuda solution that supports GPU computing, which laid the foundation for the later AI explosion.

The “winner” has also experienced setbacks. On May 27, NVIDIA CEO Huang Renxun shared three stories that almost destroyed NVIDIA during his speech at the graduation ceremony of National Taiwan University, encouraging graduates who are about to leave the ivory tower. These three stories happen to cover the three key products that can represent the company, hiding the details of why NVIDIA can win.

Throughout the 20-minute English speech, he did not forget to express his views on the current AI era: AI has fundamentally reshaped computers, and this will be a rebirth of the computer industry. He reminded young people to prepare themselves mentally, as the new era has given them new opportunities. “1984 was a perfect year to graduate, and I predict that 2023 will be the same,” he said.

“The Day of the Metaverse Explosion” compiles his speech:
The first time I came to National Taiwan University was over a decade ago. Dr. Chen (Editor’s note: President of National Taiwan University Chen Wen-tsuen) invited me to visit his computational physics laboratory. At that time, Dr. Chen’s son was also in Silicon Valley. He learned that NVIDIA had just developed CUDA technology and suggested that his father apply it to quantum physics simulation experiments.

When I arrived, Dr. Chen closed the door and showed me what he had done: the whole room was full of NVIDIA GeForce gaming graphics cards, connected to an open PC motherboard, with electric fans for cooling in the hallway. Dr. Chen used gaming graphics cards to create a very large supercomputer, and what he did can be said to be an early application case of NVIDIA chips.

Dr. Chen was very proud and said to me, “Mr. Huang, because of your work, I can complete my career in my lifetime.” His words still deeply moved me and perfectly interpreted the value of our company, which is to help the Einsteins and Da Vincis of our era complete their careers.

I am very happy to return to the campus of National Taiwan University and give a graduation speech for you. When I graduated from Oregon State University, the world was much simpler. At that time, TV was not yet a flat panel, there was no wireless TV, no music TV station, and the words “mobile” and “phone” were not yet linked together. It was not until 1994, with the birth of IBM’s personal computer PC and Apple’s Macintosh, that the PC revolution was unveiled and the chip and software industry we know today was created.

Today, you will enter a more complex world, facing challenges in geopolitics, society, and the environment. In the encirclement of technology, people have been endlessly connected to each other, forming a digital world parallel to the real world.

40 years ago, the computer industry drove the creation of the entire PC industry. Today, we have invented artificial intelligence (AI). Like software that can automatically drive cars or judge X-ray films, AI opens a door. It allows computers to perform tasks automatically and has created the largest industry in the world, worth tens of trillions of dollars. From the medical, financial services, transportation, to manufacturing industries, AI has created huge opportunities. Fast-moving companies can use AI to improve their status, and companies that do not do so will be eliminated.

During the period of rapid computer development in the past, new industries were created, and AI will also create job opportunities that did not exist before, such as data engineers, AI factory operators, and AI security workers. It is certain that AI will change all professions, and it will significantly enhance the work performance of designers, artists, marketers, and manufacturing planners.

Just as every previous generation has embraced emerging technologies to succeed, every company and individual in this room must learn to make good use of AI and achieve amazing results with its help.

We are at the beginning of a new technological era, just as people have experienced PC, the internet, mobile devices, and cloud storage before. But comparatively, the impact of AI is more profound. Because all levels of calculation have been redefined. From the way software is written to the way operations are performed, AI has changed the computer industry from top to bottom.

On various levels, this is a rebirth of the computer industry and a rare opportunity for businesses. You are the foundation of the computer industry. In the next ten years, new and faster AI computers will replace the traditional computer industry with a value exceeding trillions of dollars.

My life journey started 40 years before yours. 1984 was a perfect year for graduates, and I predict that 2023 will be the same. What can I tell you before you start your life journey? Today is the most successful day of your current life, as you are graduating from National Taiwan University. I was also very successful until I founded NVIDIA.

At NVIDIA, I experienced many failures, including very significant ones. These failures were embarrassing and shameful. Many failures almost destroyed us. Let me tell you three NVIDIA stories that define NVIDIA today.

Debugging for Help: Riva128 Saves NVIDIA

We founded NVIDIA to create accelerated computing. The first application was 3D graphics for computer games. We invented a non-traditional 3D technique called texture mapping. This method greatly reduced costs and won a contract with Sega to build game consoles. Sega also provided us with funds.

After a year of development, we realized our architecture was using the wrong strategy and was technically inadequate. Meanwhile, Microsoft was about to release Direct3D for Windows95, which was based on reverse texture mapping and triangles. This meant that if we completed Sega’s game console, we would be delivering an inferior technology that was incompatible with Windows. But if we didn’t complete the contract, we would go bankrupt.

I contacted Sega’s CEO, explained that our method of invention was wrong, and we couldn’t complete the contract, and suggested that Sega find another partner. But I also explained that we needed full payment from Sega, or NVIDIA would be bankrupt. I was embarrassed to ask for help from Sega’s CEO.

To my surprise, Sega’s CEO understood us and generously gave us six months of funding. With this six months of funding, we launched the Riva128. At the moment when we were almost bankrupt, the Riva128 shocked the early 3D market, made us famous, and saved the company.

NVIDIA Riva128 graphics card chip

We chose to face our mistakes honestly and humbly seek help from others, and this saved NVIDIA. These qualities are the most difficult for intelligent and successful young people like you.

Persisting in GPU Accelerated Computing CUDA Becomes Standard

In 2007, we announced CUDA, which supported GPU-accelerated computing. Our goal was to make CUDA a programming model from scientific computing and physics simulation to image processing. Creating a new computing model is quite difficult, and few have done it in history. CPU computing models have been the standard for 60 years.

For CUDA, developers needed to write applications to prove the benefits of GPU. But developers needed a large install base, and consumers needed to purchase new applications to install CUDA. Therefore, to solve the problem of which came first, the chicken or the egg, we used G-Force, which already had a large number of gamers, to establish a CUDA installation base.

NVIDIA CUDA defines a unified computing architecture

However, the development cost of Cuda was very high, and Nvidia’s profits suffered a huge blow, with a market value of around 1 billion US dollars (Nvidia’s current market value is close to 1 trillion US dollars) at one point. For many years, our performance was poor, and shareholders were skeptical of Cuda, hoping that we would focus on improving profitability. But we persisted, believing that the era of accelerated computing was coming.

We created a global AI developer conference called GTC and tirelessly promoted Cuda around the world. Then, applications began to appear, applications in scientific fields such as earthquake processing, CT reconstruction, molecular dynamics, particle physics, fluid dynamics, and image processing. We worked with every developer to write algorithms and achieved incredible acceleration.

In 2014, Alex trained on our GPU, triggering the big bang of AI. Fortunately, we realized the potential of deep learning and took risks to pursue it. Years later, the AI revolution began, and Nvidia became the driving engine. We invented CUDA for AI, and this journey forged our character, enduring pain and suffering, a necessary pain on the road to pursuing our vision.

Abandoning mobile chips to focus on AI computing and create new industries

In the mobile chip market, in 2010, Google aimed to make Android a mobile computer with outstanding graphics. The phone industry already had chip companies with expertise in modems, and Nvidia’s computing and graphics expertise made us an ideal partner for helping to build Android.

We immediately succeeded, and our business and stock prices soared. However, competition quickly flooded the market, with modem chip manufacturers learning how to build computing chips, while we were learning how to build modems.

The global market is huge, and we could certainly compete for market share, but we made the difficult decision to withdraw from the mobile chip market. Because Nvidia’s mission is to build computers that can solve problems that ordinary computers cannot solve, we should be fully committed to realizing our vision and making unique contributions.

Our strategic withdrawal paid off. By leaving the mobile phone market, we created a new industry. We wanted to create a new type of robot computer, with a neural network processor and a secure architecture, that could run AI algorithms. At the time, this was still a zero-sized market, and today we have billions of dollars in automotive and robotics businesses, creating a new industry.

It is often not easy to give up, especially for the smartest and most successful people like you. But retreating, sacrificing, and deciding what to give up is the core of success.

Class of 2023, you are about to enter a world that is undergoing tremendous change, just as I encountered the PC and chip revolution when I graduated and you are now at the beginning of the AI era. Every industry will be renovated and reborn. Your thoughts need to be prepared for this. In the past 40 years, we have created personal computers, the Internet, mobile devices, cloud technology, and the current AI era, and what will you create?

Whatever you create, make sure to run like us. Remember, run. Whether it’s running for food or not being taken as food by others. You often can’t know which situation you’re in, but no matter what, keep running.

On your future journey, please take my lesson: humbly face failure, be able to realize your mistakes, and have the courage to seek help. You will bear the pain and hardship required to achieve your dreams, make sacrifices, and devote yourself to a meaningful life and career.

Congratulations to each of you in the Class of 2023. Go for it!

We will continue to update Blocking; if you have any questions or suggestions, please contact us!