Nvidia Huang Renxun gave a speech at National Taiwan University, what message did he convey?

Everyone will have their own AI device in the future.

Written by: Alpha Rabbit

On May 27, Nvidia CEO Jensen Huang (Jensen Huang) delivered a speech at the graduation ceremony of National Taiwan University. After studying Jensen Huang's speech, the author interpreted the key points of Jensen's speech (that is, the underlined part of the text) for everyone For reference, comments from experts in related fields are also welcome.

CUDA (Compute Unified Device Architecture: Unified Computing Architecture) Lecture Full Text & Notes

It is a great pleasure to come back to NTU again and be your guest speaker.

To graduates:

The age you live in is very complicated, but it is also your opportunity

When I graduated from Oregon State University, the world was simpler. The TV is still super big, TVB and MTV have not yet appeared, and mobile phones do not exist. **In 1994, IBM PC and Apple's MAC Macintosh started the personal computer revolution, followed by the development of chips and computing programs. **

The world you face today is more complex, with geopolitical, social and environmental changes and challenges, all of which are shrouded in technology.

(Alpha Rabbit Note: Jensen said that, in fact, it is mainly based on two things. First, since 1994, Microsoft stopped the development and sales of DOS and turned to Windows in an all-round way. This is the most important step in the popularization of PC. Second, As a complete PC, IBM is in an absolute leading position. It is equipped with Windows 3.1, and has continuously knocked out the market share of Apple's MAC until today. Third, Apple's original machine was called Power MAC, but in 1995 Windows 95 Worldwide, Apple almost completely lost PC market share)

We are in an always-connected and immersive data world that exists parallel to the real world.

40 years ago, when the computer industry created the home PC and continued to research AI technology, our algorithms drove cars, or studied X-ray images. AI has opened the door to computer automation serving some of the world's largest mega-industries: health care, financial services, transportation, and manufacturing.

AI has brought us great opportunities. Agile companies will use AI technology to enhance their competitiveness, while companies that fail to make good use of AI will face decline. Many entrepreneurs, including many here today, will create new companies in the future.

(Alpha Rabbit Note: Why did Jensen say that? Because these high-achieving National Taiwan University students listening to speeches are the talents that Nvidia wants in the future. He is looking for young people)

Just as each past computer age has created new industries, AI has also created new jobs that did not exist before, such as: data engineers, prompt engineers, AI factory operators, and AI security engineers, which have never existed before existed.

Automating jobs will eliminate some jobs, and AI will undoubtedly reshape all jobs, dramatically enhancing the performance of designers, artists, sales and manufacturing planners. Like every generation before you, embrace technology to succeed. Every company and you must learn to take advantage of AI and do amazing things with the help of AI.

Some people worry that AI may take your job, some people may worry that AI develops self-will. We are at the beginning of a new field, just like the personal computer, the Internet, mobile devices and cloud technology.

(Alpha Rabbit’s Note: Jensen’s meaning is actually——this is the inevitability of history, everyone should embrace change and the future, hit the point, don’t walk, run)

**But the impact of AI is more fundamental, every level of computing will be rewritten. It changes the way we write software, how we execute software. **

From all perspectives, this is an opportunity for the regeneration of the computer industry. You are the cornerstone of this industry. Over the next decade, our industry will replace trillions of dollars worth of traditional computers with new AI computers.

(* Alpha Rabbit Note: Nvidia is Intel/AMD in the next 10 years. From a technical point of view, Nvidia will create a true Turing-complete computing device and continue to consolidate its first-mover advantage, that is, Said that everyone will have their own AI device in the future, the same goal as Gates and Intel in the early days. The goal of Microsoft and Intel at the time was-everyone has a Windows computer*)

My journey began 40 years before yours, 1984 was a perfect graduation year and I predict 2023 will be as well.

what can i tell you Today is the most successful day for you so far. You have graduated from National Taiwan University. I have also been successful. Before I started Nvidia. **I have also experienced failure, and it is a big failure. It is humiliating and embarrassing to say, and it almost leads us to destruction. **

Now let me tell you 3 stories that shaped Nvidia what it is today.

** Facing mistakes honestly and asking for help with humility are the hardest things for smart and successful people to learn. **We started Nvidia to create accelerated computing technology. For our first application, 3D graphics for PC games, we invented an unconventional forward texturing technique that was relatively inexpensive. We won a contract with SEGA to build a console. This attracts game developers to use our platform to develop games and provide our company with funding.

*(Alpha Rabbit Note: Why did Jensen create accelerated computing technology back then? In fact, it was because there was no such thing as a new one. But the situation at that time was that in 1993, due to the emergence and popularity of Mac and Windows, three multimedia The word is known to the public, but the problem came - at that time, PC computing power, computing graphics, and sound were very difficult, so that at that time the computer played videos, and a separate card had to be purchased. ****Jensen saw computer graphics The future ***. The so-called acceleration refers to "speeding up graphics processing". Why should it be accelerated? Because it is slow. It is the same as Mercedes-Benz instead of a horse-drawn carriage)

But after a one-year development period, we realized that the architecture we designed was a wrong strategy, and it was unqualified from a technical point of view. Meanwhile, Microsoft is about to announce Windows 95 Direct3D based on reverse texture mapping and triangles. This meant that if we finished SEGA's console, we'd create something that wasn't compatible with Windows; but if we didn't close the deal, we'd go out of business.

In any case, we will face the fate of bankruptcy.

(Alpha Rabbit Note: The architecture error mentioned by Jensen here refers to the underlying design of the display chip.

*SEGA itself is actually a manufacturer, a manufacturer of game consoles, and if Nvidia continues to do Sega's business, that is, To B's business, then when it designs chips for Sega, it will definitely follow the To B way, That is to finish the Sega business first. As a result, after Nvidia made half of it, Microsoft suddenly came into Jensen's sight, and Microsoft was also a giant (ToC). At this time, Jensen was keenly aware that Microsoft's 3D graphics concept was better. *

*For starting a business, the difficult problem before him at that time: whether to bite the bullet and finish the To B stuff? Or turn around immediately and keep up with the trend of Microsoft? **** In the end, Jensen’s decision was that Sega’s business cannot be done, because for start-up companies, resources are limited. If they do Sega’s business, they will miss out on the business with Microsoft. ***So, Jensen's first judgment, we review it after today, is correct and far-sighted: that is to say, we must follow Microsoft. Turns out, if he hadn't picked Microsoft, Nvidia would probably be out of business by now)

I contacted the CEO of SEGA, explained to him that our invention was wrong, we could not complete the contract and the game console, and suggested that SEGA find other partners. I said to him: 'We have to stop. '

But I need SEGA to pay us in full or Nvidia won't be able to stay in business.

I was embarrassed to ask the CEO of SEGA, but to my surprise, he agreed. His understanding and generosity saved us 3 extra months, during that time we built the Riva 128, and just when we were nearly out of money, the Riva 128 rocked the nascent 3D market, got us noticed and saved the company's operations.

**The market demand for our chips is strong, which made me return to Taiwan after leaving Taiwan when I was 4 years old. I met with Mr. Zhongmou Chang from TSMC and started a relationship that lasted 25 years. **

**We were honest about our mistakes and humbly asked for help to save Nvidia from surviving. These are some of the hardest traits for the smartest, most successful people like you to develop. **

Our character is shaped by the painstaking process of pursuing our vision

In 2007, we announced CUDA GPU-accelerated computing technology. Our expectation is to make CUDA a programming model that can improve the performance of applications in scientific computing, physics simulation and image processing.

Creating a completely new computing model is difficult and historically rare. The CPU's computational model has been the standard for 60 years, since the IBM 360. CUDA needs developers to write applications and demonstrate the advantages of the GPU; developers need a large user base; large CUDA user base requires someone in the market to buy new applications.

*( Alpha Rabbit Note: 1. What happened in 2007? There was a trend at that time - the standardization of parallel computing. Here is an example, for example, the picture is x * y pixels, and the calculation of the CPU is like writing a text, line by line Writing; and GPU computing means, for example, that there are 200 grids on a page, and 200 grids are written at the same time, so the writing is completed in an instant. 2. IBM 360 is a mainframe. The mainframe is called Mainframe in English, not Computer. At that time, a 3. For CUDA, a brand new thing, ***CUDA requires developers to write applications and demonstrate the advantages of GPU-this sentence means that developers do technology When investing, if there is no future behind this piece, it will be in vain. But if more people invest, even the unpromising things will become promising, such as linux. However, that era was driven by technological ideals. It may not be the same as now. 4. Under this premise, no matter what equipment Nvidia produces in the future, its parallel computing interface will never change. If cuda monopolizes, then other competing products must also follow cuda standardization)

**Therefore, in order to solve the chicken-and-egg problem. **

We leverage our gaming graphics card, GPU GeForce, which already has a huge gaming market, to build a user base. But the cost of CUDA is very high, Nvidia's profits have taken a huge hit over the years, and our market cap is barely maintained

** Hold around $1 billion. Our years of underwhelming performance have left shareholders skeptical of CUDA and wanting us to focus on improving profitability. **

(*( Alpha Rabbit Note: Here is a question, why is a company like Nvidia that has suffered a huge blow to its annual profits, why is it not directly shut down by investors and shareholders? *)

**But we persevere, we believe that the era of accelerated computing will come, we created a conference called GTC, and worked tirelessly to promote CUDA technology around the world. **

(Alpha Rabbit Note: Jensen Huang is actually a hardcore gamer, born to love to play, geek ceiling, and Nvidia’s GTC conference started in 2009, and CUDA release just passed about 2 years. And Nvidia mainly does two things at GTC: first, the future plan of the game; second, the future plan of GPU/Cuda, and the participants here are potential users of CUDA)

Then applications like CT reconstruction, molecular dynamics, particle physics, fluid dynamics, and image processing started popping up, and our developers wrote the algorithms and accelerated the chips.

In 2012, AI researchers explored CUDA, the famous AlexNet was trained on our GPU GTX 580, starting the big explosion of AI, luckily, we realized the potential of deep learning, we risked everything to pursue depth study. Years later, the AI revolution began, and Nvidia was the driving engine. **We invented CUDA for AI, this journey has forged our character, endured pain and suffering, and it is a necessary pain on the way to pursue our vision. **

Learning to give up is the key to success

To tell another story, in 2010, Google made the Android system a platform for excellent graphics, and the mobile phone industry also had chip companies for modems. Nvidia's excellent computing power makes Nvidia a good partner for the Android system. We were successful, the stock price soared, but soon enough competitors came in, and the modem makers were learning how to make computing chips while we were learning about modems.

(*Alpha Rabbit Note: Nvidia once launched a tegra set of chips, which are very eye-catching when used in mobile phones. As a result, Qualcomm saw it, this is not good, it is going to turn the table...so Qualcomm, as a communication chip factory, began to feel When it comes to crisis, develop mobile GPU, and Nvidia also plans to make its own mobile phone and start developing baseband chips, that is, modem chips for communication. But on the side note, Qualcomm must have expert advice. *Who is this person? , who persuaded the president of Qualcomm to start making GPUs? Some people are so eccentric and never show up, but they are swaying the history of science and technology...)

Because the mobile phone market is huge, we can seize market share. However, we made the difficult decision to abandon this market. Since Nvidia's mission is to create computers that solve 'problems ordinary computers solve', we should focus on our vision and our unique contribution.

(Alpha Rabbit Note: Why give up: because the mobile modem, the application range is too narrow, it can only 1. receive signals 2. send signals, there is no To C gene, no story, no future, and Qualcomm has already done it very maturely. He's been doing graphics since day one when he started nvidia)

**Our abandonment paid off, we created a new market - robotics, with neural network processors and secure architectures to run AI algorithms. **

At that time, this was still an invisible market. Retreat from the huge mobile phone market, and create a robot market that does not know the size of the market. Today, however, we have multi-billion dollar businesses in self-driving cars, robotics, and a new industry.

** "Retreat" is not easy for someone as smart and successful as you. However, strategic retreats, sacrifices, deciding what to let go is at the core of success, very critical core. **

Run! keep running no matter what

Class of 2023, you are about to enter a world that is going through a massive transformation, just like the personal computer and chip revolution when I graduated, you are at the starting line of AI. Every industry will be revolutionized, reborn, ready for new ideas - yours.

In 40 years, we created the personal computer, the web, the mobile device, the cloud. In the current AI era, what will you create?

Whatever it is, go for it with all our might, run! Don't walk slowly.

**Whether running for food or not being treated as food by others. You often have no way of knowing which situation you're in, but keep running anyway. **

**In your journey, bring some mistakes and experiences I have made. I hope you can face failure with humility, admit mistakes and ask for help. **You will endure the pain and suffering required to achieve your dreams and make the sacrifices to commit to living meaningful lives and sprinting your life's career.

**Class of 2023, my heartfelt congratulations to each and every one of you. **

**come on! **

This article thanks to Dakai——Farseer of ChaosAI

References: