各位女士、先生，老師，來賓，驕傲的父母們，以及2023?年國立臺灣大學(xué)的畢業(yè)生們，今天對你們來說是非常特別的一天，也是你們父母夢想成真的一天。你們應(yīng)該快點從家里搬出去，這確實是一個充滿驕傲和喜悅的日子！
你們的父母為了今天做出許多犧牲，我的父母、哥哥也在這里，讓我們向所有的父母和祖父母們表達(dá)感謝。
十多年前，我第一次來到臺灣大學(xué)，陳博士邀請我參觀他的實驗室。他的兒子在硅谷工作，知道NVIDIA?的?CUDA，推薦爸爸用它來做量子物理模擬實驗。我在這個實驗室看到滿滿的?NVIDIA GeForce?游戲顯卡，插在?PC?主機板上，走道上的架子開著好幾臺電扇在散熱。
陳博士以游戲顯卡用中國臺灣人的方式自制了超級計算機，這是一個展開NVIDIA?旅程的故事，他以自己的努力為榮，也對我說：「黃先生，因為你的產(chǎn)品，讓我能追尋我的志業(yè)。」
這句話至今仍讓我感動不已，它抓住了我們公司的使命，就是幫助我們這個時代的愛因斯坦和達(dá)芬奇?zhèn)兺瓿伤麄兊氖姑?/span>
我很高興能回到臺大在你們的畢業(yè)典禮致詞。當(dāng)時我從Oregon State University?畢業(yè)的時候，世界還很簡單，沒有液晶電視，也沒有有線電視跟?MTV，手機和電話這兩個詞是分開講的。
那一年是1984?年，IBM PC-AT?和蘋果?Macintosh?開啟個人計算機革命，也開創(chuàng)了我們所知的芯片和軟件產(chǎn)業(yè)?，F(xiàn)在你們所處的是一個更復(fù)雜的世界，充滿了地緣政治、社會和環(huán)境的變化和挑戰(zhàn)。
因為科技，我們能持續(xù)連線，沉浸在一個與現(xiàn)實世界平行的數(shù)位世界里，汽車也能自動駕駛了。
AI會創(chuàng)造過去不存在的新工作，每個人都要學(xué)習(xí)掌握?AI?紅利
在計算機產(chǎn)業(yè)創(chuàng)造了家用個人計算機40?年后，我們發(fā)明了人工智能，例如自動駕駛或辨識?X?光影像的軟件，AI?軟件為計算機自動化打開了大門，也開啟了價值數(shù)兆美元的產(chǎn)業(yè)——醫(yī)療保健，金融服務(wù)，運輸和制造業(yè)等等，AI?創(chuàng)造了各種機會。
敏捷的公司利用AI?提升他們的地位，反之，那些落后的公司將會滅亡。正在聽這場演講的創(chuàng)業(yè)家們也將開創(chuàng)新事業(yè)，如同過去每一個運算時代，新的產(chǎn)業(yè)會出現(xiàn)。
AI創(chuàng)造過去不存在的新工作，像數(shù)據(jù)工程，提示工程師，AI 工廠營運與?AI?安全工程師等等。
這些都是以前沒有的工作，有些工作會因為自動化而過時了，但可以肯定的是，AI將改變每一個工作，讓工程師、設(shè)計師、藝術(shù)家、營銷人員還有制造業(yè)的擘畫者有更好的表現(xiàn)。
就像過去每個世代的人一樣，他們擁抱新技術(shù)然后成功了，每一家公司，包括你們，都要學(xué)著掌握AI?的紅利，讓?AI?成為你的副駕駛，做出驚人的事業(yè)。
有些人擔(dān)心AI?可能會搶走他們的工作，其實，是擅長使用?AI?的「人」會搶走這些人的工作。我們處在重大科技時代的開端，像?PC，網(wǎng)絡(luò)，手機和云端等等。
但是AI?的技術(shù)更基本，因為每個運算的層次都被重新打造，從我們?nèi)绾螌戃浖剿绾芜\作，AI?從根本上重新創(chuàng)造了運算方式。
從各個層面來看，這是計算機產(chǎn)業(yè)的重生，對臺灣企業(yè)來說，這也是個千載難逢的機會。你們就是計算機產(chǎn)業(yè)的基石，在未來的十年，我們的產(chǎn)業(yè)將以全新、加速的AI?技術(shù)取代全球超過一兆美元的傳統(tǒng)計算機市場。
NVIDIA的第一個故事：面對錯誤，勇于求助
我的旅程比你們早40?年開始，1984?年是畢業(yè)的好時節(jié)，我預(yù)期?2023?年也是如此。在你們的旅程開始之時，我該分享些什么呢？今天是你們?nèi)松凶畛晒Φ囊惶?，你們要從臺大畢業(yè)了。我畢業(yè)那天也是個成功人士，直到我創(chuàng)辦?NVIDIA?為止。
在NVIDIA，我經(jīng)歷了失敗，非常大的失敗。那段歷程充滿羞辱和尷尬，公司幾乎要完蛋。讓我跟你們說三個關(guān)于?NVIDIA?的故事，這些故事定義了今日的?NVIDIA。
我們創(chuàng)辦NVIDIA?來打造加速計算，我們第一個技術(shù)是?PC?游戲的?3D?圖形。我們創(chuàng)造了一種非傳統(tǒng)的?3D?技術(shù)，它叫前向材質(zhì)貼圖和曲線（forward texture mapping and curves）。這個技術(shù)的成本明顯較低，因此我們拿到了?SEGA?主機的合約，我們的平臺吸引了游戲制造商，也拿到資金。
經(jīng)過一年的開發(fā)，我們發(fā)現(xiàn)這個架構(gòu)策略錯了，而且微軟發(fā)布了Windows 95 Direct 3D，這是基于逆材質(zhì)貼圖和三角（inverse texture mapping and triangles）的技術(shù)。
很多公司已經(jīng)在開發(fā)支援這個標(biāo)準(zhǔn)的3D?芯片，如果我們硬要完成?SEGA?的合約，我們的技術(shù)不僅不怎么樣，跟?Windows?不兼容，而且還追趕不上。但是，如果我們不履行合約，就會一毛錢也拿不到。不論如何，公司都會破產(chǎn)。
我跟SEGA?的?CEO?說，公司技術(shù)開發(fā)方向錯了，SEGA?應(yīng)該要換合作伙伴，因為我們無法履行合約，這個合作應(yīng)該要終止。但我們需要?SEGA?的錢，不然?NVIDIA?就要倒閉了。
我很難為情地跟SEGA?的?CEO?入交昭一郎提出這樣的要求， 讓我很驚訝的是，他竟然同意了。
他的理解和慷慨讓我們能再活六個月。因為這個機會，我們打造了RIVA 128。就在我們快要把錢燒光的時候，RIVA 128?讓剛起步的?3D?市場驚艷不已，我們被看見了，公司也活下來了。
我們的芯片在市場需求強勁，四歲就離開臺灣的我又回到這里，跟臺積電的張忠謀見面，開啟了持續(xù)25?年的合作關(guān)系。
面對自己犯下的錯，并以謙遜的態(tài)度向他人求助，這兩件事挽救NVIDIA?免于倒閉，這樣的態(tài)度對你們這些聰明又成功的人來說是最困難的。
NVIDIA的第二個故事：為了實現(xiàn)愿景必然要忍受痛苦
2007年，我們發(fā)表了?CUDA GPU?加速計算。我們的目標(biāo)是讓?CUDA?成為一種程序設(shè)計模型，能加速科學(xué)運算、物理模擬、影像處理的應(yīng)用。
從無到有創(chuàng)造一個全新的運算模型非常困難，過去很少人能做得到。CPU運算模型自IBM System 360?以來已成為世人的標(biāo)準(zhǔn)，它有?60?年的歷史了。CUDA?需要開發(fā)者撰寫應(yīng)用程序，并展示這款?GPU?的優(yōu)點。
開發(fā)者需要龐大的安裝基礎(chǔ)，大型CUDA?安裝基礎(chǔ)設(shè)施需要用戶購買新的應(yīng)用程序。為了解決先有雞還是先有蛋的問題，我們使用游戲?GPU，也就是?GeForce?來建立安裝基礎(chǔ)，因為它已經(jīng)有大量的游戲玩家市場。
但CUDA?的額外成本非常高，NVIDIA?的利潤多年來表現(xiàn)不佳，我們的市值僅略高于10?億美元。多年的低迷表現(xiàn)，讓我們的股東懷疑?CUDA?的能耐，他們希望我們專心提高獲利能力，但我們?nèi)匀粓猿肿鱿胱龅氖隆?/span>
我們相信加速運算的時代終將來臨，我們開始辦GTC?大會，也努力向全球推廣CUDA，接著各式應(yīng)用開始出現(xiàn)了。
從震測處理、CT重建、分子動力學(xué)、粒子物理學(xué)、流體動力學(xué)到影像處理，各種科學(xué)研究領(lǐng)域一個接著一個使用我們的技術(shù)。我們與每一位開發(fā)者合作撰寫他們的算法，因此進步非?？欤搅?2012?年，AI?研究者發(fā)現(xiàn)了?CUDA。
大家熟知的AlexNet?是在?GeForce GTX 580?上訓(xùn)練的，AI?技術(shù)開始大爆發(fā)。幸運的是，我們知道深度學(xué)習(xí)深具潛力，我們公司傾全力開發(fā)這個新領(lǐng)域的技術(shù)。我們冒上一切風(fēng)險追求深度學(xué)習(xí)。10?年后，AI?革命開始了，現(xiàn)在的?NVIDIA是 全球?AI?開發(fā)者的引擎。
我們打造了CUDA?并在加速運算和?AI?領(lǐng)域成為領(lǐng)導(dǎo)者，這個旅程塑造了?NVIDIA?的企業(yè)性格，為了實現(xiàn)愿景必然要忍受痛苦與經(jīng)歷磨難。
NVIDIA的第三個故事：有策略的撤退，決定什么不去做
讓我再說一個故事。2010年，Google?打算將?Android?開發(fā)成一款具有優(yōu)秀圖形的移動計算機。手機產(chǎn)業(yè)已有了調(diào)制解調(diào)器專業(yè)的芯片公司，NVIDIA?的運算和圖形專業(yè)使我們成為與?Android?合作的理想伙伴，所以我們因而進入了手機芯片市場。
我們很快的成功了，業(yè)務(wù)跟股價一起飆升，競爭對手也蜂涌而至。調(diào)制解調(diào)器芯片制造商學(xué)習(xí)如何制造運算芯片，而我們也在學(xué)習(xí)如何制造調(diào)制解調(diào)器。
手機市場非常大，我們可以提高市占率，但我們沒有這么做，我們決定放棄這個市場。
NVIDIA的使命是打造出一款計算機，它能解決一般計算機不能解決的問題。我們應(yīng)該全力以赴實現(xiàn)這個愿景，并做出獨特的貢獻。
我們有策略的撤退最終得到了回報。離開了手機市場，我們不畫地自封，而是創(chuàng)造了一個新的市場。我們創(chuàng)造一款新型的計算機，為機器人計算提供服務(wù)。這款新型計算機搭載神經(jīng)網(wǎng)絡(luò)處理器，有安全的結(jié)構(gòu)進行AI?演算。
在當(dāng)時，這是一個零億美元的市場，我們離開了龐大的手機市場，跑去創(chuàng)立一個0?元的機器人市場。
現(xiàn)在的NVIDIA?的有數(shù)十億美元的汽車和機器人業(yè)務(wù)，并開創(chuàng)了一個新的產(chǎn)業(yè)。
放棄對你們這樣聰明又成功的人來說并不容易。然而，有策略的撤退，犧牲，決定什么不要做，不就是成功的核心嗎？
2023年的畢業(yè)生們，你們將進入正在經(jīng)歷巨變的世界，就像當(dāng)年的我身處?PC?和芯片革命之中一樣，你們站在?AI?的起跑線上，每一個行業(yè)都將歷經(jīng)革命，浴火重生。
你要為創(chuàng)新做好準(zhǔn)備，擁有自己的想法。這40?年來，人類創(chuàng)造?PC、網(wǎng)絡(luò)、手機、云端技術(shù)，現(xiàn)在則進入了?AI?時代。
你想創(chuàng)造什么呢？不管是什么，跟我們一樣跑起來追逐它吧！跑起來，別用走。記得，你若不跑起來做個掠食者，就只能成為掠食者的食物。
我希望這段話可以用中文講，所以大家可以聽得懂。
你要么跑起來掠食，要么努力奔跑免得成為掠食者的食物。
大多時候，你很難區(qū)別這兩者。不管怎么樣，跑起來吧！在你自己的旅程中，帶上我今天跟你分享的經(jīng)驗：謙虛地面對失敗，承認(rèn)錯誤，向他人求助。為了實現(xiàn)夢想，你將承受痛苦與磨難。為了人生的使命與志業(yè)，你得做出犧牲并全心投入。
2023年的畢業(yè)生們，我誠摯的祝福你們每一位。加油！
Full Transcript: NVIDIA CEO Jensen Huang's Commencement Address at National Taiwan University
Ladies and gentlemen, esteemed faculty members, distinguished guests, proud parents, and above all, the 2023 graduating class of the National Taiwan University. Today is a very special day for you, and a dream come true for your parents.
You should be moving out soon. It is surely a day of pride and joy.So, your parents have made sacrifices to see you on this day. My parents are here, and so is my brother. Let's show all of our parents and our grandparents, many of them are here, our appreciation.
I came to NTU for the first time over a decade ago. Dr. Chen invited me to visit his computational physics lab. As I recall, his son, based in Silicon Valley, had learned of NVIDIA's CUDA invention and recommended his father utilize it for his quantum physics simulations.
When I arrived, he opened the door to show me what he had made. NVIDIA GeForce gaming cards filled the room, plugged into open PC motherboards, and sitting on metal shelves in the aisles were oscillating platform fans.
Dr. Chen had built a homemade supercomputer, the Taiwanese Way, out of gaming graphics cards. He started here, an early example of NVIDIA's journey.
He was so proud, and he said to me, Mr. Huang, because of your work, I can do my life's work. Those words touch me to this day, and perfectly capture our company's purpose, to help the Einstein and Da Vinci of our time do their life's work.
I am so happy to be back at NTU, and to be your commencement address. The world was simpler when I graduated from Oregon State University. TVs were not flat yet. There was no cable television, and MTV.
And the words mobile and phone didn't go together. The year was 1984. The IBM PC-AT and Apple Macintosh launched the PC revolution.
And started the chip and software industry that we know today. You enter a far more complex world, with geopolitical, social, and environmental changes and challenges.
Surrounded by technology, we are now perpetually connected, and immersed in a digital world that parallels our real world. Cars are starting to drive by themselves.
AI will create new jobs that didn't exist before
Forty years after the computer industry created the home PC, we invented artificial intelligence. Like software that automatically drives a car, or studies x-ray images, AI software has opened the door for computers to automate tasks for the world's largest, multi-trillion dollars of industries.
Healthcare, financial services, transportation, and manufacturing. AI has opened immense opportunities.
Agile companies will take advantage of AI, and boost their position. Companies less so, will perish. Entrepreneurs, many of them here today, will start new companies.
And like in every computing era before, create new industries. AI will create new jobs that didn't exist before. Like data engineering, prompt engineering, AI factory operations, and AI safety engineers.
These are jobs that never existed before. Automated tasks will obsolete some jobs. And for sure, AI will change every job. Supercharging the performance of programmers, designers, artists, marketers, and manufacturing planners.
Just as every generation before you embraced technologies to succeed, every company, and you, must learn to take advantage of AI. And do amazing things with an AI co-pilot by your side.
While some worry that AI may take their jobs, someone who expert with AI will. We are at the beginning of a major technology era, like PC, internet, mobile, and cloud. But AI is far more fundamental because every computing layer has been reinvented, from how we write software to how it's processed.
AI has reinvented computing from the ground up. In every way, this is a rebirth of the computer industry. And a golden opportunity for the companies of Taiwan.
You are the foundation and bedrock of the computer industry. Within the next decade, our industry will replace over a trillion dollars of the world's traditional computers with new, accelerated AI computers.
The first story of NVIDIA: Confronting mistakes and asking for help
My journey started 40 years before yours. 1984 was a perfect year to graduate. I predict that 2023 will be as well.
What can I tell you as you begin your journey? Today is the most successful day of your life so far. You're graduating from the National Taiwan University. I was also successful until I started NVIDIA.
At NVIDIA, I experienced failures. Great big ones. All humiliating and embarrassing. Many nearly doomed us.
Let me tell you three NVIDIA stories that define us today.
We founded NVIDIA to create accelerated computing. Our first application was 3D graphics for PC gaming.
We invented an unconventional 3D approach called forward texture mapping and curves. Our approach was substantially lower cost. We won a contract with SEGA to build their game console, which attracted games for our platform and funded our company.
After one year of development, we realized our architecture was the wrong strategy. It was technically poor. And Microsoft was about to announce Windows 95 Direct 3D based on inverse texture mapping and triangles.
Many companies were already working on 3D chips to support this standard. If we completed SEGA's game console, we would have built inferior technology, be incompatible with Windows, and be too far behind to catch up.
But we would be out of money if we didn't finish the contract. Either way, we would be out of business. I contacted the CEO of SEGA and explained that our invention was the wrong approach.
That SEGA should find another partner. And that we could not complete the contract and the console. We had to stop. But I needed Sega to pay us in whole. Or NVIDIA would be out of business.
I was embarrassed to ask. Irimajiri-san, the CEO of SEGA, to his credit and my amazement, agreed.
His understanding and generosity gave us six months to live. With that, we built RIVA 128. Just as we were running out of money. RIVA 128 shocked the young 3D market, put us on the map, and saved the company.
The strong demand for our chip led me back to Taiwan after leaving at the age of four to meet Morris Chang at TSMC and started a partnership that has lasted 25 years. Confronting our mistake and with humility asking for help, save NVIDIA.
These traits are the hardest for the brightest and most successful like yourself.
The second story of NVIDIA: Endure pain and suffering needed to realize your dreams
In 2007, we announced CUDA GPU Accelerated Computing. Our aspiration was for CUDA to become a programming model that boosts applications from scientific computing and physics simulations to image processing. Creating a new computing model is incredibly hard and rarely done in history.
The CPU computing model has been the standard for 60 years since the IBM System 360. CUDA needed developers to write applications and demonstrate the benefits of the GPU.
Developers needed a large installed base. A large CUDA installed base needed customers buying new applications. To solve the chicken or the egg problem, we used GeForce, our gaming GPU, which already had a large market of gamers, to build the installed base.
But the added cost of CUDA was very high. NVIDIA's profits took a huge hit for many years. Our market cap hovered just above $1 billion.
We suffered many years of poor performance. Our shareholders were skeptical of CUDA. And preferred we focused on improving profitability. But we persevered.
We believed a time for accelerated computing would come. We created a conference called GTC and promoted CUDA tirelessly worldwide. Then the applications came.
Seismic processing, CT reconstruction, molecular dynamics, particle physics, fluid dynamics, and image processing. One science domain after another, they came.
We worked with each developer to write their algorithms and achieved incredible speedups. Then, in 2012, AI researchers discovered CUDA.
The famous AlexNet trained on GeForce GTX 580 started the Big Bang of AI. Fortunately, we realized the potential of deep learning as a whole new software approach. And turned every aspect of our company to advance this new field.
We risked everything to pursue deep learning. A decade later, the AI revolution started. And NVIDIA is the engine of AI developers worldwide.
We invented CUDA and pioneered accelerated computing and AI. But the journey forged our corporate character to endure the pain and suffering that is always needed to realize a vision.
The third story of NVIDIA: Strategic retreat
One more story. In 2010, Google aimed to develop Android into a mobile computer with excellent graphics.
The phone industry had chip companies with modem expertise. NVIDIA's computing and graphics expertise made us an ideal partner to help build Android. So we entered the mobile chip market.
We were instantly successful. And our business and stock price surged. The competition quickly swarmed. Modem chip makers were learning how to build computing chips. And we were learning how to build modems.
The phone market is huge. We could fight for share. Instead, we made a hard decision and sacrificed the market.
NVIDIA's mission is to build computers to solve problems that ordinary computers cannot. We should dedicate ourselves to realizing our vision and to making a unique contribution.
Our strategic retreat paid off. By leaving the phone market, we opened our minds to invent a new one. We imagined creating a new type of computer for robotic computers.With neural network processor, safety architectures that run AI algorithms.
At the time, this was a zero billion dollar market. To retreat from a giant phone market to create a zero billion dollar robotics market. We now have billions of dollars of automotive and robotics business and started a new industry. Retreat does not come easily to the brightest and most successful people like yourself.
Yet, strategic retreat, sacrifice, deciding what to give up, is that a core, the very core of success?
Class of 2023, you're about to go into a world witnessing great change. And just as I was with the PC and chip revolution, you're at the beginning, at the starting line of AI. Every industry will be revolutionized. Reborn.
Ready for new ideas. Your ideas. In 40 years, we created the PC, internet, mobile, cloud, and now the AI era.
What will you create? Whatever it is, run after it like we did. Run. Don't walk. Remember, either you're running for food, or you are running from being food.
I hope that part can be translated into Chinese so that everyone can understand.
Either you're running for food, or you are running from becoming food.
And oftentimes, you can't tell which. Either way, run. And for your journey, take along some of my learnings. That you will have the humility to confront failure, admit a mistake, and ask for help.
You will endure pain and suffering needed to realize your dreams. And you will make sacrifices to dedicate yourself to a life of purpose and doing your life's work.
Class of 2023, I extend my heartfelt congratulations to each one of you. Jiayou!
