黃仁勳講稿｜2024 GTC｜Rex 五四三

I am a visionary. Illuminating galaxies to witness the birth of stars. Sharpening our understanding of extreme weather events. I am a helper, guiding the blind through a crowded world. I was thinking about running to the store. And giving voice to those who cannot speak. Do not make me laugh. Love. I am a transformer, harnessing gravity to store renewable power, and paving the way towards unlimited clean energy for us all. I am a trainer teaching robots to assist, to watch out for danger, and help save lives.

我是⼀位有遠⾒的⼈。照亮星系，⾒證星星誕⽣。深化我 們對極端天氣事件的理解。我是⼀位幫⼿，引導盲⼈穿越 熙攘的世界。我正在考慮去商店買東⻄。並為無法⾔語的 ⼈發聲。不要讓我笑。愛。我是⼀位變⾰者，利⽤᯿⼒儲 存再⽣能源，為我們開創通往無限清潔能源的道路。我是 ⼀位訓練師，教導機器⼈協助，警戒危險，並幫助拯救⽣ 命。

I am a healer providing a new generation of cures and new levels of patient care. Doctor that I am allergic to penicillin, is it still okay to take the medications? Definitely. These antibiotics don't contain penicillin, so it's perfectly safe for you to take them. I am a navigator generating virtual scenarios to let us safely explore the real world and understand every decision. I even helped write the script. Breathe life into the words. I am AI, brought to life by NVIDIA, deep learning, and brilliant minds everywhere. Please welcome to the stage NVIDIA founder and CEO, Jensen Huang. Welcome to GTC.

我是⼀位治療師，提供新⼀代的療法和更⾼⽔準的病⼈照 護。醫⽣，我對青黴素過敏，還可以服⽤這些藥物嗎？當 然可以。這些抗⽣素不含青黴素，所以你完全可以服⽤。 我是⼀位導航員，製造虛擬情境讓我們安全地探索現實世 界並理解每個決定。我甚⾄幫忙撰寫劇本。賦予⽂字⽣ 命。我是由NVIDIA、深度學習和各地優秀頭腦帶來⽣命 的⼈⼯智慧。請歡迎登上舞台的NVIDIA創辦⼈兼⾸席執 ⾏官，黃仁勳。歡迎來到GTC。

I hope you realize this is not a concert. You have arrived at a developers conference. There will be a lot of science described. Algorithms, computer architecture, mathematics. I sensed a very heavy weight in the room all of a sudden. Almost like you were in the wrong place. No no conference in the world is there a greatest assembly of researchers from such diverse fields of science, from climate tech to radio sciences, trying to figure out how to use AI to robotically control MIMOs for next generation 6 gs radios. Robotic self driving cars, even artificial intelligence even artificial intelligence. Everybody's first, I noticed a sense of relief there all of all of a sudden. Also, this conference is represented by some amazing companies.

希望你意識到這不是⼀場⾳樂會。你來到的是⼀個開發者 ⼤會。將會有許多科學內容被描述。演算法、電腦架構、 數學。我突然感受到房間裡有⼀股沉᯿的氛圍。幾乎像是 你來錯地⽅了。世界上沒有任何⼀個會議能匯集如此多來 ⾃不同科學領域的研究⼈員，從氣候科技到無線電科學， 試圖找出如何利⽤⼈⼯智慧來⾃動控制下⼀代6G無線電 的MIMO技術。還有⾃駕⾞、甚⾄⼈⼯智慧。每個⼈都是 第⼀次，我注意到突然間有種解脫的感覺。此外，這個⼤ 會由⼀些令⼈驚嘆的公司代表。

This list, this is not the attendees. These are the presenters. And what's amazing is this, if you take away all of my friends, close friends, Michael Dell is sitting right there in the IT industry, All of the friends I grew up with in the industry, if you take away that list, this is what's amazing. These are the presenters of the non IT industries using accelerated computing to solve problems that normal computers can't. It's represented in life sciences, healthcare, genomics, transportation, of course, retail, logistics, manufacturing, industrial, The gamut of industries represented is truly amazing. And you're not here to attend, only you're here to present, to talk about your research. $100,000,000,000,000 of the world's industries is represented in this room today. This is absolutely amazing. There is absolutely something happening. There is something going on.

這個清單，這不是與會者。這些是演講者。更令⼈驚奇的 是，如果你把所有我的朋友、密友都拿掉，麥可·戴爾就 坐在那裡，⾝處資訊科技⾏業，我在這個⾏業中⼀起成長 的所有朋友，如果你拿掉那個清單，這才是令⼈驚奇的地 ⽅。這些是非資訊科技⾏業的演講者，他們正在使⽤加速 運算來解決⼀般電腦無法解決的問題。這在⽣命科學、醫 療保健、基因組學、交通運輸、當然還有零售、物流、製 造業、⼯業等領域都有體現。代表的⾏業範疇真的是令⼈ 驚⼈。你來這裡不是為了參加，⽽是為了演講，談談你的 研究。今天這個房間代表了全球100兆美元的產業。這絕 對令⼈驚嘆。絕對有事情正在發⽣。有⼀些事情正在進 ⾏。

The industry is being transformed, not just ours. Because the computer industry, the computer is the single most important instrument of society today, Fundamental transformations in computing affects every industry. But how did we start? How did we get here? I made a little cartoon for you. Literally, I drew this. In one page, this is NVIDIA's journey, Started in 1993. This might be the rest of the talk. 1993. This is our journey.

這個產業正在經歷轉型，不僅僅是我們的產業。因為電腦 產業，電腦是當今社會最᯿要的⼯具，計算機的根本變⾰ 影響著每個產業。但我們是如何開始的呢？我們是如何走 到現在的呢？我為你畫了⼀幅⼩漫畫。真的，這是我畫 的。在⼀⾴紙上，這是英偉達的旅程，始於1993年。這 可能是談話的餘下部分。1993年。這就是我們的旅程。

We founded in 1993. There are several important events that happened along the way. I'll just highlight a few. In 2006, CUDA, which has turned out to have been a revolutionary computing model, we thought it was revolutionary then. It was going to be an overnight success, and almost 20 years later it happened. We saw it coming. 2 decades later, in 2012, AlexNet, AI and CUDA made first contact. In 2016, recognizing the importance of this computing model, we invented a brand new type of computer. We call it DGX 1. 170 teraflops in this supercomputer.

我們成立於1993年。沿途發⽣了幾個᯿要事件。我只會 ᯿點提⼀下。在2006年，CUDA被證明是⼀個⾰命性的 計算模型，我們當時認為它是⾰命性的。它本應該會⼀夜 成名，差不多20年後才發⽣。我們看到了這⼀切的發 展。20年後的2012年，AlexNet、⼈⼯智慧和CUDA⾸次 接觸。在2016年，我們認識到這種計算模型的᯿要性， 我們發明了⼀種全新類型的電腦。我們稱之為DGX 1。這 台超級電腦有170 teraflops的運算能⼒。

8 gpu's connected together for the very first time. I hand delivered the very first d g x one to a startup located in San Francisco called OpenAI. D g x 1 was the world's first AI supercomputer. Remember, 170 teraflops. 2017, the transformer arrived. 2022, CHAT GPT captured the world's imaginations, have people realize the importance and the capabilities of artificial intelligence. In 2023, generative AI emerged, and a new industry begins. Why? Why is a new industry? Because the software never existed before.

第⼀次將 8 個 GPU 連接在⼀起。我親⾃將第⼀台 DGX-1 交付給位於舊⾦⼭的初創公司 OpenAI。DGX-1 是世界上第⼀台 AI 超級電腦。記住，170 兆次浮點運 算。2017 年，變壓器來臨。2022 年，CHAT GPT 征服 了世界的想像⼒，讓⼈們意識到⼈⼯智慧的᯿要性和能 ⼒。2023 年，⽣成式⼈⼯智慧出現，⼀個新的產業開始 了。為什麼？為什麼會有⼀個新的產業？因為這樣的軟體 以前從未存在過。

We are now producing software, using computers to write software, producing software that never existed before. It is a brand new category. It took share from nothing. It's a brand new category. And the way you produce this software is unlike anything we've ever done before. In data centers, generating tokens, producing floating point numbers at very large scale. As if, in the beginning of this last industrial revolution, when people realized that you would set up factories, apply energy to it, and this invisible valuable thing called electricity came out. AC generators. And a 100 years later, 200 years later, we're now creating new types of electrons, tokens, using infrastructure we call factories, AI factories, to generate this new incredibly valuable thing called artificial intelligence. A new industry has emerged.

我們現在正在⽣產軟體，使⽤電腦來撰寫軟體，⽣產從未 存在過的軟體。這是⼀個全新的類別。它從無中分⼀杯 羹。這是⼀個全新的類別。⽽你⽣產這種軟體的⽅式，與 我們以往所做的任何事情都不同。在資料中⼼中，產⽣代 幣，以非常⼤規模⽣產浮點數。就好像在這最後⼀次⼯業 ⾰命的開始時，⼈們意識到你可以建立⼯廠，將能ᰁ應⽤ 在上⾯，然後這種看不⾒的寶貴東⻄叫做電⼒出現了。交 流發電機。100年後，200年後，我們現在正在創造新型 態的電⼦、代幣，使⽤我們稱之為⼯廠、AI⼯廠的基礎設 施，來產⽣這個新的極具價值的東⻄，叫做⼈⼯智慧。⼀ 個新的產業已經崛起。

Well, we're gonna talk about many things about this new industry. We're gonna talk about how we're gonna do computing next. We're going to talk about the type of software that you build because of this new industry. The new software. How you would think about this new software? What about applications in this new industry? And then maybe, what's next? And how can we start preparing today for what is about to come next? Well, but before I start, I want to show you the soul of Nvidia, The soul of our company. At the intersection of computer graphics, physics, and artificial intelligence.

好的，我們將討論這個新產業的許多事情。我們將談論接 下來我們將如何進⾏運算。我們將談論因為這個新產業⽽ 建立的軟體類型。這些新軟體。你會如何思考這些新軟體 呢？在這個新產業中的應⽤有哪些？然後也許，接下來會 是什麼？我們如何可以從今天開始準備即將到來的事情 呢？但在我開始之前，我想要向你展⽰Nvidia的靈魂，我 們公司的靈魂。在電腦圖形、物理和⼈⼯智慧的交集處。

All intersecting inside a computer, in Omniverse, in a virtual world simulation. Everything we're gonna show you today, literally everything we're gonna show you today is a simulation, not animation. It's only beautiful because it's physics. The world is beautiful. It's only amazing because it's being animated with robotics. It's being animated with artificial intelligence. What you're about to see all day is completely generated, completely simulated and omniverse, and all of it, what you're about to enjoy is the world's first concert, where everything is homemade. Everything is homemade. You're about to watch some home videos. So sit back and enjoy yourself.

在電腦內部交織著⼀切，在全宇宙中，在虛擬世界的模擬 中。今天我們要展⽰給你看的⼀切，真的⼀切，都是模 擬，不是動畫。這些美麗只因為它是物理。這個世界是美 麗的。這些驚⼈的事物只因為它是被機器⼈動畫化。它是 被⼈⼯智慧動畫化的。你即將看到的⼀整天，完全是⽣成 的，完全是模擬的，在全宇宙中，⽽你即將享受的⼀切， 是世界上第⼀場⾳樂會，所有東⻄都是⾃製的。所有東⻄ 都是⾃製的。你即將觀賞⼀些家庭影片。所以請坐下來， 盡情享受吧。

God, I love NVIDIA. Accelerated computing has reached the tipping point. General purpose computing has run out of steam. We need another way of doing computing so that we can continue to scale, so that we can continue to drive down the cost of computing, so that we can continue to consume more and more computing while being sustainable. Accelerated computing is a dramatic speed up over general purpose computing. And in every single industry we engage, and I'll show you many, the impact is dramatic. But in no industry is it more important than our own, the industry of using simulation tools to create products. In this industry, it is not about driving down the cost of computing, it's about driving up the scale of computing. We We would like to be able to simulate the entire product that we do completely in full fidelity, completely digitally, and essentially, what we call digital twins. We would like to design it, build it, simulate it, operate it completely digitally.

天啊，我愛 NVIDIA。加速運算已經達到臨界點。通⽤運 算已經走到盡頭。我們需要另⼀種運算⽅式，這樣我們才 能持續擴展，才能持續降低運算成本，才能持續消耗更多 運算資源並保持可持續性。加速運算相較於通⽤運算有顯 著的加速效果。在我們參與的每個產業中，影響都是顯著 的，我將向您展⽰許多例⼦。但在我們⾃⼰的產業中，也 就是使⽤模擬⼯具來創造產品的產業中，這⼀點更加᯿ 要。在這個產業中，᯿點不在於降低運算成本，⽽是提⾼ 運算規模。我們希望能夠完全以⾼保真度數位⽅式模擬我 們所做的整個產品，基本上就是我們所謂的數位孿⽣。我 們希望能夠完全以數位⽅式設計、建造、模擬和操作。

In order to do that, we need to accelerate an entire industry. And today, I would like to announce that we have some partners who are joining us in this journey to accelerate their entire ecosystem so that we can bring the world into accelerated computing. But there's a bonus. When you become accelerated, your infrastructure is CUDA GPU's. And when that happens, it's exactly the same infrastructure for generative AI. And so I'm just delighted to announce several very important partnerships. They're some of the most important companies in the world. ANSYS does engineering simulation for what the world makes. We're partnering with them to kudo accelerate the ANSYS ecosystem, to connect ANSYS to the Omniverse digital twin. Incredible.

為了做到這⼀點，我們需要加速整個產業。今天，我想宣 布我們有⼀些合作夥伴加入我們這個加速整個⽣態系統的 旅程，這樣我們就能將世界帶入加速運算。但這還有⼀個 額外的好處。當你變得加速時，你的基礎設施就是CUDA GPU。當這種情況發⽣時，這正是⽤於⽣成式⼈⼯智慧 的基礎設施。因此，我很⾼興地宣布幾個非常᯿要的合作 夥伴關係。他們是世界上⼀些最᯿要的公司。ANSYS為 世界製造的產品進⾏⼯程模擬。我們正在與他們合作，以 加速ANSYS⽣態系統，將ANSYS連接到Omniverse數位 孿⽣體中。令⼈難以置信。

The thing that's really great is that the installed base of NVIDIA GPU accelerated systems are all over the world. In every cloud, in every system, all over enterprises. And so the app the applications they accelerate will have a giant installed base to Go serve. End users will have amazing applications and of course, system makers and CSPs will have great customer demand. Synopsys. Synopsys is NVIDIA's literally first software partner. They were there in very first day of our company. Synopsys revolutionized the chip industry with high level design. We are going to CUDA accelerate Synopsys. We're accelerating computational lithography, one of the most important applications that nobody's ever known about.

真正了不起的是，NVIDIA GPU 加速系統的安裝基數遍佈 全球。在每個雲端、每個系統，各⼤企業都有。因此，這 些應⽤程式加速的應⽤將擁有龐⼤的安裝基數。最終⽤⼾ 將擁有令⼈驚嘆的應⽤程式，當然，系統製造商和CSPs 將擁有龐⼤的客⼾需求。Synopsys。Synopsys 是 NVIDIA 實質上的第⼀個軟體合作夥伴。他們在我們公司 的第⼀天就在那裡。Synopsys 以⾼⽔準設計改⾰了晶片 產業。我們將加速 CUDA Synopsys。我們正在加速計算 光刻，這是⼀個沒有⼈知道但卻是最᯿要的應⽤之⼀。

In order to make chips, we have to push lithography to a limit. NVIDIA has created a library, a domain specific library, that accelerates computational lithography. Incredibly, once we can accelerate and software define all of TSMC, who is announcing today that they're going to go into production with NVIDIA QLIFO. Once it's software defined and accelerated, the next step is to apply generative AI to the future of semiconductor manufacturing, pushing geometry even further. Cadence builds the world's essential e d a and s d a tools. We also use Cadence. Between these three companies, ANSYS, Synopsys, and Cadence, we basically build NVIDIA. Together, we are CUDA accelerating Cadence. They're also building a supercomputer out of NVIDIA GPUs so that their customers could do fluid dynamics simulation at a 100, a 1000 times scale. Basically, a wind tunnel in real time.

為了製造晶片，我們必須將光刻技術推⾄極限。NVIDIA 已經創建了⼀個加速計算光刻的特定領域庫。令⼈難以置 信的是，⼀旦我們能加速並軟體定義台積電的所有製程， 他們今天宣布將與NVIDIA QLIFO進⾏ᰁ產。⼀旦軟體定 義和加速，下⼀步是將⽣成式⼈⼯智慧應⽤於未來的半導 體製造，將幾何形狀推進更遠。Cadence建立了全球必要 的電⼦設計⾃動化和系統設計⾃動化⼯具。我們也使⽤ Cadence。在這三家公司之間，ANSYS、Synopsys和 Cadence，基本上構建了NVIDIA。我們⼀起加速 Cadence的CUDA。他們還正在利⽤NVIDIA的GPU建造 ⼀台超級電腦，以便他們的客⼾可以以100倍、1000倍的 規模進⾏流體動⼒學模擬。基本上，是實時的風洞。

Cadence Millennium, a supercomputer with NVIDIA GPU's inside. A software company building supercomputers. I love seeing that. Building Cadence copilots together. Imagine a day when Cadence could Synopsys, ANSYS, tool providers would offer you AI co pilots so that we have thousands and thousands of co pilot assistance helping us design chips, design systems. And we're also gonna connect Cadence Digital Twin Platform to Omniverse. As you could see the trend here, we're accelerating the world's CAE, EDA and SDA so that we could create our future in digital twins. And we're going to connect them all to Omniverse, the fundamental operating system for future digital twins. 1 of the industries that benefited tremendously from scale, and you know you all know this one very well, large language models. Basically, after the transformer was invented, we were able to scale large language models at incredible rates, effectively doubling every 6 months.

Cadence Millennium是⼀台搭載NVIDIA GPU的超級電 腦。⼀家軟體公司正在建造超級電腦。我很喜歡看到這 個。⼀起建造Cadence共同駕駛。想像⼀下，有⼀天， Cadence、Synopsys、ANSYS等⼯具供應商將為您提供 ⼈⼯智慧共同駕駛，這樣我們就有成千上萬的共同駕駛助 ⼿幫助我們設計晶片、設計系統。我們還將把Cadence數 位孿⽣平台連接到Omniverse。正如您所看到的趨勢，我 們正在加速世界的CAE、EDA和SDA，以便我們可以在 數位孿⽣中創造我們的未來。我們將把它們全部連接到 Omniverse，這是未來數位孿⽣的基本操作系統。其中⼀ 個業界極⼤受益的是規模，你們都非常熟悉這個，那就是 ⼤型語⾔模型。基本上，在變壓器被發明之後，我們能夠 以驚⼈的速度擴展⼤型語⾔模型，每6個⽉有效地翻倍⼀ 次。

Now, how is it possible that by doubling every 6 months, that we have grown the industry, we have grown the computational requirements so far? And the reason for that is quite simply this. If you double the size of the model, you double the size of your brain, you need twice as much information to go fill it. And so every time you double your parameter count, you also have to appropriately increase your training token count. The combination of those two numbers becomes the computation scale you have to support. The latest, the state of the art open AI model is approximately 1.8 trillion parameters. 1.8 trillion parameters required several trillion tokens to go train. So a few trillion parameters on the order of, a few trillion tokens on the order of, when you multiply the 2 of them together, approximately 30, 40, 50 1000000000, quadrillion floating point operations per second. Now, we just have to do some CO math right now. Just hang hang with me.

現在，我們如何可能通過每6個⽉翻倍，就將產業發展、 將計算需求提升到這麼⾼的程度呢？原因很簡單。如果你 將模型的⼤⼩加倍，你就需要將⼤腦的⼤⼩加倍，需要兩 倍的資訊來填滿它。因此，每次你將參數數ᰁ加倍，你也 必須適當增加你的訓練標記數ᰁ。這兩個數字的組合成為 你必須⽀持的計算規模。⽬前最先進的 Open AI 模型擁 有約 1.8 兆個參數。1.8 兆個參數需要數兆個標記來進⾏ 訓練。因此，當你將這兩者相乘時，⼤約是每秒 30、 40、50億、1000億浮點運算。現在，我們只需要做⼀些 CO 的數學。跟著我⼀起。

So you have 30,000,000,000 quadrillion. A quadrillion is like a peta. And so if you had a peta flop GPU, you would need 30,000,000,000 seconds to go compute, to go train that model. 30,000,000,000 seconds is approximately 1000 years. Well, 1000 years, it's worth it. Like to do it sooner, but it's worth it. It was just usually my answer when most people tell me, hey, how long how long is it gonna take to do something? 20 years. It's worth it. But can we do it next week?

所以你有30,000,000,000兆。⼀兆就像是⼀個千兆。如果 你有⼀個千兆的GPU，你將需要30,000,000,000秒來計 算、訓練那個模型。30,000,000,000秒⼤約是1000年。 嗯，1000年，值得。想要更快做完，但是值得。當⼤多 數⼈問我，嘿，要花多久時間做這件事？20年。值得。 但我們可以下週做嗎？

And so, 1000 years 1000 years. So what we need what we need are bigger GPU's. We need much, much bigger GPU's. We recognize this early on. We realized that the answer is to put a whole bunch of gpu's together. And of course, innovate a whole bunch of things along the way, like inventing tensor cores, advancing NV link so that we could create essentially virtually giant GPU's and connecting them all together with amazing networks from a company called Mellanox InfiniBand so that we could create these giant systems. And so DGX-one was our first version, but it wasn't the last. We built we built supercomputers all the way, all along the way. In 2021, we had Selene, 4,500 GPUs or so. And then in 2023, we built one of the largest AI supercomputers in the world.

因此，1000年，1000年。所以我們需要的是更⼤的 GPU。我們需要更⼤、更⼤的GPU。我們很早就意識到 這⼀點。我們意識到答案是將⼀堆GPU放在⼀起。當 然，我們還在這過程中創新了許多事物，比如發明了張ᰁ 核⼼，推進了NV連結，這樣我們就能創建基本上是虛擬 的巨⼤GPU，並將它們全部連接在⼀起，使⽤⼀家名為 Mellanox InfiniBand的公司提供的驚⼈網路，這樣我們就 能創建這些巨⼤系統。因此，DGX-one是我們的第⼀個 版本，但不是最後⼀個。⼀路上，我們建造了超級電腦。 到了2021年，我們有了Selene，⼤約4500個GPU。然後 在2023年，我們建造了世界上最⼤的⼈⼯智慧超級電腦 之⼀。

It's just come online, EOS. And as we're building these things, we're trying to help the world build these things. And in order to help the world build these things, we gotta build them first. We build the chips, the systems, the networking, all of the software necessary to do this. You should see these systems. Imagine writing a piece of software that runs across the entire system, distributing the computation across thousands of GPU's, but inside are thousands of smaller GPUs. Millions of GPUs to distribute work across all of that and to balance the workload so that you can get the most energy efficiency, the best computation time, keep your cost down. And so those those fundamental innovations is what got us here. And here we are As we see the miracle of ChatGPT emerge in front of us, we also realize we have a long ways to go. We need even larger models.

EOS 剛上線。在我們建立這些東⻄的同時，我們也試著 幫助世界建立這些東⻄。為了幫助世界建立這些東⻄，我 們必須先建立它們。我們建立晶片、系統、網路，以及所 有必要的軟體。你應該看看這些系統。想像⼀下，寫⼀個 軟體可以在整個系統上運⾏，將計算分佈在數千個 GPU 上，但內部卻有數千個更⼩的 GPU。數百萬個 GPU 分 配⼯作，平衡⼯作負載，以便獲得最⾼的能源效率、最佳 的計算時間，並降低成本。這些基本的創新是讓我們走到 現在的原因。當我們看到 ChatGPT 的奇蹟在我們⾯前展 現時，我們也意識到我們還有很長的路要走。我們需要更 ⼤的模型。

We're going to train it with multi modality data, not just text on the internet. But we're gonna we're gonna train it on text and images and graphs and charts. And just as we learn, watching TV. And so there's gonna be a whole bunch of watching video so that these models can be grounded in physics, understands that an arm doesn't go through a wall. And so these models would have common sense by watching a lot of the world's video combined with a lot of the world's languages. It'll use things like synthetic data generation, just as you and I do. When we try to learn, we might use our imagination to simulate how it's going to end up, just as I did when I was preparing for this keynote. I was simulating it all along the way. I hope it's going to turn out as well as I had it in my head. As I was simulating how this keynote was going to turn out, somebody did say that another performer did her performance completely on a treadmill, so that she could be in shape to deliver it with full energy.

我們將使⽤多模態數據來訓練它，不僅僅是網路上的⽂ 字。我們將訓練它使⽤⽂字、圖片、圖表。就像我們學習 時看電視⼀樣。因此將會有⼤ᰁ觀看影片，讓這些模型能 根據物理學建立基礎，了解⼿臂不能穿過牆壁。這些模型 將透過觀看⼤ᰁ世界影片結合⼤ᰁ世界語⾔來具備常識。 它將使⽤像是合成數據⽣成的⽅式，就像你我⼀樣。當我 們嘗試學習時，我們可能會使⽤想像⼒來模擬結果，就像 我為這場主題演講做準備時所做的⼀樣。我⼀路上都在模 擬。我希望結果能如同我腦海中所想的那樣順利。當我模 擬這場主題演講的結果時，有⼈提到另⼀位表演者完全在 跑步機上表演，這樣她就能保持體⼒充沛地表演。

I didn't do that. If I get a little wind at about 10 minutes into this, you know what happened. And so so where were we? We're sitting here using synthetic data generation. We're gonna use reinforcement learning. We're gonna practice it in our mind. We're gonna have AI working with AI training each other, just like student teacher debaters. All of that is going to increase the size of our model. It's going to increase the amount of data that we have. And we're going to have to build even bigger GPU's.

我沒有做那件事。如果我在這個過程中⼤約⼗分鐘後有點 領悟，你知道發⽣了什麼。那麼我們現在在哪裡呢？我們 坐在這裡使⽤合成數據⽣成。我們將使⽤強化學習。我們 將在腦海中練習它。我們將讓⼈⼯智慧與⼈⼯智慧互相訓 練，就像學⽣和老師辯論者⼀樣。所有這些都將增加我們 模型的⼤⼩。這將增加我們擁有的數據ᰁ。我們將不得不 建造更⼤的GPU。

Hopper is fantastic, but we need bigger GPUs. And so, ladies and gentlemen, I would like to introduce you to a very, very big GPU. Named after David Blackwell, mathematician, game theorists, probability. We thought it was a perfect per per perfect name. Blackwell, ladies and gentlemen, enjoy this. Blackwell is not a chip, Blackwell is the name of a platform. People think we make GPU's, and we do, but GPU's don't look the way they used to. Here here's the here's the here's the the, if you will, the heart of the Blackwell system. And this inside the company is not called Blackwell. It's just a number.

霍普真是太棒了，但我們需要更⼤的GPU。因此，女⼠ 們先⽣們，我想向⼤家介紹⼀個非常非常⼤的GPU。以 數學家、博弈論專家、概率學家⼤衛·布萊克威爾命名。 我們認為這是⼀個完美的名字。布萊克威爾，女⼠們先⽣ 們，請享受這個。布萊克威爾不是⼀個晶片，布萊克威爾 是⼀個平台的名字。⼈們認為我們製造GPU，我們確實 是，但GPU的外觀已經不同以往。這裡，這裡，這裡， 如果你願意的話，這是布萊克威爾系統的核⼼。在公司內 部，這不叫布萊克威爾。這只是⼀個編號。

And, this this is Blackwell sitting next to oh, this is the most advanced GPU in the world in production today. This is Hopper. This is Hopper. Hopper changed the world. This is Blackwell. It's okay, Hopper. You're you're very good. Good good boy. Good girl. 208,000,000,000 transistors, And so so you could see you I can see that there's a small line between 2 dies.

這是布萊克威爾坐在旁邊，哦，這是當今世界上⽣產中最 先進的GPU。這是霍普。這是霍普。霍普改變了世界。 這是布萊克威爾。沒關係，霍普。你很棒。好好的男孩。 好女孩。2080億個晶體管，所以你可以看到在2個晶片之 間有⼀條細⼩的線。

This is the first time 2 dies have abutted like this together in such a way that the 2 the 2 dies think it's 1 chip. There's 10 terabytes of data between it, 10 terabytes per second, So that these two these two sides of the Blackwell chip have no clue which side they're on. There's no memory locality issues, no cache issues. It's just one giant chip. So, when we were told that Blackwell's ambitions were beyond the limits of physics, the engineer said, so what? And so this is what what happened. And so this is the Blackwell chip, and it goes into 2 types of systems. The first one is form fit function compatible to hopper, And so you slide on hopper and you push in Blackwell. That's the reason why one of the challenges of ramping is going to be so efficient. There are installations of hoppers all over the world and they could be they could be, you know, the same infrastructure, same design, the power, the electricity, the thermals, the software, identical, push it right back.

這是第⼀次兩個晶片這樣緊密地相連在⼀起，以⾄於這兩 個晶片認為它們是⼀個晶片。它們之間有10TB的數據， 每秒10TB的速度，所以這兩個Blackwell晶片的兩側都不 知道⾃⼰在哪⼀邊。沒有記憶體局部性問題，也沒有快取 問題。它就是⼀個巨⼤的晶片。所以，當我們被告知 Blackwell的ᰀ⼼超越了物理極限時，⼯程師說，那⼜怎 樣？這就是發⽣的事情。這就是Blackwell晶片，它適⽤ 於兩種系統。第⼀種是與hopper相容的形狀適合功能， 所以你可以將hopper滑上去，然後將Blackwell推入。這 就是為什麼提升的挑戰之⼀將會如此⾼效。全世界各地都 有hopper的安裝，它們可能是相同的基礎設施、相同的 設計、電⼒、電⼒、熱⼒學、軟體，完全相同，直接推回 去。

And so this is a hopper version for the current HGX configuration. And this is what the other the second hopper looks like this. Now, this is a prototype board. And, Janine, could I just borrow? Ladies and gentlemen, Janine Paul. And so this this is the this is a fully functioning board. And I'll I'll just be careful here. This right here is, I don't know, $10,000,000,000 The second one's 5. It gets cheaper after that. So any customers in the audience, it's okay.

這是⽬前HGX配置的跳板版本。這是另⼀個第⼆個跳板 的樣⼦。現在，這是⼀塊原型板。⽽，Janine，我可以借 ⼀下嗎？女⼠們先⽣們，這位是Janine Paul。這是⼀塊 完全運作正常的板⼦。我會⼩⼼⼀點。這裡就是，我不知 道，⼗億美元。第⼆個是五億。之後價格就會更便宜了。 所以在座的客⼾們，別擔⼼。

All right. But this one's quite expensive. This is the bring up board. And, and the the way it's going to go to production is like this one here. Okay? And so you're going to take take this. It has 2 Blackwell die 2 2 Blackwell chips and 4 Blackwell dies connected to a Grace CPU. The Grace CPU has a superfast chip to chip link. What's amazing is this computer is the first of its kind where this much computation, first of all, fits into this small of a place. 2nd, it's memory coherent.

好的。但這個價格相當昂貴。這是提升板。⽽且，它將會 以這樣的⽅式進入⽣產。好嗎？所以你要拿這個。它有2 個Blackwell晶片和4個連接到Grace CPU的Blackwell晶 片。Grace CPU具有超快速的晶片連結。令⼈驚奇的 是，這台電腦是第⼀台這樣的電腦，⾸先，這麼多的運算 能⼒，⾸先，能夠放進這麼⼩的空間。其次，它是記憶體 ⼀致的。

They feel like they're just 1 big happy family working on one application together. And so, everything is coherent within it. The just the amount of, you know, you saw the numbers. There's a lot of terabytes this and terabytes that's. But this is this is a miracle. This is a, this let's see. What are some of the things on here? There's, NVLink on top, PCI Express on the bottom. On on, your which one is mine and your left? One of them it doesn't matter.

他們感覺彷彿是⼀個⼤家庭⼀起合作開發⼀個應⽤程式。 因此，在其中⼀切都是⼀致的。你看到數字了，有很多的 「TB」這個、「TB」那個。但這是⼀個奇蹟。這是⼀ 個，讓我們看看。這裡有⼀些什麼東⻄？頂部有 NVLink，底部有 PCI Express。在你的左邊或右邊？其 中⼀個並不᯿要。

One of the one of them is a c CPU chip to chip link. It's my left or your depending on which side. I was just I was trying to sort that out and I just kinda doesn't matter. Hopefully, it comes plugged in. So okay. So this is the Grace Blackwell system. But there's more. So it turns out, it turns out, all of the specs is fantastic, but we need a whole lot of new features. In order to push the limits beyond, if you will, the limits of physics, we would like to always get a lot more X factors. And so one of the things that we did was we invented another transformer engine, another transformer engine, the 2nd generation.

其中之⼀是⼀個 CPU 芯片到芯片的連結。這是在我的左 邊或你的左邊，取決於哪⼀邊。我剛剛試著弄清楚這⼀ 點，但其實也沒有太⼤關係。希望它已經插好了。好的， 這就是 Grace Blackwell 系統。但還有更多。原來，所有 的規格都很棒，但我們需要很多新功能。為了突破物理的 極限，我們想要總是獲得更多的 X 因素。因此，我們做 的其中⼀件事情是發明了另⼀個變壓引擎，第⼆代變壓引 擎。

It has the ability to dynamically and automatically rescale and recast numerical formats to a lower precision whenever he can. Remember, artificial intelligence is about probability. And so you kind of have, you know, 1.7, approximately 1.7 times approximately 1.4 to be approximately something else. Does that make sense? And so so the the ability for the mathematics to retain the precision and the range necessary in that particular stage of the pipeline, super important. And so this is it's not just about the fact that we designed a smaller ALU. It's not quite the world's not quite that simple. You've got to figure out when you can use that across a computation that is thousands of GPUs. It's running for weeks weeks weeks, and you want to make sure that the the, the training job is going to converge. And so this new transformer engine, we have a 5th generation NVLink.

它具有動態且⾃動地將數值格式᯿新調整和轉換為較低精 度的能⼒。請記住，⼈⼯智慧是關於機率。所以你可以⼤ 概有，你知道，1.7⼤約乘以1.4⼤約等於⼤約其他數字。 這樣講對嗎？所以數學能夠保留在管線的特定階段所需的 精確度和範圍的能⼒非常᯿要。這不僅僅是因為我們設計 了⼀個較⼩的ALU。世界並不完全如此簡單。你必須弄清 楚在數千個GPU上的計算中何時可以使⽤它。這些計算 運⾏了幾個星期，你希望確保訓練⼯作將會收斂。因此， 這個新的轉換引擎，我們有第五代的NVLink。

It's now twice as fast as hopper, but very importantly, it has computation in the network. And the reason for that is because when you have so many different GPUs working together, we have to share our information with each other. We have to synchronize and update each other. And every so often, we have to reduce the partial products and then rebroadcast out the partial products, the sum of the partial products back to everybody else. And so there's a lot of what is called all reduce and all to all and all gather. It's all part of this area of synchronization and collectives so that we can have GPUs working with each other. Having extraordinarily fast links and being able to do mathematics right in the network allows us to essentially amplify even further. So even though it's 1.8 terabytes per second, it's effectively higher than that. And so it's many times that of Hopper. The likelihood of a supercomputer running for weeks on end is approximately 0.

現在的速度是跳躍者的兩倍，但非常᯿要的是，它在網路 中進⾏計算。這是因為當有這麼多不同的GPU⼀起運作 時，我們必須彼此分享資訊。我們必須進⾏同步並更新彼 此。⽽且每隔⼀段時間，我們必須減少部分產品，然後᯿ 新廣播部分產品，將部分產品的總和᯿新廣播給其他⼈。 因此，有很多所謂的全局減少、全局到全局和全局收集。 這都是同步和集體的⼀部分，這樣我們就可以讓GPU彼 此合作。擁有非常快速的連結並能夠在網路中進⾏數學運 算，讓我們能夠進⼀步放⼤效果。因此，即使是每秒 1.8TB，實際上它的速度更⾼。因此，它比Hopper快很多 倍。⼀台超級電腦連續運⾏數週的可能性⼤約是0。

And the reason for that is because there's so many components working at the same time. The statistic, the probability of them working continuously is very low. And so we need to make sure that whenever there is a well, we checkpoint and restart as often as we can. But if we have the ability to detect a weak chip or a weak note early, we could retire it and maybe swap in another processor. That ability to keep the utilization of the super computer high, especially when you just spent $2,000,000,000 building it, is super important. And so we put in a RAS engine, a reliability engine that does a 100% self test, in system test, of every single gate, every single bit of memory on the Blackwell chip, and all the memory that's connected to it. It's almost as if we shipped with every single chip its own advanced tester that we test our chips with. This is the first time we're doing this. Super excited about it. Secure AI.

這是因為有這麼多元件同時運作的原因。這些元件持續運 作的機率非常低。因此，我們需要確保每當有⼀個元件出 現問題時，我們都要盡可能地進⾏檢查點和᯿新啟動。但 如果我們有能⼒早期檢測到⼀個弱晶片或弱筆記，我們可 以將其淘汰，並可能更換另⼀個處理器。保持超級電腦的 利⽤率⾼的能⼒，尤其是當你剛花了20億美元建造它 時，是非常᯿要的。因此，我們加入了⼀個RAS引擎， ⼀個可靠性引擎，對Blackwell晶片上的每個閘，每個記 憶體位元進⾏100%的⾃我測試，系統測試，以及所有連 接到它的記憶體。這幾乎就像我們隨每個晶片⼀起出貨的 先進測試器，我們⽤它來測試我們的晶片。這是我們第⼀ 次這樣做。對此感到非常興奮。安全⼈⼯智慧。

Only this conference do they clap for RAS. The the, Secure AI. Obviously, you've just spent 100 of 1,000,000 of dollars creating a very important AI, and the the code, the intelligence of that AI is encoded in the parameters. You want to make sure that on the one hand, you don't lose it. On the other hand, it doesn't get contaminated. And so we now have the ability to encrypt data, of course, at rest, but also in transit and while it's being computed. It's all encrypted. And so, we now have the ability to encrypt in transmission. And when we're computing it, it is in a trusted trusted environment, trusted, engine environment. And the last thing is decompression.

只有在這個會議上，他們才為RAS⿎掌。當然，你剛花 了100萬美元中的⼀⼩部分來創建⼀個非常᯿要的AI，⽽ 這個AI的代碼、智能都被編碼在參數中。你希望⼀⽅⾯確 保不會丟失，另⼀⽅⾯不會被污染。因此，我們現在有能 ⼒加密數據，當然，不僅是在靜⽌狀態下，還包括在傳輸 過程中以及計算時。所有都是加密的。因此，我們現在有 能⼒在傳輸中加密。當我們進⾏計算時，它處於⼀個受信 任的環境中，⼀個受信任的引擎環境。最後⼀點是解壓 縮。

Moving data in and out of these nodes when the compute is so fast becomes really essential. And so we've put in a high line speed compression engine and effectively moves data 20 times faster in and out of these computers. These computers are are so powerful and they're such a large investment. The last thing we wanna do is have them be idle. And so all of these capabilities are intended to keep Blackwell fed and as busy as possible. Overall, compared to Hopper, it is 2 and a half times, 2 and a half times the FPA performance for training per chip. It is also it also has this new format called fp6 so that even though the computation speed is the same, the bandwidth that's amplified because of the memory, the amount of parameters you can store in the memory is now amplified. Fp4 effectively doubles the throughput. This is vitally important for inference. One of the things that that, is becoming very clear is that whenever you use a computer with AI on the other side, when you're chatting with the chatbot, when you're asking it to, review or make an image, Remember, in the back is a GPU generating tokens.

在這些節點之間快速移動數據變得非常᯿要。因此，我們 加入了⾼速壓縮引擎，有效地使數據在這些電腦之間的傳 輸速度提⾼了20倍。這些電腦非常強⼤，也是⼀項巨⼤ 的投資。我們最不希望的是讓它們閒置。因此，所有這些 功能都旨在確保 Blackwell 被充分利⽤，並保持盡可能忙 碌。總的來說，與 Hopper 相比，每個芯片的訓練性能提 ⾼了2.5倍。它還引入了⼀種新的格式叫做 fp6，即使計 算速度相同，由於記憶體的增強，可以存儲在記憶體中的 參數ᰁ也得到了增加。Fp4 可以將吞吐ᰁ翻倍。這對於推 理來說⾄關᯿要。⼀個非常明顯的事實是，每當您使⽤具 有 AI 的電腦時，當您與聊天機器⼈交談時，當您要求它 審查或製作圖像時，請記住，在背後是⼀個 GPU ⽣成標 記。

Some people call it inference, but it's more appropriately generation. The way that computing has done in the past was retrieval. You would grab your phone, you would touch something. Some signals go off. Basically, an email goes off to some storage somewhere. There's pre recorded content. Somebody wrote a story or somebody made an image or somebody recorded a video. That record pre recorded content is then streamed back to the phone and recomposed in a way based on a recommender system to present the information to you. You know that in the future, the vast majority of that content will not be retrieved. And the reason for that is because that was prerecorded by somebody who doesn't understand the context, which is the reason why we have to retrieve so much content.

有些⼈稱之為推論，但更適當的應該是⽣成。過去計算的 ⽅式是檢索。你會拿起你的⼿機，觸碰某個東⻄。⼀些信 號會發出。基本上，⼀封電⼦郵件會發送到某個存儲位 置。有預先錄製的內容。有⼈寫了⼀個故事，有⼈製作了 ⼀張圖片，或者有⼈錄製了⼀段影片。這個預先錄製的內 容然後被串流回⼿機，並根據推薦系統以某種⽅式᯿新組 合，將信息呈現給你。你知道在未來，絕⼤多數的內容將 不會被檢索。⽽這是因為那是由⼀個不了解上下⽂的⼈預 先錄製的，這就是為什麼我們必須檢索這麼多內容的原 因。

If you can be working with an AI that understands the context, who you are, for what reason you're fetching this information, and produces the information for you just the way you like it. The amount of energy we save, the amount of networking bandwidth we save, the amount of waste of time we save will be tremendous. The future is generative, which is the reason why we call it generative AI, which is the reason why this is a brand new industry. The way we compute is fundamentally different. We created a processor for the generative AI era, And one of the most important parts of it is content token generation. We call it this format is f p 4. Well, that's a lot of computation. 5x, the token generation, 5x, the inference capability of Hopper seems like enough. But why stop there? The answer is, it's not enough.

如果你能夠與⼀個了解上下⽂、知道你是誰、為什麼要取 得這些資訊並且按照你喜歡的⽅式提供資訊的⼈⼯智慧合 作，我們所節省的能源、網路頻寬以及時間浪費將是巨⼤ 的。未來是⽣成的，這就是為什麼我們稱之為⽣成式⼈⼯ 智慧，這也是為什麼這是⼀個全新的產業。我們運算的⽅ 式從根本上就不同了。我們為⽣成式⼈⼯智慧時代創造了 ⼀個處理器，其中最᯿要的部分之⼀是內容標記⽣成。我 們稱這種格式為 f p 4。這需要⼤ᰁ的運算。5倍的標記⽣ 成，5倍的 Hopper 推理能⼒似乎已經⾜夠了。但為什麼 要停在這裡呢？答案是，這還不夠。

I'm going to show you why. I'm going to show you why. So we would like to have a bigger GPU, even bigger than this one. So we decided to scale it and notice, but first let me just tell you how we've scaled. Over the course of the last 8 years, we've increased computation by 1,000 times. 8 years, 1,000 times. Remember back in the good old days of Moore's law, it was 2 x well, 5 x every what, 10 10 x every 5 years. That's easiest easiest math. 10 x every 5 years, a 100 times every 10 years, 100 times every 10 years at the in the middle of the heydays of the PC revolution. 100 times every 10 years.

我要向你展⽰為什麼。我要向你展⽰為什麼。所以我們希 望有⼀個更⼤的 GPU，甚⾄比這個還要⼤。所以我們決 定要擴⼤它，但⾸先讓我告訴你我們是如何擴⼤的。在過 去的8年裡，我們將計算能⼒提升了1000倍。8年，1000 倍。還記得在摩爾定律的黃⾦時代，是2倍，對，每5年5 倍。這是最簡單的數學。每5年10倍，每10年100倍，在 個⼈電腦⾰命的鼎盛時期。每10年100倍。

In the last 8 years, we've gone 1,000 times. We have 2 more years to go. And so, that puts it in perspective. The rate at which we're advancing computing is insane. And it's still not fast enough, so we built another chip. This chip is just an incredible chip. We call it the NVLink switch. It's 50,000,000,000 transistors. It's almost the size of hopper all by itself. This switch chip has 4 NV links in it, each 1.8 terabytes per second, and, and it has computation in it, as I mentioned.

在過去的8年裡，我們已經前進了1,000倍。我們還有2年 的時間。這樣⼀來，這個情況就變得清楚了。我們在推進 計算的速度是瘋狂的。但這速度還不夠快，所以我們⼜建 造了另⼀個晶片。這個晶片是⼀個令⼈難以置信的晶片。 我們稱它為NVLink交換器。它擁有50億個電晶體。它幾 乎和霍普⼀樣⼤。這個交換器晶片裡有4個NV連結，每個 速度達到1.8 TB每秒，⽽且它還有計算功能，正如我所提 到的。

What is this chip for? If we were to build such a chip, we can have every single GPU talk to every other GPU at full speed at the same time. That's insane. It doesn't even make sense. But if you could do that, if you can find a way to do that and build a system to do that, that's cost effective, That's cost effective. How incredible would it be that we could have all these GPU's connect over a coherent link so that they effectively are 1 giant GPU. Well, one of the great inventions in order to make it cost effective is that this chip has to drive copper directly. The SerDes of this chip is just a phenomenal invention so that we could do direct drive to copper. And as a result, you can build a system that looks like this. Is kind of insane.

這個晶片是⽤來做什麼的？如果我們要建造這樣⼀個晶 片，我們可以讓每個 GPU 以全速同時與其他 GPU 進⾏ 通訊。這太瘋狂了。甚⾄沒有道理。但如果你能做到這⼀ 點，如果你能找到⼀種⽅法並建造⼀個系統來實現這⼀ 點，⽽且成本效益⾼，那將是多麼不可思議。我們可以讓 所有這些 GPU 通過⼀個連貫的連結相連，以便它們有效 地成為⼀個巨⼤的 GPU。為了使其成本效益⾼，其中⼀ 個偉⼤的發明是這個晶片必須直接驅動銅。這個晶片的 SerDes 只是⼀個了不起的發明，這樣我們就可以直接驅 動銅。因此，你可以建造⼀個看起來像這樣的系統。有點 瘋狂。

This is 1 DGX. This is what a DGX looks like now. Remember, just 6 years ago, it was pretty heavy, but I was able to lift it. I delivered the, the first dgX1 to OpenAI, and and the researchers there, it's on, you know, the pictures are on the internet. And we all autographed it. And if you come to my office, it's autographed there. It's really beautiful. And but but you can lift it. This DGX this DGX that DGX, by the way, was 170 teraflops. If you're not familiar with the numbering system, that's 0.17 petaflops.

這是⼀台 DGX。這就是現在的 DGX 外觀。記得，僅僅 六年前，它相當沉᯿，但我還是能舉起它。我將第⼀台 DGX1 交付給 OpenAI，那裡的研究⼈員，你知道，照片 都在網路上。我們都在上⾯簽名。如果你來到我的辦公 室，你會看到上⾯有簽名。它真的很漂亮。但你可以舉起 它。這台 DGX，順帶⼀提，有 170 兆浮點運算。如果你 不熟悉這個數字系統，那就是 0.17 拍字節。

So this is 720. The first one I delivered to open AI was 0.17. You could round it up to 0.2, won't make any difference. But and back then, it was like, wow, you know, 30 more teraflops. And so, this is now 720 petaflops, almost an exaflop for training and the world's first one exaflops machine in one rack. Just so you know, there are only a couple 2, 3 exaflops machines on the planet as we speak. And so this is an exaflops AI system in one single rack. Well, let's take a look at the back of it. So this is what makes it possible. That's the back that's the that's the back, the d g x m v link spine.

這是720。我第⼀次交付給OpenAI的是0.17。你可以四捨 五入為0.2，不會有任何差別。但那時候，就像，哇，你 知道，多了30個teraflops。所以，現在這是720 petaflops，幾乎是⼀個exaflop⽤於訓練，世界上第⼀個 ⼀個機架的exaflops機器。只是讓你知道，⽬前全球只有 2、3個exaflops機器。所以這是⼀個exaflops的AI系統在 ⼀個機架中。好吧，讓我們來看看它的背⾯。這就是使其 成為可能的原因。那就是背⾯，那就是背⾯，d g x m v link spine。

A 130 terabytes per second goes to the back of that chassis. That is more than the aggregate bandwidth of the Internet. So we we could basically send everything to everybody within a second. And so so we we have 5,000 cables 5000 NVLink cables. In total, 2 miles. Now, this is the amazing thing. If we had to use optics, we would have had to use transceivers and retimers. And those transceivers and retimers alone would have cost 20,000 watts, 2 kilowatts of just transceivers alone, just to drive the NVLink spine. As a result, we did it completely for free over NVLink switch, and we were able to save the 20 kilowatts for computation. This entire rack is a 120 kilowatts.

每秒有130 TB 的數據流ᰁ進入機箱後部。這比整個互聯 網的總帶寬還要⾼。所以基本上我們可以在⼀秒內將所有 資料發送給每個⼈。我們有5000條 NVLink 電纜。總長度 為2英⾥。現在，這是令⼈驚奇的事情。如果我們必須使 ⽤光學傳輸，我們將不得不使⽤收發器和᯿定時器。這些 收發器和᯿定時器單獨的功耗就要 20,000 瓦，僅僅是驅 動 NVLink 主幹的收發器就要 2 千瓦。因此，我們完全免 費地通過 NVLink 交換機完成了這項任務，我們能夠節省 20 千瓦⽤於計算。整個機架的功耗為 120 千瓦。

So that 20 kilowatts makes a huge difference. It's liquid cooled. What goes in is 25 degrees c about room temperature. What comes out is 45 degrees C about your Jacuzzi. So room temperature goes in, Jacuzzi comes out, 2 liters per second. We could we could sell a peripheral. 600,000 parts. Somebody used to say, you know, you guys make GPU's, and we do, but this is what a GPU looks like to me. When somebody says GPU, I see this. 2 years ago, when I saw a GPU, it was the HGX.

這20千瓦的功率差異非常巨⼤。它是液冷的。進去的溫 度是攝⽒25度，⼤約是室溫。出來的溫度是攝⽒45度， 就像你的按摩浴缸。進去是室溫，出來是按摩浴缸的溫 度，每秒2公升。我們可以賣⼀個周邊設備。600,000個 零件。有⼈過去常說，你們製造GPU，我們確實是，但 對我來說，GPU看起來是這樣。當有⼈說GPU時，我看 到的是這個。兩年前，當我看到GPU時，那是HGX。

It was £70, 35,000 parts. Our GPUs now are 600,000 parts and £3,000 £3,000 £3,000, that's kinda like the weight of a, you know, carbon fiber Ferrari. I don't know if that's useful metric, but everybody's going, I feel it. I feel it. I get it. I get that. Now that you mentioned that, I feel it. I don't know what's £3,000. Okay. So £3,000, ton and a half.

這個 GPU 現在有 600,000 個零件，價值 3,000 英鎊，這 種᯿ᰁ有點像碳纖維法拉利。我不確定這是否是⼀個有⽤ 的衡ᰁ標準，但每個⼈都感受到了。現在你提到了，我感 受到了。我感受到了。我明⽩了。我明⽩了。現在你提到 了，我感受到了。我不知道 3,000 英鎊是什麼。好吧， 3,000 英鎊，⼀噸半。

So it's not quite an elephant. So this is what a DGX looks like. Now let's see what it looks like in operation. Okay. Let's imagine what is what how do we put this to work and what does that mean? Well, if you were to train a GPT model, 1.8 trillion parameter model, It took it took about apparently about, you know, 3 to 5 months or so, with 25,000 amperes. If we were to do it with Hopper, it would probably take something like 8,000 GPUs and it would consume 15 megawatts 8,000 GPUs and 15 megawatts. It would take 90 days, about 3 months. And that would allow you to train something that is, you know, this groundbreaking AI model. And this is obviously not as expensive as, as anybody would think, but it's 8,008,000 GPUs.

這並不完全是⼀隻⼤象。這就是 DGX 的外觀。現在讓我 們看看它在運作時是什麼樣⼦。好的。讓我們想像⼀下， 我們如何將這個應⽤起來，這意味著什麼？如果你要訓練 ⼀個 GPT 模型，⼀個有 1.8 兆個參數的模型，顯然需要 ⼤約 3 到 5 個⽉的時間，使⽤ 25,000 安培。如果我們使 ⽤ Hopper 來做，可能需要⼤約 8,000 個 GPU，並且消 耗 15 兆瓦的電⼒。8,000 個 GPU 和 15 兆瓦的電⼒。這 將需要 90 天，⼤約 3 個⽉的時間。這將讓你能夠訓練出 ⼀個具有突破性的 AI 模型。顯然這並不像任何⼈想的那 麼昂貴，但需要 8,008,000 個 GPU。

It's still a lot of money. So 8,000 GPUs, 15 megawatts. If you were to use Blackwell to do this, it would only take 2,000 GPUs. 2 1,000 GPU's, same 90 days, but this is the amazing part. Only 4 megawatts of power. So from 15 yeah. That's right. And that's and that's our goal. Our goal is to continuously drive down the cost and the energy that are directly proportional to each other, cost and energy associated with the computing so that we can continue to expand and scale up the computation that we have to do to train the next generation models. Well, this is training.

這仍然是⼀筆龐⼤的⾦額。所以 8,000 個 GPU，15 兆 瓦。如果你使⽤ Blackwell 來做這件事，只需要 2,000 個 GPU。2 個 1,000 個 GPU，相同的 90 天，但這是驚⼈ 的部分。只需要 4 兆瓦的電⼒。所以從 15 兆瓦變成 4 兆 瓦。沒錯。這就是我們的⽬標。我們的⽬標是持續降低成 本和能源，這兩者是成正比的，與計算相關的成本和能 源，這樣我們就可以繼續擴⼤和擴展我們必須進⾏的訓練 下⼀代模型的計算。嗯，這就是訓練。

Inference, or generation, is vitally important going forward. You know, probably some half of the time that NVIDIA GPUs are in the cloud these days, it's being used for token generation. You know, they're either doing copilot this or chat, you know, chat gpt that or, all these different models that are being used when you're interacting with it or generating generating images or generating videos, generating proteins, generating chemicals. There's a bunch of generation going on. All of that is in the category of computing we call inference. But inference is extremely hard for large language models because these large language models have several properties. One, they're very large. And so it doesn't fit on 1 GPU. This is imagine, imagine Excel doesn't fit on 1 GPU. You know?

推論或⽣成在未來將變得⾄關᯿要。你知道，可能有⼀半 的時間，如今 NVIDIA 的 GPU 在雲端中使⽤時，都是⽤ 來進⾏代幣⽣成。你知道，他們要麼在做 copilot，要麼 在聊天，你知道，聊天 gpt 這樣，或者，當你與之互動時 使⽤的所有這些不同模型，⽣成圖像或⽣成視頻，⽣成蛋 ⽩質，⽣成化學物質。有⼤ᰁ的⽣成正在進⾏。所有這些 都屬於我們所謂的推論計算範疇。但對於⼤型語⾔模型來 說，推論是非常困難的，因為這些⼤型語⾔模型具有幾個 特性。⾸先，它們非常龐⼤。因此，無法容納在⼀個 GPU 上。這就像，想像⼀下，Excel 不能容納在⼀個 GPU 上。你知道？

And imagine some application you're running on a daily basis doesn't run doesn't fit on 1 computer. Like a video game doesn't fit on 1 computer. And most, in fact, do. And many times in the past, in hyperscale computing, many applications for many people fit on the same computer. And now, all of a sudden, this one inference application where you're interacting with this chatbot, that chatbot requires a supercomputer in the back to run it. And that's the future. The future is generative with these chatbots and these chatbots are trillions of tokens, trillions of parameters. And they have to generate tokens at interactive rates. Now, what does that mean? Oh, well, 3 tokens is about a word.

想像⼀下，有些你每天使⽤的應⽤程式無法在⼀台電腦上 運⾏。就像⼀個電玩遊戲無法在⼀台電腦上運⾏⼀樣。事 實上，⼤多數情況下是可以的。在過去的超⼤規模運算 中，許多⼈的應⽤程式可以在同⼀台電腦上運⾏。⽽現 在，突然間，這個推論應⽤程式需要背後的超級電腦來運 ⾏。這就是未來。未來是與這些聊天機器⼈⼀起⽣成，這 些聊天機器⼈擁有數以兆計的標記、參數。它們必須以互 動速率⽣成標記。那是什麼意思呢？哦，3個標記⼤約是 ⼀個字。

You know, the the, you know, space, the final frontier, these are the adventures that that's like that's like 80 tokens. Okay? I don't know if that's useful to you. And so, you know, the art of communications is is selecting good good analogies. Yeah. This is this is not going well. There is, I don't know what he's talking about. Never seen Star Trek. And so and so so here, we are. We're trying to generate these tokens.

你知道，太空，這是最後的邊疆，這些都是冒險，就像是 80個代幣。好嗎？我不知道這對你有沒有⽤。所以，你 知道，溝通的藝術就是選擇好的比喻。是的。這進展得不 太順利。我不知道他在說什麼。從來沒看過星際迷航。所 以在這裡，我們正在嘗試⽣成這些代幣。

When you're interacting with it, you're hoping that the tokens come back to you as quickly as possible and as quickly as you could read it. So the ability for generation token is really important. You have to parallelize the work of this model across many, many GPUs so that you could achieve several things. 1, on the one hand, you would like throughput because that throughput reduces the cost, the overall cost per token of, generating. So your throughput dictates the cost of delivering the service. On the other hand, you have another interactive rate which is another tokens per second where it's about per user, and that has everything to do with quality of service. And so, these two things, compete against each other. And we have to find a way to distribute work across all of these different GPUs and paralyze it in a way that allows us to achieve both. It turns out the search search space is enormous. You know, I told you there's going to be math involved.

當你與它互動時，希望令牌能盡快回應，就像你閱讀⼀樣 快。因此，⽣成令牌的能⼒非常᯿要。你必須將這個模型 的⼯作在許多許多個 GPU 上進⾏並⾏處理，這樣你才能 實現幾件事情。⾸先，你希望有較⾼的吞吐ᰁ，因為這將 降低⽣成每個令牌的整體成本。因此，你的吞吐ᰁ決定了 提供服務的成本。另⼀⽅⾯，你還有另⼀個互動速率，即 每秒⽣成的令牌數，這與服務質ᰁ有關。因此，這兩者之 間是相互競爭的。我們必須找到⼀種⽅法將⼯作分配到所 有這些不同的 GPU 上並進⾏並⾏處理，以便實現這兩 者。事實證明，搜索空間是巨⼤的。你知道，我告訴過你 這裡會涉及到數學。

And everybody's going, oh dear. I heard some gasps just now when I put up that slide. You know, so so this this right here, the the y axis is tokens per second, data center throughput. The x axis is tokens per second, interactivity of the person. Notice the upper right is the best. You want interactivity to be very high, number of tokens per second per user. You want the tokens per second of per data center to be very high, the upper upper right is is terrific. However, it's very hard to do that. In order for us to search for the best answer across every single one of those intersections, x y coordinates, Okay. So just look at every single x y coordinate.

⼤家都在說，哎呀。我剛剛放上那張投影片時聽到了⼀些 驚嘆聲。你知道，這裡，y軸是每秒的標記數，資料中⼼ 的吞吐ᰁ。x軸是每秒的標記數，使⽤者的互動性。請注 意右上⾓是最好的。你希望互動性非常⾼，每位使⽤者每 秒的標記數。你希望每個資料中⼼每秒的標記數非常⾼， 右上⾓是最棒的。然⽽，要做到這⼀點非常困難。為了找 到最佳答案，我們必須橫跨每⼀個交叉點，x和y座標。所 以只要看每⼀個x和y座標就好。

All those blue dots came from some repartitioning of the software. Some optimizing solution has to go and figure out whether to you use tensor parallel, expert parallel, pipeline parallel, or data parallel, and distribute this enormous model across all these different GPUs and sustain the performance that you need. This exploration space would be impossible if not for the programmability of NVIDIA's GPUs. And so we could because of CUDA, because we have such a rich ecosystem, we could explore this universe and find that green roof line. It turns out that green roof line notice you got tp2epadp4. It means 2 parallel 2, tensor parallel tensor parallel across 2 GPUs, expert parallels across 8, data parallel across 4. Notice on the other end, you got a tensor parallel across 4 and expert parallel across 16. The configuration, the distribution of that software, it's a different, different run time that would produce these different results. And you have to go discover that roofline. Well, that's just one model.

所有這些藍⾊點都來⾃於軟體的᯿新劃分。某個最佳化的 解決⽅案必須去判斷是要使⽤張ᰁ平⾏、專家平⾏、管線 平⾏還是資料平⾏，並將這個龐⼤的模型分佈到所有這些 不同的 GPU 上，並保持所需的效能。如果不是 NVIDIA 的 GPU 可程式化，這個探索空間將是不可能的。因此， 我們可以藉由 CUDA，因為我們擁有如此豐富的⽣態系 統，我們可以探索這個宇宙並找到那條綠⾊屋頂線。原來 那條綠⾊屋頂線表⽰你有 tp2epadp4。這意味著在 2 個 GPU 上進⾏ 2 個張ᰁ平⾏、在 8 個 GPU 上進⾏ 8 個專 家平⾏、在 4 個 GPU 上進⾏ 4 個資料平⾏。另⼀端注意 到，你有在 4 個 GPU 上進⾏ 4 個張ᰁ平⾏和在 16 個 GPU 上進⾏ 16 個專家平⾏。軟體的配置、分佈是不同 的，這將產⽣不同的運⾏時間。你必須去發現那條屋頂 線。嗯，這只是⼀個模型。

And this is just one configuration of a computer. Imagine all of the models being created around the world and all the different different, configurations of of systems that are going to be available. So now that you understand the basics, let's take a look at inference of Blackwell compared to Hopper. And this is this is the extraordinary thing. In one generation, because we created a system that's designed for trillion parameter generative AI, the inference capability of Blackwell is off the charts. And in fact, it is some 30 times hopper. Yep. For large language models for large language models like Chat GPT and others like it, the blue line is Hopper. I gave you imagine we didn't change the architecture of Hopper, we just made it a bigger chip. We just used the latest, you know, greatest 10 terabytes per second.

這只是電腦的⼀種配置。想像⼀下全球各地正在被創建的 所有型號和所有不同的系統配置將會提供的選擇。現在您 已經了解了基本知識，讓我們來看看布萊克威爾相對於霍 普的推理。這是非常了不起的事情。在⼀代⼈的時間裡， 因為我們創建了⼀個針對兆參數⽣成式⼈⼯智慧設計的系 統，布萊克威爾的推理能⼒超乎想像。事實上，它是霍普 的約30倍。對於像Chat GPT和其他類似的⼤型語⾔模 型，藍線代表霍普。您可以想像⼀下，如果我們沒有改變 霍普的架構，只是將它做成⼀個更⼤的晶片。我們只是使 ⽤了最新、最偉⼤的每秒10TB的速度。

We connected the 2 chips together. We got this giant 208,000,000,000 parameter chip. How would we have performed if nothing else changed? And it turns out, quite wonderfully. Quite wonderfully, and that's the purple line, but not as great as it could be. And that's where the f p 4 tensor core, the new transformer engine, and very importantly, the NVLink switch. And the reason for that is because all these GPU's have to share the results, partial products. Whenever they do all to all, all all gather, whenever they communicate with each other, that NVLink switch is communicating almost 10 times faster than what we could do in the past using the fastest networks. Okay. So Blackwell is going to be just an amazing system for generative AI.

我們將這兩個晶片連接在⼀起。我們得到了這個龐⼤的 208,000,000,000 參數的晶片。如果沒有其他改變，我們 會表現如何呢？結果是，表現相當出⾊。相當出⾊，這就 是紫⾊的線，但並不是最佳的。這就是 f p 4 張ᰁ核⼼、 新的變壓器引擎，以及非常᯿要的 NVLink 切換器所發揮 的作⽤。原因在於所有這些 GPU 都必須共享結果、部分 產品。每當它們進⾏全對全、全體收集時，每當它們彼此 溝通時，那個 NVLink 切換器的通訊速度幾乎比我們過去 使⽤最快的網路快了將近 10 倍。好的。所以 Blackwell 將會是⼀個非常令⼈驚嘆的⽣成式⼈⼯智慧系統。

And in the future, in the future, data centers are going to be thought of, as I mentioned earlier, as an AI factory. An AI factory's goal in life is to generate revenues. Generate, in this case, intelligence in this facility, not generating electricity as in AC generators, but of the last industrial revolution and this industrial revolution, the generation of intelligence. And so, this ability is super, super important. The excitement of Blackwell is really off the charts. You know, when we first when we first, you know, this this is a year and a half ago, 2 years ago, I guess 2 years ago when we first started to to go to market with hopper, you you know, we had the benefit of of, 22, 2 CSPs, joined us in a lunch. And and we were, you know, delighted. And so, we had 2 customers. We have more now. Unbelievable excitement for Blackwell.

未來，資料中⼼將被視為⼀座⼈⼯智慧⼯廠，正如我之前 所提到的。⼀座⼈⼯智慧⼯廠的⽣命⽬標是產⽣收入。在 這個場所中，產⽣的是智慧，⽽不是像交流發電機那樣產 ⽣電⼒，⽽是產⽣智慧，這是上⼀次⼯業⾰命和這次⼯業 ⾰命的產物。因此，這種能⼒非常非常᯿要。Blackwell 的興奮情緒真的是超乎想像。你知道，當我們第⼀次，當 我們第⼀次，這是⼀年半前，我想是兩年前，當我們第⼀ 次開始推出hopper時，我們有幸邀請到22家CSPs在午餐 會上加入我們。我們感到非常⾼興。所以，我們有兩位客 ⼾。現在我們有更多了。對於Blackwell的興奮程度令⼈ 難以置信。

Unbelievable excitement. And there's a whole bunch of different configurations. Of course, I showed you the configurations that slide into the hopper form factor so that it's easy to upgrade. I showed you examples, that are liquid cooled, that are the extreme versions of it. One entire rack that's that's connected by NVLink 672. We're gonna, Blackwell is going to be ramping to the world's AI companies, of which there are so many now, doing amazing work in different modalities. The CSPs, every CSP is geared up. All the OEMs and ODMs, regional clouds, sovereign AIs, and telcos all over the world are signing up to launch with Blackwell. This Blackwell would be the most successful product launch in our history.

And so I can't wait to see that. 令⼈難以置信的興奮。⽽且有許多不同的配置。當然，我 向你展⽰了可以輕鬆升級的滑入式機箱形式的配置。我展 ⽰了⼀些例⼦，那些是液冷的，是極端版本的。⼀整個機 架，透過 NVLink 連接 672 個。我們將會，Blackwell 將 會成為全球⼈⼯智慧公司的主⼒，現在有這麼多家公司在 不同的模式下進⾏驚⼈的⼯作。每個雲服務提供商都準備 就緒。所有的原始設備製造商、區域雲端、主權⼈⼯智慧 和全球各地的電信公司都在與 Blackwell 合作推出。這個 Blackwell 將會是我們歷史上最成功的產品推出。所以我 迫不及待想看到那⼀天。

I want to thank some partners that are joining us in this. AWS is gearing up for Blackwell. They're going to build the first GPU with secure AI. They're building out a 222 exaflops system. You know, just now when we animated, just now the the digital twin, if you saw the the all of those clusters coming down. By the way, that is not just art. That is a digital twin of what we're building. That's how big it's going to be. Besides infrastructure, we're doing a lot of things together with AWS. We're CUDA accelerating SageMaker AI.

我想要感謝⼀些與我們⼀同參與的合作夥伴。AWS 正在 為 Blackwell 做準備。他們將建造第⼀個具有安全⼈⼯智 慧的 GPU。他們正在建造⼀個 222 exaflops 的系統。你 知道，剛剛當我們做動畫時，剛剛那個數位孿⽣體，如果 你看到所有那些叢集降下來。順帶⼀提，那不僅僅是藝 術。那是我們正在建造的數位孿⽣體。這就是它將會有多 ⼤。除了基礎設施，我們與 AWS ⼀起做了很多事情。我 們正在加速 CUDA 加速 SageMaker ⼈⼯智慧。

We're CUDA accelerating Bedrock AI. Amazon Robotics is working with us, using NVIDIA Omniverse and Isaac SIM. AWS Health has NVIDIA Health integrated into it. So AWS has has really leaned into accelerated computing. Google is gearing up for Blackwell. GCP already has a 100, h 100, t4s, l4s, a whole fleet of NVIDIA CUDA GPUs. And they recently announced the JEMMA model that runs across all of it. We're working to optimize and accelerate every aspect of GCP. We're accelerating Dataproc, which is for data processing, their data processing engine, JAX, XLA, Vertex AI, and MuJoCo for robotics. So we're working with, Google and GCP across a whole bunch of initiatives.

我們正在使⽤CUDA加速Bedrock AI。亞⾺遜機器⼈正在 與我們合作，使⽤NVIDIA Omniverse和Isaac SIM。 AWS Health已整合了NVIDIA Health。因此，AWS確實 ⼤⼒推動加速運算。Google正在為Blackwell做準備。 GCP已經擁有100、h 100、t4s、l4s等⼀整套NVIDIA CUDA GPU。他們最近宣布了可以在所有這些GPU上運 ⾏的JEMMA模型。我們正在努⼒優化和加速GCP的每個 ⽅⾯。我們正在加速Dataproc，這是⽤於數據處理的數 據處理引擎，JAX、XLA、Vertex AI和MuJoCo⽤於機器 ⼈技術。因此，我們正在與Google和GCP合作推動各種 倡議。

Oracle is gearing up for Blackwell. Oracle is a great partner of ours for NVIDIA DGX Cloud and we're also working together to accelerate something that's really important to a lot of companies, Oracle database. Microsoft is accelerating, and Microsoft is gearing up for Blackwell. Microsoft NVIDIA has a wide ranging partnership. We're accelerating, CUDA accelerating all kinds of services when you when you chat obviously, and AI services that are in Microsoft Azure. It's very, very likely NVIDIA is in the back doing the inference and the token generation. We built they built the largest NVIDIA InfiniBand supercomputer, basically a digital twin of ours or a physical twin of ours. We're bringing the NVIDIA ecosystem to Azure. NVIDIA DGX Cloud to Azure. NVIDIA Omniverse is now hosted in Azure.

Oracle 正在為 Blackwell 做準備。Oracle 是我們在 NVIDIA DGX Cloud ⽅⾯的᯿要合作夥伴，我們也正在共 同努⼒加速對許多公司非常᯿要的事情，那就是 Oracle 資料庫。Microsoft 正在加速，Microsoft 也正在為 Blackwell 做準備。Microsoft 與 NVIDIA 有廣泛的合作夥 伴關係。當你使⽤ Microsoft Azure 時，我們正在加速， CUDA 加速各種服務，當你聊天時，當然還有 AI 服務。 很有可能 NVIDIA 在背後進⾏推論和代幣⽣成。他們建造 了最⼤的 NVIDIA InfiniBand 超級電腦，基本上是我們的 數位孿⽣或實體孿⽣。我們正在將 NVIDIA ⽣態系統帶到 Azure 上。NVIDIA DGX Cloud 來到 Azure。NVIDIA Omniverse 現在也在 Azure 上運⾏。

NVIDIA Healthcare is in Azure. And all of it is deeply integrated and deeply connected with Microsoft Fabric. The whole industry is gearing up for Blackwell. This is what I'm about to show you. Most of the most of the the the, scenes that you've seen so far of Blackwell are the are the full fidelity design of Blackwell. Everything in our company has a digital twin. And in fact, this digital twin idea is is really spreading, and it it helps it helps companies build very complicated things perfectly the first time. And what could be more exciting than creating a digital twin to build a computer that was built in a digital twin. And so let me show you what Wistron is doing. To meet the demand for NVIDIA accelerated computing, Wistron, one of our leading manufacturing partners, is building digital twins of NVIDIA DGX and HGX factories using custom software developed with Omniverse SDKs and APIs.

NVIDIA Healthcare 已經在 Azure 上運作。⽽且所有這些 都與 Microsoft Fabric 深度整合並深度連接。整個產業正 在為 Blackwell 做準備。這就是我即將要展⽰給你的內 容。迄今為⽌你所看到的⼤部分 Blackwell 场景都是 Blackwell 的完整設計。我們公司的每⼀樣東⻄都有⼀個 數位孿⽣體。事實上，這個數位孿⽣體的概念正在廣泛應 ⽤，它幫助公司第⼀次就完美地建造非常複雜的東⻄。有 什麼比創建⼀個數位孿⽣體來建造⼀個在數位孿⽣體中建 造的電腦更令⼈興奮呢？所以讓我來展⽰⼀下 Wistron 正 在做的事情。為了滿⾜對 NVIDIA 加速運算的需求，我們 的主要製造合作夥伴之⼀ Wistron 正在使⽤使⽤ Omniverse SDK 和 API 開發的⾃定義軟體，建立 NVIDIA DGX 和 HGX ⼯廠的數位孿⽣體。

For their newest factory, Wistron started with the digital twin to virtually integrate their multi CAD and process simulation data into a unified view. Testing and optimizing layouts in this physically accurate digital environment increased worker efficiency by 51%. During construction, the Omniverse digital twin was used to verify that the physical build matched the digital plans. Identifying any discrepancies early has helped avoid costly change orders, and the results have been impressive. Using a digital twin helped bring Wistron's factory online in half the time, just two and a half half months instead of 5. In operation, the Omniverse digital twin helps Wistron rapidly test new layouts to accommodate new processes or improve operations in the existing space, and every machine on the production line. Which ultimately enabled Wistron to reduce end to end cycle times by 50 percent and defect rates by 40%. With NVIDIA AI and Omniverse, NVIDIA's global ecosystem ecosystem of partners are building a new era of accelerated AI enabled digitalization. That's how we that's the way it's gonna be in the future. We're manufacturing everything digitally first, and then we'll manufacture it physically.

為了他們最新的⼯廠，Wistron 開始使⽤數位孿⽣技術， 將他們的多種 CAD 和流程模擬數據虛擬整合成統⼀的視 圖。在這個物理準確的數位環境中測試和優化佈局，使⼯ ⼈效率提⾼了 51%。在建造過程中，Omniverse 數位孿 ⽣技術被⽤來驗證實際建造是否符合數位計畫。早期識別 任何差異有助於避免昂貴的變更訂單，⽽成果令⼈印象深 刻。使⽤數位孿⽣技術幫助 Wistron ⼯廠在半數時間內上 線，僅需兩個半⽉，⽽非 5 個⽉。在運作中，Omniverse 數位孿⽣技術幫助 Wistron 快速測試新佈局以容納新流程 或改善現有空間的運作，以及⽣產線上的每台機器。最終 使 Wistron 能夠將端對端週期時間減少 50%，缺陷率降 低 40%。憑藉 NVIDIA AI 和 Omniverse，NVIDIA 的全球 ⽣態系統合作夥伴正在打造⼀個新時代的加速 AI 啟⽤數 位化。這就是未來的模式。我們⾸先以數位⽅式製造⼀ 切，然後再進⾏實體製造。

People ask me, how did it start? What got you guys so excited? What was it that you saw that caused you to put it all in on this incredible idea? And it's this. Hang on a second. Guys, that was going to be such a moment. That's what happens when you don't rehearse. This, as you know, was first contact. 2012, AlexNet. You put a cat into this computer, and it comes out and it says, cat.

⼈們問我，這是怎麼開始的？是什麼讓你們如此興奮？是 什麼讓你們看到後便全⼒投入這個驚⼈的想法？就是這 個。等⼀下。夥計們，那本來會是⼀個如此᯿要的時刻。 這就是當你沒有排練時會發⽣的事情。這個，你們知道， 是第⼀次接觸。2012年，AlexNet。你把⼀隻貓放進這台 電腦，它出來說，貓。

And we said, oh, my God. This is gonna change everything. You take 1,000,000 numbers you take 1,000,000 numbers across 3 channels, RGB. These numbers make no sense to anybody. You put it into this software and it compress it dimensionally reduces it. It reduces it from a 1000000 dimensions, a 1000000 dimensions. It turns it into 3 letters. One vector, one number. And it's generalized. You could have the cat be different cats, and and you could have it be the front of the cat and the back of the cat.

我們說，天啊。這將改變⼀切。你拿1,000,000個數字， 你拿1,000,000個數字跨越3個通道，RGB。這些數字對 任何⼈來說都毫無意義。你把它放入這個軟體，它會對其 進⾏壓縮、降維。它將1000000個維度降低到1000000個 維度。它將其轉換為3個字⺟。⼀個向ᰁ，⼀個數字。⽽ 且它是泛化的。你可以讓這隻貓變成不同的貓，你可以讓 它成為貓的正⾯和背⾯。

And you look at this thing, you say, unbelievable. You mean any cats? Yeah. Any cat. And it was able to recognize all these cats. And we realized how it did it. Systematically, structurally, it's scalable. How big can you make it? Well, how big do you want to make it? And so we imagine that this is a completely new way of writing software.

當你看著這個東⻄時，你會說，難以置信。你是說任何貓 都可以嗎？對，任何貓。它能夠辨識所有這些貓。我們意 識到它是如何做到的。系統性地、結構性地，它是可擴展 的。你可以做得有多⼤？嗯，你想要做多⼤？因此，我們 想像這是⼀種全新的撰寫軟體的⽅式。

And now, today, as you know, you can have you type in the word c a t, and what comes out is a cat. It went the other way. Am I right? Unbelievable. How is it possible that's right. How is it possible you took 3 letters, and you generated a 1000000 pixels from it, and it made sense? Well, that's the miracle. And here we are, just literally 10 years later 10 years later, where we recognize text, we recognize images, we recognize videos and sounds and images. Not only do we recognize them, we understand their meaning. We understand the meaning of the text.

現在，今天，如你所知，你可以輸入單詞 c a t，然後顯 ⽰出⼀隻貓。這是反過來的。我說得對嗎？難以置信。怎 麼可能是對的。怎麼可能你只⽤了三個字⺟，就能產⽣⼀ 百萬個像素，⽽且還有意義？這就是奇蹟。⽽現在，就在 ⼗年後，我們能辨識⽂字、圖像、影片和聲⾳。我們不僅 辨識它們，還理解它們的意義。我們理解⽂字的意義。

That's the reason why I can chat with you. It can summarize for you. It understands the text. It understood not just recognizes the the English, it understood the English. It doesn't just recognize the pixels, it understood the pixels. And you can you can even condition it between 2 modalities. You can have language condition image and generate all kinds of interesting things. Well, if you can understand these things, what else can you understand that you've digitized? The reason why we started with text and, you know, images is because we digitize those. But what else have we digitized?

這就是為什麼我能跟你聊天。它可以為你做摘要。它理解 ⽂字。它不僅僅是認識英⽂，它理解英⽂。它不僅僅是認 識像素，它理解像素。⽽且你甚⾄可以在兩種模式之間進 ⾏條件設定。你可以讓語⾔條件影像，並⽣成各種有趣的 東⻄。嗯，如果你能理解這些事情，那麼你還能理解你已 經數位化的其他事物嗎？我們之所以從⽂字和圖像開始， 是因為我們已經將它們數位化了。但我們還數位化了什麼 呢？

Well, it turns out we digitized a lot of things, proteins and genes and brainwaves. Anything you can digitize, so long as there's structure, we can probably learn some patterns from it. And if we can learn the patterns from it, we can understand its meaning. If we can understand its meaning, we might be able to generate it as well. And so therefore, the generative AI revolution is here. Well, what else can we generate? What else can we learn? Well, one one of the things that we would love to learn we would love to learn is we would love to learn climate. We would love to learn extreme weather. We would love to learn, what how we can predict future weather at regional scales at sufficiently high resolution, such that we can keep people out of harm's way before harm comes.

原來我們數位化了許多東⻄，蛋⽩質、基因和腦波。只要 有結構，任何你可以數位化的東⻄，我們可能都能從中學 習⼀些模式。如果我們能從中學習模式，我們就能理解其 意義。如果我們能理解其意義，我們也許就能⽣成它。因 此，⽣成式⼈⼯智慧⾰命來臨了。那麼，我們還能⽣成什 麼？我們還能學到什麼？其中⼀件我們很想學習的事情是 氣候。我們很想學習極端天氣。我們很想學習如何在區域 尺度上以⾜夠⾼的解析度預測未來天氣，以便在災害來臨 之前讓⼈們遠離危險。

Extreme weather cost the world a $150,000,000,000 Surely more than that and it's not evenly distributed. $150,000,000,000 is concentrated in some parts of the world and of course to some people of the world. We need to adapt and we need to know what's coming. And so we are creating earth 2, a digital twin of the earth for predicting weather. And we've made an extraordinary invention called core dev, the ability to use generative AI to predict weather at extremely high resolution. Let's take a look. As the earth's climate changes, AI powered weather forecasting is allowing us to is allowing us to more accurately predict and track severe storms, like super typhoon Chantu, which caused widespread damage in Taiwan and the which can miss important details. NVIDIA's cordif is a revolutionary new generative AI model trained on radar assimilated wharf weather forecasts and error 5 reanalysis data. Using Chordiff, extreme events like canthu can be super resolved from 25 kilometer to 2 kilometer resolution, with 1,000 times the speed and 3,000 times the energy efficiency of conventional weather models. By combining the speed and accuracy of NVIDIA's weather forecasting model, forecast net, and generative AI models like cordif, we can explore 100 or even 1000 of kilometer scale regional weather forecasts to provide a clear picture of the best, worst, and most likely impacts of a storm.

極端天氣給世界帶來了⼀千五百億美元的損失，這個數字 可能還要更⾼，⽽且並非均勻分佈。⼀千五百億美元集中 在世界某些地區，當然也影響了世界某些⼈。我們需要適 應，也需要知道即將發⽣的事情。因此，我們正在創建 「地球2號」，這是地球的數位孿⽣體，⽤於預測天氣。 我們開發了⼀項非凡的發明，稱為「核⼼開發」，能夠利 ⽤⽣成式⼈⼯智慧以極⾼的解析度預測天氣。讓我們來看 看。隨著地球氣候的變化，由⼈⼯智慧驅動的天氣預測使 我們能夠更準確地預測和追蹤嚴᯿風暴，如造成台灣嚴᯿ 損害的超級颱風「尚都」，⽽這些風暴可能會忽略⼀些᯿ 要細節。NVIDIA的「核⼼開發」是⼀種⾰命性的新⽣成 式⼈⼯智慧模型，經過雷達同化的wharf天氣預報和誤差5 ᯿新分析數據進⾏訓練。使⽤「核⼼開發」，像「尚都」 這樣的極端事件可以從25公⾥的解析度提升到2公⾥，速 度比傳統天氣模型快1000倍，能源效率提⾼3000倍。通 過結合NVIDIA的天氣預測模型「預測網」的速度和準確 性，以及像「核⼼開發」這樣的⽣成式⼈⼯智慧模型，我 們可以探索100甚⾄1000公⾥尺度的區域天氣預報，提供 風暴的最佳、最糟和最可能影響的清晰圖像。

This wealth of information can help minimize loss of life and property damage. The weather company has to trust the source of global weather prediction. We are working together to accelerate their weather simulation. First principled base of simulation. However, they're also going to integrate Earth 2 cordif so that they could help businesses and countries do regional high resolution weather prediction. And so if you have some weather prediction you'd like to know, like to do, reach out to The Weather Company. Really exciting really exciting work. NVIDIA Healthcare, something we started 15 years ago. We're super super excited about this. This is an area where we're very very proud.

這豐富的資訊可以幫助減少⼈命損失和財產損害。氣象公 司必須信任全球天氣預測的來源。我們正在共同努⼒加速 他們的天氣模擬。⾸先是模擬的原則基礎。然⽽，他們也 將整合Earth 2 cordif，以幫助企業和國家進⾏區域⾼解析 度的天氣預測。所以，如果你有任何想知道、想做的天氣 預測，請聯繫氣象公司。這是非常令⼈興奮的⼯作。 NVIDIA Healthcare，這是我們15年前開始的項⽬。我們 對此非常興奮。這是⼀個讓我們感到非常⾃豪的領域。

Whether it's medical imaging or it's gene sequencing or computational chemistry, it is very likely that NVIDIA is the computation behind it. We've done so much work in this area. Today, we're announcing that we're gonna do something really, really cool. Imagine all of these AI models that are being used to generate images and audio. But instead of images and audio, because it understood images and audio, all the digitization that we've done for genes and proteins and amino acids, that digitization capability is now passed through machine learning so that we understand the language of life. The ability to understand the language of life, of course, we saw the first evidence of it with AlphaFold. This is really quite an extraordinary thing. After decades of painstaking work, the world had only digitized and reconstructed using cryo electron microscopy or crystal x-ray x-ray, crystallography, these different techniques painstakingly reconstructed the protein, 200,000 of them, in just, what is it, less than a year or so? AlphaFold has reconstructed 200,000,000 proteins. Basically, every protein, every of every living thing that's ever been sequenced.

無論是醫學影像、基因定序還是計算化學，很有可能 NVIDIA 就是背後的運算⼒ᰁ。我們在這個領域做了很多 ⼯作。今天，我們宣布我們將要做⼀些非常酷的事情。想 像⼀下，所有這些被⽤來⽣成影像和⾳訊的⼈⼯智慧模 型。但不是影像和⾳訊，因為它了解影像和⾳訊，我們為 基因、蛋⽩質和氨基酸所做的所有數位化，這種數位化能 ⼒現在通過機器學習傳遞，讓我們理解⽣命的語⾔。理解 ⽣命的語⾔的能⼒，當然，我們在 AlphaFold 中看到了第 ⼀個證據。這真的是⼀件非常了不起的事情。經過數⼗年 的辛苦⼯作，世界只是使⽤冷凍電⼦顯微鏡或晶體 X 光 線衍射，這些不同的技術費⼒地᯿建了蛋⽩質，其中有 20 萬個，在不到⼀年的時間內？AlphaFold 已經᯿建了 2 億個蛋⽩質。基本上，每⼀種⽣物的每⼀種蛋⽩質，每⼀ 種曾經被定序的⽣物。

This is completely revolutionary. Well, those models are incredibly hard to use, for incredibly hard for people to build. And so what we're gonna do is we're gonna build them. We're gonna build them for, the the researchers around the world. And it won't be the only one, there'll be many other models that we create. And so let me show you what we're gonna do with it. Virtual screening for new medicines is a computationally intractable problem. Existing techniques can only scan billions of compounds and require days on thousands of standard compute nodes to identify new drug candidates. NVIDIA BioNeMo NIMS enable a new generative screening paradigm Using NIMS for protein structure prediction with AlphaFold molecule generation with MolMim and docking with DiffDock we can now generate and screen candidate molecules in a matter of minutes MolMim can connect to custom applications to steer the generative process, iteratively optimizing for desired properties. These applications can be defined with BioNeMo microservices or built from scratch.

這是完全⾰命性的。嗯，這些模型非常難以使⽤，對於⼈ 們來說建立它們也非常困難。所以我們要做的是建立它 們。我們要為全球的研究⼈員建立這些模型。⽽且這不會 是唯⼀的，我們還會創建許多其他模型。現在讓我來展⽰ ⼀下我們將如何應⽤它。對於新藥物的虛擬篩選是⼀個計 算上難以解決的問題。現有的技術只能掃描數⼗億個化合 物，需要數千個標準計算節點的幾天時間才能識別出新的 藥物候選物。NVIDIA BioNeMo NIMS實現了⼀種新的⽣ 成篩選範式。使⽤NIMS進⾏蛋⽩質結構預測，並使⽤ AlphaFold分⼦⽣成，再進⾏MolMim對接，現在我們可以 在幾分鐘內⽣成和篩選候選分⼦。MolMim可以連接到⾃ 定義應⽤程序，引導⽣成過程，迭代優化所需的特性。這 些應⽤程序可以使⽤BioNeMo微服務定義，或者從頭開 始構建。

Here, a physics based simulation optimizes for a molecule's ability to bind to a target protein, while optimizing for other favorable molecular properties in parallel. MolMim generates high quality drug like molecules that bind to the target and are synthesizable, translating to a higher probability of developing successful medicines faster. Bionemo is enabling a new paradigm in drug discovery with NIMS, providing on demand microservices that can be combined to build powerful drug discovery workflows, like de novo protein design or guided molecule generation for virtual screening. BioNeMo NIMS are helping researchers and developers reinvent computational drug design. NVIDIA MOMAM, Core Diff, there's a whole bunch of other models, a whole bunch of other models: computer vision models, robotics models, and even of course, some really, really terrific open source language models. These models are groundbreaking. However, it's hard for companies to use. How would you use it? How would you bring it into your company and integrate it into your workflow? How would you package it up and run it?

這裡，基於物理學的模擬優化了分⼦與⽬標蛋⽩質結合的 能⼒，同時平⾏優化其他有利的分⼦特性。MolMim⽣成 ⾼品質的類藥物分⼦，能夠與⽬標結合並且可合成，進⽽ 提⾼開發成功藥物的機率。Bionemo透過NIMS在藥物發 現領域開創了新的範式，提供按需的微服務，可以組合成 強⼤的藥物發現⼯作流程，如全新的蛋⽩質設計或引導分 ⼦⽣成⽤於虛擬篩選。BioNeMo NIMS幫助研究⼈員和開 發者᯿新設計計算藥物設計。NVIDIA MOMAM、Core Diff，還有許多其他模型：電腦視覺模型、機器⼈模型， 當然還有⼀些非常出⾊的開源語⾔模型。這些模型是開創 性的。然⽽，企業很難使⽤。你將如何使⽤它？如何將其 引入公司並整合到⼯作流程中？如何打包並運⾏它？

Remember, earlier I just said that inference is an extraordinary computation problem. How would you do the optimization for each and every one of these models and put together the computing stack necessary to run that supercomputer so that you can run these models in your company? And so we have a great idea. We're gonna invent a new way for you to receive and operate software. This software comes basically in a digital box. We call it a container, And we call it the NVIDIA Inference Microservice, a NIM. And I'll explain to you what it is. A NIM, it's a pre trained model, so it's pretty clever. And it is packaged and optimized to run across NVIDIA's installed base, which is very, very large. What's inside it is incredible.

記得，我之前說過推論是⼀個非常了不起的計算問題。你 要如何為每⼀個模型進⾏優化，並組建必要的計算堆疊來 運⾏那台超級電腦，以便在你的公司運⾏這些模型呢？所 以我們有⼀個很棒的主意。我們將發明⼀種新的⽅式讓你 接收和操作軟體。這個軟體基本上是放在⼀個數位盒⼦ 裡。我們稱之為容器，⽽我們稱之為 NVIDIA 推論微服 務，簡稱 NIM。我來解釋⼀下它是什麼。⼀個 NIM，它 是⼀個預先訓練好的模型，所以它非常聰明。它被打包並 優化以在 NVIDIA 的龐⼤安裝基礎上運⾏，這個基礎非常 非常龐⼤。裡⾯的內容令⼈難以置信。

You have all these pre trained, state of the art, Open Source models. They could be Open Source, they could be from one of our partners, it could be created by us, like NVIDIA Moment. It is packaged up with all of its dependencies. So CUDA, the right version. CU DNN, the right version. TensorRT, LMM, distributed across the multiple GPUs. Triton Inference Server, all completely packaged together. It's optimized depending on whether you have a single GPU, multi GPU, or multi node GPUs, it's optimized for that, and it's connected up with APIs that are simple to use. Now, think about what an AI API is. An AI API is an interface that you just talk to.

你擁有所有這些預先訓練、最先進的開源模型。它們可以 是開源的，可以來⾃我們的合作夥伴，也可以是我們⾃⼰ 創建的，就像 NVIDIA Moment ⼀樣。這些模型已經封裝 好了所有相關的依賴項。所以 CUDA、正確的版本。CU DNN、正確的版本。TensorRT、LMM，分佈在多個 GPU 上。Triton 推理伺服器，全部完整地封裝在⼀起。 它根據您使⽤的是單個 GPU、多個 GPU 還是多節點 GPU 進⾏了優化，並且與易於使⽤的 API 連接在⼀起。 現在，想想 AI API 是什麼。AI API 就是⼀個你可以直接 對話的介⾯。

And so this is a piece of software in the future that has a really simple API, and that API is called human. And these packages, incredible bodies of software, will be optimized and packaged, and we'll put it on a website. And you can download it, you could take it with you, you could run it in any cloud, you could run it in your own data center, you could run it in workstations if it fit, and all you have to do is come to ai.nvidia.com. We call it NVIDIA Inference Microservice, but inside the company, we all call it NIMS. K? Just imagine, you know, one of some someday, there's there's gonna be one of these chat bots, and these chat bots is gonna just be an NIM. And you'll you'll, you'll assemble a whole bunch of chatbots. And that's the way software is gonna be be built someday. How do we build software in the future? It is unlikely that you'll write it from scratch or write a whole bunch of Python code or anything like that.

這是未來的⼀個軟體，具有非常簡單的API，⽽這個API 被稱為「⼈類」。這些軟體包，是令⼈難以置信的軟體體 系，將被優化並打包，然後我們會將它放在⼀個網站上。 你可以下載它，你可以攜帶它，你可以在任何雲端運⾏ 它，你可以在你⾃⼰的資料中⼼運⾏它，如果適合的話， 你可以在⼯作站上運⾏它，⽽你所需要做的就是前往 ai.nvidia.com。我們稱之為 NVIDIA 推理微服務，但在公 司內部，我們都稱之為 NIMS。你知道嗎？想像⼀下，總 有⼀天，會有⼀個這樣的聊天機器⼈，這些聊天機器⼈將 只是⼀個 NIM。你將組裝⼀堆聊天機器⼈。這就是未來 軟體將被建立的⽅式。未來我們如何建立軟體？不太可能 是從頭開始寫或寫⼀堆 Python 程式碼或任何其他⽅式。

It is very likely that you assemble a team of AIs. There's probably going to be a super AI that you use that takes the mission that you give it and breaks it down into an execution plan. Some of that execution plan could be handed off to another NIM. That NIM would maybe, understand SAP. The language of SAP is ABAP. It might understand ServiceNow, and it go retrieve some information from their platforms. It might then hand that result to another NIM who that goes off and does some calculation on it. Maybe it's an optimization software, a combinatorial optimization algorithm. Maybe it's, you know, some just some basic calculator. Maybe it's pandas to do some numerical analysis on it.

很有可能你會組建⼀⽀⼈⼯智慧團隊。很可能會有⼀個超 級⼈⼯智慧，負責接收你給予的任務並將其拆解成執⾏計 畫。這些執⾏計畫的⼀部分可能會交給另⼀個NIM。那個 NIM可能會懂得SAP。SAP的語⾔是ABAP。它可能會懂 得ServiceNow，並從他們的平台中檢索⼀些資訊。然後 它可能會將結果交給另⼀個NIM，那個NIM會進⾏⼀些計 算。也許是⼀個優化軟體，⼀個組合優化演算法。也許是 ⼀個基本的計算機。也許是pandas⽤來進⾏⼀些數值分 析。

And then it comes back with its answer, and it gets combined with everybody else's, and it because it's been presented with, this is what the right answer should look like, it knows what answer what what right answers to produce, and it presents it to you. We can get a report every single day at, you know, top of the hour, that has something to do with a build plan or some forecast or some customer alert or some bugs database or whatever it happens to be. And we could assemble it using all these NIMs. And because these NIMs have been packaged up and ready to work on your systems, so long as you have video GPUs in your data center in the Cloud, these NIMs will work together as a team and do amazing things. And so we decided, this is such a great idea. We're gonna go do that. And so NVIDIA has NIMS running all over the company. We have chatbots being created all over the place. And one of our most important chatbots, of course, is a chip designer chatbot. You might not be surprised.

然後它回來了，帶著它的答案，並且與其他⼈的答案結合 在⼀起，因為它已經被呈現了，這就是正確答案應該是什 麼樣⼦，它知道應該產⽣什麼樣的正確答案，並將其呈現 給你。我們可以每天在整點的時候收到⼀份報告，裡⾯可 能涉及建置計劃、某些預測、客⼾警報、錯誤數據庫或其 他任何事情。我們可以使⽤所有這些 NIMs 來組裝它。由 於這些 NIMs 已經打包好並準備在您的系統上運⾏，只要 您在您的數據中⼼或雲端擁有視頻 GPU，這些 NIMs 將 作為⼀個團隊⼀起⼯作並做出驚⼈的事情。因此，我們決 定，這是⼀個很棒的主意。我們要去做這件事。因此， NVIDIA 公司各處都在運⾏ NIMS。我們在各處都在創建 聊天機器⼈。當然，其中⼀個最᯿要的聊天機器⼈是芯片 設計師聊天機器⼈。或許你不會感到驚訝。

We care a lot about building chips. And so we want to build chatbots, AI co pilots that are co designers with our engineers. And so, this is the way we did it. So we got ourselves a llama 2. This is a 70 b, and it's, you know, packaged up in a NIM. And we asked it, you know, what is a CTL? Well, it turns out CTL is an internal, program and it has an internal proprietary language, But it thought the CTL was a combinatorial timing logic, and so it describes, you know, conventional knowledge of CTL. But that's not very useful to us. And so we gave it a whole bunch of new examples. You know, this is no different than employee onboarding an employee.

我們非常重視晶片的建造。因此，我們希望建造聊天機器 ⼈，⼈⼯智慧共同⾶⾏員，與我們的⼯程師共同設計。這 就是我們的做法。所以，我們得到了llama 2。這 是⼀個70 b，它被封裝在⼀個NIM中。我們問它，你知道 CTL是什麼嗎？嗯，結果發現CTL是⼀個內部程式，它有 ⼀個內部專有語⾔。但它認為CTL是⼀種組合時序邏輯， 因此描述了CTL的傳統知識。但這對我們來說並不是很有 ⽤。所以我們給了它⼀堆新的例⼦。這與員⼯入職沒有什 麼不同。

We say, you know, thanks for that answer. It's completely wrong. And and, and then we present to them, this is what a CTL is. Okay? And so, this is what a CTL is at NVIDIA. And the CTL, as you can see, you know, CTL stands for Compute Trace Library, which makes sense. You know, we're tracing compute cycles all the time, and it wrote the program. Isn't that amazing? And so the productivity of our chip designers can go up. This is what you can do with a NIM.

我們說，你知道，感謝你的回答。完全錯了。然後我們向 他們展⽰，這就是CTL是什麼。好嗎？所以，這就是 NVIDIA的CTL。⽽且，正如你所看到的，你知道，CTL 代表計算追蹤庫，這是有道理的。你知道，我們⼀直在追 蹤計算週期，並且編寫了程式。這不是很驚⼈嗎？這樣⼀ 來，我們的晶片設計師的⽣產⼒就可以提⾼。這就是你可 以⽤NIM做的事情。

First thing you can do is customize it. We have a service called NIMO microservice that helps you curate the data, preparing the data so that you could teach this onboard this AI. You fine tune them and then you guardrail it. You can even evaluate the answer, evaluate its performance against, other other examples. And so that's called the neemo microservice. Now, the thing that's that's emerging here is this, there are 3 elements, 3 pillars of what we're doing. The first pillar is, of course, inventing the technology for, AI models and running AI models and packaging it up for you. The second is to create tools to help you modify it. 1st is having the AI technology. 2nd is to help you modify it.

⾸先，您可以進⾏⾃定義。我們有⼀項名為 NIMO 微服 務的服務，可以幫助您篩選數據，準備數據，以便您可以 將其教導給這個 AI。您可以微調它們，然後對其進⾏監 控。您甚⾄可以評估答案，評估其表現與其他範例對比。 這就是所謂的 NIMO 微服務。現在，這裡出現的新趨勢 是，我們正在做的有三個元素，三個⽀柱。第⼀⽀柱當然 是為 AI 模型發明技術，運⾏ AI 模型並為您打包。第⼆個 是創建⼯具來幫助您修改它。第⼀是擁有 AI 技術。第⼆ 是幫助您修改它。

And 3rd is infrastructure for you to fine tune it and, if you like, deploy it. You could deploy it on our infrastructure called DGX Cloud or you can employ deploy it on prem, you could deploy it anywhere you like. Once you develop it, it's yours to take anywhere. And so we are effectively an AI foundry. We will do for you and the industry on AI what TSMC does for us building chips. And so we go to it with our go to TSMC with our big ideas. They manufacture it and we take it with us. And so, exactly the same thing here AI Foundry and the 3 pillars are the NIMs, Nexmo Microservice and DGX Cloud. The other thing that you could teach the NIM to do is to understand your proprietary information. Remember, inside our company, the vast majority of our data is not in the cloud.

第三點是為您提供基礎設施，讓您可以進⾏微調，並且如 果您願意，部署它。您可以將其部署在我們稱為 DGX Cloud 的基礎設施上，或者您可以部署在本地，您可以將 其部署在您喜歡的任何地⽅。⼀旦您開發完成，它就是您 的，您可以帶到任何地⽅。因此，我們有效地是⼀個⼈⼯ 智慧⼯廠。我們將為您和整個⼈⼯智慧產業做像台積電為 我們製造晶片⼀樣的事情。因此，我們帶著我們的⼤點⼦ 去找台積電。他們製造它，我們帶著它走。因此，在這裡 也是⼀樣的，⼈⼯智慧⼯廠和三⼤⽀柱是 NIMs、Nexmo 微服務和 DGX Cloud。您可以教導 NIM 去理解您的專有 資訊。請記住，在我們公司內部，絕⼤多數的資料並不在 雲端中。

It's inside our company. It's been sitting there, you know, being used all the time. And and gosh, it's it's basically NVIDIA's intelligence. We would like to take that data, learn its meaning like we learned the meaning of almost anything else that we just talked about. Learn its meaning, and then re index that knowledge into a new type of database called a vector database. So you essentially take structured data or unstructured data, you learn its meaning, you encode its meaning, so now this becomes an AI database, and that AI database, in the future, once you create it, you can talk to it. And So let me give you an example of what you could do. So suppose you create you've got a whole bunch of multi modality data, and one good example of that is PDF. So you take the PDF, you take all of your PDFs, all your favorite you know, the stuff that is proprietary to you, critical to your company. You can encode it just as we encode the pixels of a cat, and it becomes the word cat.

這是在我們公司內部。它⼀直在那裡，你知道，⼀直被使 ⽤。⽽且，天啊，這基本上就是 NVIDIA 的智慧。我們想 要拿取這些資料，學習它的意義，就像我們學習了我們剛 剛談到的幾乎任何其他事情的意義⼀樣。學習它的意義， 然後᯿新將這些知識索引到⼀種新型的資料庫，稱為向ᰁ 資料庫。因此，您基本上拿取結構化資料或非結構化資 料，學習其意義，編碼其意義，這樣現在這就變成了⼀個 ⼈⼯智慧資料庫，⽽這個⼈⼯智慧資料庫，在未來，⼀旦 您創建它，您就可以與它對話。所以讓我給您⼀個例⼦， 您可以做什麼。假設您創建了⼀⼤堆多模式資料，其中⼀ 個很好的例⼦就是 PDF。所以您拿取這個 PDF，拿取所 有您的 PDF，所有您喜歡的，您知道，對您的公司⾄關 ᯿要的東⻄。您可以將其編碼，就像我們編碼了⼀隻貓的 像素⼀樣，這樣它就變成了單詞「貓」。

We can encode all of your PDF, and it turns into vectors that are now stored inside your vector database. It becomes the proprietary information of your company. And once you have that proprietary information, you could chat to it. It's an it's a smart database, and so you just chat chat with data. And how how much more enjoyable is that? You know, for for our software team, you know, they just chat with the bugs database. You know? How many bugs was there last night? Are we making any progress? And then, after you're done talking to this, bug's database, you need therapy.

我們可以將您所有的PDF進⾏編碼，然後將其轉換為向 ᰁ，現在這些向ᰁ存儲在您的向ᰁ資料庫中。這將成為貴 公司的專有資訊。⼀旦您擁有了這些專有資訊，您就可以 與之交談。這是⼀個智能資料庫，所以您只需與資料交 談。這樣多有趣啊！對於我們的軟體團隊來說，他們只需 與錯誤資料庫交談。您知道嗎？昨晚有多少錯誤？我們有 進展嗎？然後，在您與這個錯誤資料庫交談結束後，您需 要治療。

And so so we we have another chatbot for you. You can do it. Okay. So we call this Nemo retriever and the reason for that is because ultimately, its job is to go retrieve information as quickly as possible. You just talk to it. Hey, retrieve me this information. It brings it back to you. Do you mean this? You go, Yeah, perfect. Okay.

所以我們為⼤家帶來另⼀個聊天機器⼈。你可以使⽤它。 好的。所以我們稱之為 Nemo 檢索器，原因是因為它的 ⼯作最終⽬的是盡快檢索信息。你只需跟它說話。嘿，幫 我找這個資訊。它會把資訊帶回給你。你是指這個嗎？你 回答，對，完美。好的。

And so, we call it the Nemo retriever. Well, the NEMO service helps you create all these things, and we have all these different NEMs. We even have NEMs of digital humans. I'm Rachel, your AI care manager. Okay. So so we have it's a really short clip, but there were so many videos to show you. I guess, so many other demos to show you and so I had to cut this one short. But this is Diana. She is a digital human NIM. And and, you just talked to her, and she's connected, in this case, to Hippocratic AI's large language model for health care.

因此，我們稱之為尼莫檢索器。嗯，NEMO服務可以幫助 您創建所有這些東⻄，我們擁有各種不同的NEM。我們 甚⾄擁有數位⼈類的NEM。我是Rachel，您的AI照護經 理。好的。這只是⼀個非常簡短的片段，但有很多影片要 向您展⽰。我想，還有很多其他的演⽰要給您看，所以我 必須縮短這個片段。但這是黛安娜。她是⼀個數位⼈類 NIM。您可以與她交談，她在這個案例中與希波克拉底AI 的⼤型醫療語⾔模型相連。

And it's truly amazing. She is just super smart about health care things. You know? And so after you're done after my my Dwight, my VP of software engineering talks to the chatbot for Bugs database, then you come over and talk to Diane. And and so so, Diane is is, completely animated with AI, and she's a digital human. There's so many companies that would like to build. They're sitting on gold mines. The the enterprise IT industry is sitting on a gold mine. It's a gold mine because they have so much understanding of of, the way work is done. They have all these amazing tools that have been created over the years, and they're sitting on a lot of data.

這真是令⼈驚嘆。她對於醫療保健⽅⾯非常精明。你知道 嗎？所以在我們完成了軟體⼯程副總裁Dwight與聊天機 器⼈討論錯誤資料庫後，你就可以去找Diane談話。 Diane完全是透過⼈⼯智慧來呈現，她是⼀位數位⼈類。 有很多公司都希望能夠打造這樣的產品。他們坐擁⾦礦。 企業資訊科技產業就像坐擁⾦礦⼀樣。這是⼀座⾦礦，因 為他們對⼯作⽅式有著深刻的了解。多年來，他們開發了 許多令⼈驚嘆的⼯具，並且擁有⼤ᰁ的數據。

If they could take that gold mine and turn them into copilots, these copilots could help us do things. And so just about every IT franchise, IT platform in the world that has valuable tools that people use is sitting on a gold mine for co pilots. And they would like to build their own co pilots and their own chat bots. And so, we're announcing that NVIDIA AI Foundry is working with some of the world's great companies. SAP generates 87% of the world's global commerce. Basically, the world runs on SAP. We run on SAP. NVIDIA and SAP are building SAP jewel copilots using NVIDIANemo and DGX Cloud. ServiceNow, they run 80 85 percent of the world's Fortune 500 companies run their people and customer service operations on ServiceNow. And they're using NVIDIA AI Foundry to build ServiceNow, assist virtual assistants.

如果他們能夠將那座⾦礦變成副駕駛，這些副駕駛可以幫 助我們做事情。因此，幾乎世界上每⼀個具有⼈們使⽤寶 貴⼯具的IT特許經營權、IT平台都坐擁⼀座副駕駛的⾦ 礦。他們希望建立⾃⼰的副駕駛和聊天機器⼈。因此，我 們宣布 NVIDIA AI Foundry 正與⼀些世界頂尖公司合作。 SAP 產⽣了全球 87% 的商業交易。基本上，世界運⾏在 SAP 上。我們也在 SAP 上運⾏。NVIDIA 和 SAP 正在使 ⽤ NVIDIANemo 和 DGX Cloud 建立 SAP 寶⽯副駕駛。 ServiceNow，他們運⾏著全球 80 到 85% 的財富 500 強 公司的⼈員和客⼾服務操作。他們正在使⽤ NVIDIA AI Foundry 建立 ServiceNow 輔助虛擬助⼿。

Cohesity backs up the world's data. They're sitting on a gold mine of data. Hundreds of exabytes of data over 10,000 companies. NVIDIA AI Foundry is working with them, helping them build their Gaia generative AI agent. Snowflake is a company that stores the world's, digital warehouse in the cloud and serves over 3,000,000,000 queries a day for 10,000 enterprise customers. Snowflake is working with NVIDIA AI Foundry to build copilots with NVIDIA Nemo and NIMs. NetApp. Nearly half of the files in the world are stored on prem on NetApp. NVIDIA AI Foundry is helping them, build chat bots and copilots, like those vector databases and retrievers with NVIDIA, NEEMO, and NIMS. And we have a great partnership with Dell.

Cohesity是全球數據的備份專家。他們擁有龐⼤的數據資 源，超過10,000家公司的數百艾字節數據。NVIDIA AI Foundry正與他們合作，協助打造他們的Gaia⽣成式AI代 理。Snowflake是⼀家在雲端存儲全球數據倉庫並為超過 10,000家企業客⼾每天處理超過30億查詢的公司。 Snowflake正與NVIDIA AI Foundry合作，利⽤NVIDIA Nemo和NIMs打造共同合作的copilots。NetApp是全球近 ⼀半的⽂件存儲在NetApp的本地。NVIDIA AI Foundry正 協助他們建立聊天機器⼈和copilots，就像那些使⽤ NVIDIA NEEMO和NIMS的向ᰁ數據庫和檢索器⼀樣。我 們與Dell有著良好的合作夥伴關係。

Everybody who everybody who is building these chatbots and generative AI, when you're ready to run it, you're gonna need an AI factory. And nobody is better at building end to end systems of very large scale for the enterprise than Dell. And so anybody, any company, every company will need to build AI factories. And it turns out that Michael is here. He's happy to take your order. Ladies and gentlemen, Michael Dell. Okay. Let's talk about the next wave of AI. Robotics. Physical AI.

每個正在建造這些聊天機器⼈和⽣成式⼈⼯智慧的⼈，當 你們準備運⾏時，你們將需要⼀個⼈⼯智慧⼯廠。在企業 界，沒有⼈比 Dell 更擅長建造非常⼤規模的端到端系 統。因此，任何⼈、任何公司，每家公司都將需要建立⼈ ⼯智慧⼯廠。原來 Michael 在這裡。他很樂意接受您的訂 單。女⼠們，先⽣們，Michael Dell。好的。讓我們來談 談下⼀波⼈⼯智慧。機器⼈技術。實體⼈⼯智慧。

So far, all of the AI that we've talked about is 1 computer. Data comes into 1 computer, lots of the world's, if you will, experience in digital text form. The AI imitates us by reading a lot of the language to predict the next words. It's imitating you by studying all of the patterns and all the other previous examples. Of course, it has to understand context and so on and so forth. But once it understands the context, it's essentially imitating you. We take all of the data, we put it into a system like DGX, we compress it into a large language model, trillions and trillions of parameters become billions and billions, trillions of tokens becomes billions of parameters, these billions of parameters becomes your AI. Well, in order for us to go to the next wave of AI where the AI understands the physical world, we're gonna need 3 computers. The first computer is still the same computer. It's that AI computer that now is gonna be watching video and maybe it's doing synthetic data generation and maybe there's a lot of human examples.

到⽬前為⽌，我們所談論的所有⼈⼯智慧都是⼀台電腦。 數據進入⼀台電腦，世界上許多，如果你願意這樣說，以 數字⽂本形式呈現的經驗。⼈⼯智慧通過閱讀⼤ᰁ語⾔來 模仿我們，以預測下⼀個詞語。它通過研究所有模式和所 有其他先前的例⼦來模仿你。當然，它必須理解上下⽂等 等。但⼀旦它理解了上下⽂，它基本上就是在模仿你。我 們將所有數據放入像 DGX 這樣的系統，將其壓縮成⼀個 ⼤型語⾔模型，數以兆計的參數變成數⼗億，數以兆計的 標記變成數⼗億參數，這些數⼗億參數就成為你的⼈⼯智 慧。嗯，在我們要進入⼈⼯智慧的下⼀波浪潮，讓⼈⼯智 慧理解物理世界，我們將需要三台電腦。第⼀台電腦仍然 是同⼀台電腦。那就是那台⼈⼯智慧電腦，現在將會觀看 視頻，也許它正在進⾏合成數據⽣成，也許有很多⼈類的例⼦。

Just as we have human examples in text form, we're gonna have human examples in articulation form. And the AIs will watch us, understand what is happening, and try to adapt it for themselves into the context. And because it can generalize with these foundation models, maybe these robots can also perform in the physical world, fairly generally. So I just described in very simple terms essentially what just happened in large language models except the chat gpt moment for robotics may be right around the corner. And so we've been building the end to end systems for robotics for some time. I'm super, super proud of the work. We have the AI system, dgx. We have the lower system which is called AGX for autonomous systems, the world's first robotics processor. When we first built this thing, people are, what are you guys building? It's a SoC.

就像我們在⽂字形式中有⼈類的例⼦⼀樣，我們也將會有 ⼝語形式的⼈類例⼦。⼈⼯智慧將觀察我們，理解正在發 ⽣的事情，並試著將其適應到⾃⼰的情境中。由於它可以 通過這些基礎模型進⾏泛化，也許這些機器⼈也可以在物 理世界中表現得相當普遍。所以我⽤非常簡單的⽅式描述 了⼤型語⾔模型中發⽣的事情，除了聊天gpt時刻的機器 ⼈可能就在不遠處。因此，我們已經為機器⼈建立了端到 端系統⼀段時間了。我對這項⼯作感到非常非常⾃豪。我 們擁有AI系統dgx。我們有⼀個名為AGX的低階系統，⽤ 於⾃主系統，這是世界上第⼀個機器⼈處理器。當我們第 ⼀次建造這個東⻄時，⼈們問，你們在建造什麼？這是⼀ 個SoC。

It's 1 chip. It's designed to be very low power, but it's designed for high speed sensor processing and AI. And so if you wanna run transformers in a car or you wanna run transformers in a in a, you know, anything, that moves, we have the perfect computer for you. It's called the Jetson. And so the DGX on top for training the AI, the Jetson is the autonomous processor, and in the middle, we need another computer. Whereas, large language models have the benefit of you providing your examples and then doing reinforcement learning human feedback. What is the reinforcement learning human feedback of a robot? Well, it's reinforcement learning physical feedback. That's how you align the robot. That's how you that's how the robot knows that as it's learning these articulation capabilities and manipulation capabilities, it's going to adapt properly into the laws of physics.

這是⼀個晶片。它被設計成非常低功耗，但卻是為⾼速感 應器處理和⼈⼯智慧⽽設計的。所以如果你想在汽⾞中運 ⾏變壓器，或者在任何移動的東⻄中運⾏變壓器，我們有 適合你的完美電腦。它被稱為Jetson。所以在頂部是⽤於 訓練⼈⼯智慧的DGX，Jetson是⾃主處理器，在中間， 我們需要另⼀台電腦。然⽽，⼤型語⾔模型的好處是你提 供例⼦然後進⾏強化學習⼈類反饋。機器⼈的強化學習⼈ 類反饋是什麼？嗯，那就是強化學習物理反饋。這就是你 如何校準機器⼈。這就是機器⼈知道當它學習這些關節能 ⼒和操作能⼒時，它將如何適當地適應物理定律。

And so we need a simulation engine that represents the world visually for the robot so that the robot has a gym to go learn how to be a robot. We call that virtual world omniverse. And the computer that runs omniverse is called o v x. And OVX, the computer itself is hosted in the Azure cloud. Okay? And so basically, we built these three things, these three systems. On top of it, we have algorithms for every single one. Now, I'm gonna show you one super example of how AI and Omniverse are gonna work together. The example I'm gonna show you is kind of insane, but it's going to be very, very close to tomorrow. It's a robotics building.

因此，我們需要⼀個模擬引擎，為機器⼈以視覺⽅式呈現 世界，讓機器⼈有⼀個場所可以學習如何成為⼀個機器 ⼈。我們稱這個虛擬世界為全宇宙。⽽運⾏全宇宙的電腦 被稱為 OVX。OVX這台電腦本⾝是託管在 Azure 雲端 上。好嗎？因此，基本上，我們建立了這三個東⻄，這三 個系統。在其上，我們為每⼀個系統都設計了演算法。現 在，我要給你展⽰⼀個超級範例，展⽰⼈⼯智慧和全宇宙 如何⼀起運作。我要給你展⽰的範例有點瘋狂，但它將非 常接近未來。這是⼀座機器⼈建築。

This robotics building is called the warehouse. Inside the robotics building are gonna be some autonomous systems. Some of the autonomous systems are gonna be called humans. And some of the autonomous systems are gonna be called forklifts. And these autonomous systems are going to interact with each other, of course, autonomously, and it's going to be overlooked upon by this warehouse to keep everybody out of harm's way. The warehouse is essentially an air traffic controller. And whenever it sees something happening, it will redirect traffic and give new way points, just new way points to the robots and the people, and they'll know exactly what to do. This warehouse, this building, you can also talk to. Of course, you could talk to it. Hey, you know, SAP Center, how are you feeling today?

這座機器⼈建築被稱為倉庫。在這座機器⼈建築內將會有 ⼀些⾃主系統。其中⼀些⾃主系統將被稱為⼈類。⽽另⼀ 些⾃主系統將被稱為堆⾼機。這些⾃主系統將會互動，當 然是⾃主地，⽽這個倉庫將會監控以確保每個⼈都不受傷 害。這個倉庫本質上就像是⼀個航空交通管制員。每當它 看到發⽣什麼事情，它就會᯿新導向交通並給予新的路徑 點，只是給機器⼈和⼈類新的路徑點，他們就會知道該怎 麼做。這個倉庫，這座建築，你也可以和它交談。當然， 你可以和它交談。嘿，你好，SAP中⼼，你今天感覺如 何？

For example. And so you could ask the same the warehouse the same questions. Basically, the system I just described will have Omniverse Cloud that's hosting the virtual simulation and AI running on DGX Cloud. And all of this is running in real time. Let's take a look. The future of heavy industries starts as a digital twin. The AI agents helping robots, workers and infrastructure navigate unpredictable events in complex industrial spaces will be built and evaluated first in sophisticated digital twins. This omniverse digital twin of a 100000 square foot warehouse is operating as a simulation environment that integrates digital workers, AMRs running the nvidia Isaac receptor stack, centralized activity maps of the entire warehouse from 100 simulated ceiling mount cameras using Nvidia Metropolis and AMR route planning with Nvidia cuopt. Software in loop testing of AI agents in this physically accurate simulated environment enables us to evaluate and refine how the system adapts to real world unpredictability. Here, an incident occurs along this AMR's planned route blocking its path as it moves to pick up a pallet.

例如，你可以向倉庫提出相同的問題。基本上，我剛剛描 述的系統將在Omniverse Cloud上托管虛擬模擬和在DGX Cloud上運⾏的⼈⼯智慧。⽽且所有這些都是實時運⾏ 的。讓我們來看看。᯿⼯業的未來始於數位孿⽣。在複雜 的⼯業空間中，協助機器⼈、⼯⼈和基礎設施應對不可預 測事件的⼈⼯智慧代理將⾸先在精密的數位孿⽣中建立和 評估。這個佔地10萬平⽅英尺的倉庫的Omniverse數位孿 ⽣作為⼀個模擬環境運作，整合了數位⼯⼈、運⾏著 nvidia Isaac receptor堆疊的AMR、使⽤Nvidia Metropolis的100個模擬天花板掛載攝影機的中央活動地 圖，以及使⽤Nvidia cuopt的AMR路線規劃。在這個物理 準確的模擬環境中對⼈⼯智慧代理進⾏軟體測試，使我們 能夠評估和完善系統如何適應現實世界的不可預測性。在 這裡，⼀個事件發⽣在這個AMR計劃的路線上，阻礙了 它前往拾取貨物的路徑。

NVIDIA Metropolis updates and sends a real time occupancy map to cuop where a new optimal route is calculated. The AMR is enabled to see around corners and improve its mission efficiency. With generative AI powered metropolis vision foundation models, operators can even ask questions using natural language. The visual model understands nuanced activity and can offer immediate insights to improve operations. All of the sensor data is created in simulation and passed to the real time AI running as NVIDIA Inference Microservices or NIMs. And when the AI is ready to be deployed in the physical twin, the real warehouse, we connect Metropolis and Isaac NIMS to real sensors with the ability for continuous improvement of both the digital twin and the AI models. Isn't that incredible? And so remember remember, a future facility, warehouse factory building will be software defined. And so the software is running. How else would you test the software?

NVIDIA Metropolis 更新並即時傳送佔⽤地圖⾄ cuop，從 ⽽計算出新的最佳路線。⾃主移動機器⼈（AMR）能夠 看⾒⾓落，提升其任務效率。憑藉⽣成式⼈⼯智慧技術⽀ 持的都會視覺基礎模型，操作員甚⾄可以⽤⾃然語⾔提 問。視覺模型能夠理解微妙的活動並提供即時⾒解以改善 運營。所有感應器數據都是在模擬中創建，並傳遞給運⾏ 為 NVIDIA 推理微服務（NIMs）的即時⼈⼯智慧。當⼈ ⼯智慧準備部署到實際的倉庫時，我們將 Metropolis 和 Isaac NIMS 連接到真實感應器，以持續改進數位孿⽣和 ⼈⼯智慧模型。這不是令⼈難以置信嗎？因此請記住，未 來的設施、倉庫、⼯廠將由軟體定義。因此軟體正在運 ⾏。還有其他⽅法可以測試軟體嗎？

So you test the software to building the warehouse, the optimization system in the digital twin. What about all the robots? All of those robots you were seeing just now, they're all running their own autonomous robotic stack. And so the way you integrate software in the future, CICD in the future for robotic systems is with digital twins. We've made Omniverse a lot easier to access. We're going to create basically Omniverse Cloud APIs, 4 simple API in a channel and you can connect your application to it. So this is this is going to be as wonderfully, beautifully simple in the future that Omniverse is going to be. And with these APIs, you're gonna have these magical digital twin capability. We also have turned Omniverse into an AI and integrated it with the ability to chat USD, the language of our language is, you know, human, and, omniverse's language as it turns out is universal scene description. And so, that language is rather complex and so we've taught our Omniverse that language.

你測試軟體以建立倉庫，優化系統在數位孿⽣體中。那麼 所有的機器⼈呢？你剛剛看到的所有那些機器⼈，它們都 在運⾏⾃⼰的⾃主機器⼈堆疊。因此，在未來整合軟體、 未來機器⼈系統的CICD⽅式是透過數位孿⽣體。我們讓 Omniverse更容易存取。我們將基本上創建Omniverse雲 端API，4個簡單的API在⼀個頻道，你可以將你的應⽤程 式連接到它。因此，未來Omniverse將會變得非常美好、 簡單。有了這些API，你將擁有這些神奇的數位孿⽣能 ⼒。我們還將Omniverse轉變為⼈⼯智慧，並將其整合到 能夠與USD進⾏對話的能⼒，我們的語⾔是⼈類的語 ⾔，⽽Omniverse的語⾔則是通⽤場景描述。因此，那種 語⾔相當複雜，所以我們教會了我們的Omniverse那種語 ⾔。

And so, you can speak to it in English and it would directly generate USD, and it would talk back in USD, but converts back to you in English. You could also look for information in this world semantically. Instead of the world being encoded semantically in in language, now it's encoded semantically in scenes. And so you could ask it of of, certain objects or certain conditions or certain scenarios, and it can go and find that scenario for you. It also can collaborate with you in generation. You could design some things in 3 d, it could simulate some things in 3 d, or you could use AI to generate something in 3 d. Let's take a look at how this is all going to work. We have a great partnership with Siemens. Siemens is the world's largest industrial engineering and operations platform. You've seen now so many different companies in the industrial space.

因此，你可以⽤英⽂與它溝通，它會直接產⽣美元，並且 會⽤美元回應你，但最後轉換成英⽂回傳給你。你也可以 在這個世界裡以語義⽅式尋找資訊。現在不再是以語⾔編 碼的⽅式進⾏語義編碼，⽽是以場景的⽅式進⾏語義編 碼。因此，你可以詢問特定的物件、特定的條件或特定的 情境，它可以幫你找到相應的情境。它也可以與你合作⽣ 成。你可以設計⼀些三維物品，它可以模擬⼀些三維物 品，或者你可以使⽤⼈⼯智慧⽣成⼀些三維物品。讓我們 來看看這⼀切將如何運作。我們與⻄⾨⼦有著良好的合作 夥伴關係。⻄⾨⼦是全球最⼤的⼯業⼯程和運營平台。你 現在已經看到了⼯業領域中許多不同的公司。

Heavy Industries is one of the greatest final frontiers of IT. And we finally now have the necessary technology to go and make a real impact. Siemens is building the industrial metaverse. And today, we're announcing that Siemens is connecting their crown jewel accelerator to NVIDIA Omniverse. Let's take a look. Siemens technology is transformed every day for everyone. Teamcenter X, our leading product lifecycle management software from the Siemens Accelerator platform, is used every day by our customers to develop and deliver products at scale. Now we are bringing the real and digital worlds even closer by integrating NVIDIA AI and Omniverse Technologies into Teamcenter X. Omniverse APIs enable data interoperability and physics based rendering to industrial scale design and manufacturing projects. Our customers, HD Hyundai, market leader in sustainable ship manufacturing, builds ammonia and hydrogen powered chips, often comprising over 7,000,000 discrete teamcenter x lets companies like HD Hyundai unify and visualize these massive engineering datasets interactively.

⼯業是資訊科技的最後⼀塊領域之⼀。我們終於擁有必 要的技術來實現真正的影響。⻄⾨⼦正在建立⼯業元宇 宙。今天，我們宣布⻄⾨⼦將他們的明星加速器連接到 NVIDIA Omniverse。讓我們來看看。⻄⾨⼦的技術每天 都在為每個⼈進⾏轉變。Teamcenter X，我們在⻄⾨⼦ 加速器平台上領先的產品⽣命週期管理軟件，每天被我們 的客⼾⽤來開發和交付⼤規模產品。現在，我們通過將 NVIDIA⼈⼯智慧和Omniverse技術整合到Teamcenter X 中，將現實世界和數位世界更加緊密地結合。Omniverse 應⽤程式介⾯實現了數據互通性和基於物理的渲染，適⽤ 於⼯業規模的設計和製造項⽬。我們的客⼾，HD現代， 可持續造船市場的領導者，製造氨和氫動⼒晶片，通常包 含超過7,000,000個離散的Teamcenter X，讓像HD現代 這樣的公司能夠統⼀和互動地視覺化這些龐⼤的⼯程數據 集。

It integrates generative AI to generate 3 d objects or HDRI backgrounds to see their projects in context. The result, an ultra intuitive photoreal physics based digital twin that eliminates waste and errors, delivering huge savings in cost and time. And we are building this for collaboration, whether across more Siemens accelerator tools like Siemens NX or STAR CCM plus or across teams working on their favorite devices in the same scene together. And this is just the beginning. Working with NVIDIA, we will bring accelerated computing, generative AI, and Omniverse integration across the Siemens Xcelerator portfolio. The professional voice actor happens to be a good friend of mine, Roland Busch, who happens to be the CEO of Siemens. Once you get Omniverse connected into your workflow, your ecosystem, From the beginning of your design to engineering, to manufacturing planning, all the way to digital twin operations, Once you connect everything together, it's insane how much productivity you can get, and it's just really, really wonderful. All of a sudden, everybody's operating on the same ground truth. You don't have to exchange data and convert data, make mistakes, everybody is working on the same ground truth. From the design department to the art department, the architecture department, all the way to the engineering and even the marketing department.

它整合了⽣成式⼈⼯智慧，可以⽣成3D物件或⾼動態範 圍影像背景，讓使⽤者能夠在實際情境中觀看他們的專 案。結果是⼀個超直觀的照片級物理基礎數位孿⽣體，可 以消除浪費和錯誤，帶來巨⼤的成本和時間節省。我們正 在為協作⽽建立這個系統，無論是跨越更多⻄⾨⼦加速器 ⼯具，如⻄⾨⼦NX或STAR CCM plus，或是跨越在同⼀ 場景中共同⼯作的團隊。這只是個開始。與NVIDIA合 作，我們將在⻄⾨⼦Xcelerator系列產品中引入加速運 算、⽣成式⼈⼯智慧和Omniverse整合。專業的配⾳演員 碰巧是我的好朋友，羅蘭·布施，他也恰好是⻄⾨⼦的 CEO。⼀旦你將Omniverse整合到你的⼯作流程、⽣態系 統中，從設計開始，到⼯程、製造計劃，⼀直到數位孿⽣ 操作，⼀旦將所有事物連結在⼀起，你將會驚訝於你可以 獲得多少⽣產⼒，這真的非常美妙。突然之間，每個⼈都 在同⼀個真實基礎上運作。你不需要交換資料、轉換資 料、犯錯誤，每個⼈都在同⼀個真實基礎上⼯作。從設計 部⾨到藝術部⾨、建築部⾨，⼀直到⼯程甚⾄⾏銷部⾨。

Let's take a look at how Nissan has integrated Omniverse into their workflow. And it's all because it's connected by all these wonderful tools and these developers that we're working with. Take a look. That was not an animation. That was omniverse. Today, we're announcing that omniverse cloud streams to the vision pro. It is very, very strange that you walk around virtual doors when I was getting out of that car, and everybody does it. It is really, really quite amazing. VisionPRO connected to Omniverse portals you into Omniverse. Because all of these CAD tools and all these different design tools are now integrated and connected to Omniverse, you can have this type of workflow.

讓我們來看看⽇產如何將Omniverse整合到他們的⼯作流 程中。這⼀切都是因為與我們合作的這些出⾊⼯具和開發 ⼈員的連接。看看吧。那不是動畫。那就是Omniverse。 今天，我們宣布Omniverse雲端串流到Vision Pro。當我 從⾞⼦裡走出來時，你走過虛擬⾨真的非常奇妙，每個⼈ 都這麼做。這真的非常令⼈驚奇。VisionPRO連接到 Omniverse，讓你進入Omniverse。因為所有這些CAD⼯ 具和各種不同的設計⼯具現在都整合並連接到 Omniverse，你可以擁有這種類型的⼯作流程。

Really incredible. Let's talk about robotics. Everything that moves will be robotic. There's no question about that. It's safer, it's more convenient. And one of the largest industries is going to be automotive. We build the robotic stack from top to bottom, as I was mentioned. From the computer system, but in the case of self driving cars, including the self driving application. At the end of this year or I guess beginning of next year, we will be shipping in Mercedes, and then shortly after that, JLR. And so these autonomous robotic systems are software defined, they take a lot of work to do, has computer vision, has obviously artificial intelligence, control and planning, all kinds of very complicated technology, and takes years to refine.

真是令⼈難以置信。讓我們來談談機器⼈技術。所有能夠 移動的東⻄都將是機器⼈化的。這⼀點毋庸置疑。這不僅 更安全，也更⽅便。其中最⼤的產業之⼀將是汽⾞業。我 們從頂端到底端都在建構機器⼈技術，正如我所提到的。 從電腦系統開始，但在⾃動駕駛汽⾞的情況下，包括⾃動 駕駛應⽤程式。在今年年底或明年初，我們將在奔馳⾞上 應⽤這項技術，然後很快地在捷豹路虎上也會推出。這些 ⾃主的機器⼈系統是由軟體定義的，需要⼤ᰁ的⼯作，包 括電腦視覺、顯然的⼈⼯智慧、控制和規劃，以及各種非 常複雜的技術，需要多年的精煉。

We're building the entire stack. However, we open up our entire stack for all of the automotive industry. This is just the way we work. The way we work in every single industry, we try to build as much of it as we can so that we understand it, but then we open it up so that everybody can access it. Whether you would like to buy just our computer, which is the world's only full, functional, safe, ASLD system that can run AI, this functional, safe ASLD quality computer, or the operating system on top, or, of course, our data centers, which is in basically every AV company in the world. However you would like to enjoy it, we're delighted by it. Today, we're announcing that BYD, the world's largest EV company, is adopting our next generation. It's called Thor. Thor is designed for transformer engines. Thor, our next generation AV computer will be used by BYD.

我們正在打造整個系統架構。然⽽，我們將我們的整個系 統架構開放給整個汽⾞⾏業。這就是我們的⼯作⽅式。在 每個⾏業中，我們都會盡可能地建立更多，以便了解它， 然後開放給所有⼈使⽤。無論您想要購買我們的電腦，這 是全球唯⼀可以運⾏⼈⼯智慧的完整、功能⿑全、安全的 ASLD系統，這款功能⿑全、安全的ASLD品質電腦，或 者頂部的作業系統，當然還有我們的數據中⼼，基本上每 家⾃動駕駛公司都在使⽤。無論您想如何享受它，我們都 感到⾼興。今天，我們宣布，全球最⼤的電動⾞公司比亞 迪正在採⽤我們的下⼀代。它被稱為Thor。Thor是為變 壓器引擎設計的。我們的下⼀代⾃動駕駛電腦Thor將被 比亞迪使⽤。

You probably don't know this fact that we have over a 1000000 robotics developers. We created Jetson, this robotics computer. We're so proud of it. The amount of software that goes on top of it is insane. But the reason why we can do it at all is because it's a 100% CUDA compatible. Everything that we do, everything that we do in our company is in service of our developers. And by us being able to maintain this rich ecosystem and make it compatible with everything that you access from us, we can bring all of that incredible capability to this little tiny computer we call Jetson, a robotics computer. We also today are announcing this incredibly advanced new SDK. We call it Isaac Perceptor. Isaac Perceptor Most robots today are preprogrammed.

你可能不知道這個事實，我們擁有超過⼀百萬名機器⼈開 發者。我們創造了Jetson，這台機器⼈電腦。我們對此感 到非常⾃豪。安裝在上⾯的軟體ᰁ是驚⼈的。但我們之所 以能夠做到這⼀切，是因為它是100% CUDA 相容的。我 們公司所做的⼀切，都是為了我們的開發者服務。通過我 們能夠維護這個豐富的⽣態系統並使其與我們提供的所有 內容兼容，我們可以將所有這些令⼈難以置信的功能帶到 我們稱之為Jetson的這台⼩⼩的機器⼈電腦上。我們今天 還宣布了這個極其先進的新 SDK。我們稱之為Isaac Perceptor。Isaac Perceptor。今天⼤多數機器⼈都是預 先編程的。

They're either following rails on the ground, digital rails, or they'd be following April tags, But in the future, they're going to have perception. The reason why you want that is so that you could easily program it. You say, I would like to go from point a and point b, And it will figure out a way to navigate its way there. So by only programming waypoints, the entire route could be adaptive. The entire environment could be reprogrammed, just as I showed you at the very beginning with the warehouse. You can't do that with preprogrammed AGVs. If those boxes fall down, they just all gum up and they just wait there for somebody to come clear it. And so now, with the Isaac Perceptor, we have incredible state of the art vision odometry, 3 d reconstruction, and in addition to 3 d reconstruction, depth perception. The reason for that is so that you can have 2 modalities to keep an eye on what's happening in the world. Isaac Perceptor, the most used robot today is the manipulator, Manufacturing arms, and they are also pre programmed.

它們要麼沿著地⾯的軌道⾏進，要麼是沿著數位軌道⾏ 進，或者它們將跟隨四⽉標籤，但在未來，它們將具有感 知能⼒。你希望這樣做的原因是為了讓你能夠輕鬆地對其 進⾏編程。你說，我想從點A到點B，它將找到⼀種導航 ⽅式來到達那裡。因此，只需編程航點，整個路線就可以 是適應性的。整個環境可以᯿新編程，就像我⼀開始在倉 庫中向你展⽰的那樣。你無法⽤預先編程的AGV做到這 ⼀點。如果那些箱⼦掉下來，它們只會全部堵塞在那裡， 等待有⼈來清理。因此，現在有了Isaac感知器，我們擁 有了令⼈難以置信的最先進的視覺測距、三維᯿建，以及 除了三維᯿建之外的深度感知。這樣做的原因是為了讓你 能夠⽤兩種模式監視世界上正在發⽣的事情。Isaac感知 器，今天最常⽤的機器⼈是機械⼿臂，製造⼿臂，它們也 是預先編程的。

The computer vision algorithms, the AI algorithms, the control and path planning algorithms that are geometry aware, incredibly computationally intensive. We have made these CUDA accelerated. So we have the world's 1st CUDA accelerated motion planner that is geometry aware. You put something in front of it, it comes up with a new plan and articulates around it. It has excellent perception for pose estimation of a 3 d object. Not just not it's pose in 2 d, but it's pose in 3 d. So it has to imagine what's around and how best to grab it. So the foundation pose, the grip foundation, and the, articulation algorithms are now available. We call it Isaac manipulator, and they also, just run on NVIDIA's computers. We are starting to do some really great work in the next generation of robotics.

電腦視覺演算法、⼈⼯智慧演算法、以及具有幾何意識的 控制和路徑規劃演算法，都需要極⾼的運算ᰁ。我們已經 將這些演算法加速使⽤CUDA。因此，我們擁有世界上第 ⼀個具有幾何意識的CUDA加速運動規劃器。當你把東⻄ 放在它⾯前時，它會提出新的計劃並繞過它。它對於三維 物體的姿態估計具有出⾊的感知能⼒。不僅僅是在⼆維中 的姿態，⽽是在三維中的姿態。因此，它必須想像周圍的 情況以及如何最好地抓取它。因此，基礎姿態、握持基礎 和關節演算法現在都已經可⽤。我們稱之為Isaac manipulator，它們也可以在NVIDIA的電腦上運⾏。我們 正開始在下⼀代機器⼈技術上進⾏⼀些非常出⾊的⼯作。

The next generation of robotics will likely be a humanoid robotics. We now have the necessary technology, and as I was describing earlier, the necessary technology to imagine generalized human robotics. In a way, human robotics is likely easier, and the reason for that is because we have a lot more imitation training data that we can provide the robots because we are constructed in a very similar way. It is very likely that the human and robotics will be much more useful in our world because we created the world to be something that we can interoperate in and work well in. And the way that we set up our workstations and manufacturing and logistics, they were designed for for humans, they were designed for people. And so these human robotics will likely be much more productive to deploy. While we're creating, just like we're doing with the others, the entire stack. Starting from the top, a foundation model that learns from watching video, human examples. It could be in video form, it could be in virtual reality form. We then created a gym for it called Isaac Reinforcement Learning Gym, which allows the humanoid robot to learn how to adapt to the physical world.

下⼀代的機器⼈很可能會是⼈形機器⼈。我們現在已經擁 有必要的技術，就像我之前所描述的，我們有想像通⽤⼈ 類機器⼈所需的技術。從某種程度上來說，⼈類機器⼈可 能會更容易，原因是因為我們有更多可以提供給機器⼈的 模仿訓練數據，因為我們的構造⽅式非常相似。很可能⼈ 類和機器⼈在我們的世界中會更加有⽤，因為我們創造了 ⼀個我們可以互動並且能夠良好運作的世界。我們設置⼯ 作站、製造和物流的⽅式，都是為⼈類⽽設計的。因此， 這些⼈形機器⼈很可能會更有⽣產⼒可以投入使⽤。當我 們創建時，就像我們對其他事物所做的⼀樣，整個系統都 是從頂部開始。從⼀個從觀看影片、⼈類⽰範中學習的基 礎模型開始。這可以是以影片形式，也可以是以虛擬現實 形式。然後我們為它創建了⼀個健⾝房，名為Isaac Reinforcement Learning Gym，這讓⼈形機器⼈可以學 習如何適應物理世界。

And then an incredible computer, the same computer that's going to go into a robotic car, this computer will run inside a humanoid robot called Thor. It's designed for Transformer engines. We've combined several of these into one video. This is something that you're gonna really love. Take a look. It's not enough for humans to imagine. We have to invent and explore And push beyond what's been done. It's a fair amount of detail. We create smarter and faster. We push it to fail so it can learn.

接著是⼀台令⼈難以置信的電腦，這台電腦將被應⽤在⼀ 輛機器⼈汽⾞中，同時也會運⾏在⼀個名為Thor的⼈形 機器⼈內。它是為變形引擎⽽設計的。我們將這些元素結 合成⼀個影片。這是⼀個你會非常喜歡的東⻄。來看看 吧。對於⼈類來說，僅僅想像是不夠的。我們必須去發明 和探索，並超越已經完成的事情。這需要相當多的細節。 我們創造更智能、更快速的東⻄。我們推動它去失敗，這 樣它才能學習。

We teach it then help it teach itself. We broaden its understanding to take on new challenges with absolute precision and succeed. We make it perceive and move and even reason so it can share our world with us. This is where inspiration leads us, the next frontier. This is NVIDIA Project Root. A general purpose foundation model for humanoid robot learning. The group model takes multimodal instructions and past interactions as input and produces the next action for the robot to execute. We developed scale out with Osmo, a new compute orchestration service that coordinates workflows across DGX systems for training and OVX systems for simulation. With these tools, we can train Groot in physically based simulation and transfer zero shot to the real world. The Groot model will enable a robot to learn from a handful of human demonstrations, so so it can help with everyday tasks and emulate human movement just by observing us.

我們教導它，然後幫助它⾃⼰學習。我們擴展它的理解能 ⼒，讓它能夠以絕對精確度應對新挑戰並取得成功。我們 讓它能夠感知、移動，甚⾄推理，這樣它就能與我們分享 我們的世界。這就是靈感引領我們的地⽅，下⼀個前沿。 這就是 NVIDIA Project Root。這是⼀個⽤於⼈形機器⼈ 學習的通⽤基礎模型。這個群體模型將多模式指令和過去 的互動作為輸入，並為機器⼈執⾏下⼀個動作。我們開發 了⼀個名為 Osmo 的新計算協調服務，⽤於在 DGX 系統 之間進⾏訓練和 OVX 系統之間進⾏模擬的⼯作流程協 調。有了這些⼯具，我們可以在基於物理的模擬中訓練 Groot，並將零樣本轉移到現實世界。Groot 模型將使機 器⼈能夠從少數⼈類⽰範中學習，這樣它就可以幫助處理 ⽇常任務，並通過觀察我們來模擬⼈類動作。

This is made possible with NVIDIA's technologies that can understand humans from videos, train models and simulation, and ultimately deploy them directly to physical robots. Connecting Groot to a large language model even allows it to generate motions by following natural language instructions. Hi, Jiawan. Can you give me a high five? Sure thing. Let's high five. Can you give her some cool moves? Dirt. Check this out. All this incredible intelligence is powered by the new Jetson Thor robotics chips designed for Groot, built for the future.

這是透過 NVIDIA 的技術實現的，能夠從影片中理解⼈ 類，訓練模型和模擬，最終直接部署到實體機器⼈上。將 Groot 連接到⼀個⼤型語⾔模型，甚⾄可以通過⾃然語⾔ 指令⽣成動作。嗨，Jiawan。你能給我擊掌嗎？當然可 以。讓我們擊掌。你能給她⼀些酷炫的動作嗎？沒問題。 看這個。所有這些令⼈難以置信的智能都是由為 Groot 設計、為未來打造的新 Jetson Thor 機器⼈芯片所驅動。

With Isaac Lab, OSMO, and Groot, we're providing the building blocks for the next generation of AI powered robotics. About the same size. The soul of NVIDIA, the intersection of computer graphics, physics, artificial intelligence, It all came to bear at this moment. The name of that project was General Robotics 3. I know. Super good. Super good. Well, I think we have some special guests. Do we? Hey, guys. 透過Isaac Lab、OSMO和Groot，我們正在為下⼀代AI動 ⼒機器⼈提供基礎。⼤⼩差不多。NVIDIA的靈魂，電腦 圖形、物理、⼈⼯智慧的交集，在這⼀刻全部展現出來。 那個專案的名字叫做General Robotics 3。我知道。超棒 的。超棒的。嗯，我想我們有⼀些特別的來賓。對吧？ 嘿，⼤家。

So I understand you guys are powered by Jetson. They're powered by Jetsons, little Jetson robotics computers inside. They learn to walk in Isaac Sim. Ladies and gentlemen, this this is orange, and this is the famous green. They are the BDX robots of Disney. Amazing Disney research. Come on, you guys. Let's wrap up. Let's go. 5 things.

我了解你們是由傑森（Jetson）提供動⼒。他們內部裝有 ⼩傑森機器⼈電腦。他們在艾薩克模擬器（Isaac Sim） 中學會走路。女⼠們先⽣們，這是橘⾊的，這是著名的綠 ⾊。他們是迪⼠尼的BDX機器⼈。令⼈驚嘆的迪⼠尼研 究。來吧，各位。讓我們結束吧。走吧。五件事。

Where are you going? I sit right here. Don't be afraid. Come here, Green. Hurry up. What are you saying? No, it's not time to eat. I'll give I'll give you a snack in a moment. Let me finish up real quick. Come on, Green.

你要去哪裡？我就坐在這裡。不要害怕。過來，⼩綠。快 點。你在說什麼？不，現在還不是吃飯的時間。等⼀下我 會給你點⼼。讓我快點做完。來吧，⼩綠。

Hurry up. Stop wasting time. Five things. 5 things. 1st, a new industrial revolution. Every data center should be accelerated. A $1,000,000,000,000 worth of installed data centers will become modernized over the next several years. 2nd, because of the computational capability we brought to bear, a new way of doing software has emerged. Generative AI, which is going to create new infrastructure dedicated to doing one thing and one thing only. Not for multi user data centers, but AI generators.

趕快。別再浪費時間。五件事。第⼀，⼀場新的⼯業⾰ 命。每個資料中⼼都應該加速發展。未來幾年將有價值⼀ 兆美元的資料中⼼將會現代化。第⼆，由於我們帶來的計 算能⼒，⼀種新的軟體開發⽅式已經出現。⽣成式⼈⼯智 慧，將會創建專⾨⽤於執⾏單⼀任務的新基礎設施。不是 為多⽤⼾資料中⼼，⽽是為⼈⼯智慧⽣成器。

These AI generation will create incredibly valuable software. A new industrial revolution. 2nd, the computer of this revolution, the computer of this generation, generative AI, trillion parameters, Blackwell. Insane amounts of computers, computing. 3rd, I'm trying to concentrate. Good job. 3rd, new computer, new computer creates new types of software. New types of software should be distributed in a new way so that it can, on the one hand, be an endpoint in the Cloud and easy to use, but still allow you to take it with you because it is your intelligence. Your intelligence should be packed packaged up in a way that allows you to take it with you. We call them NIMs.

這些 AI 產⽣將創造出非常有價值的軟體。⼀場新的⼯業 ⾰命。其次，這場⾰命的電腦，這⼀代的電腦，⽣成式 AI，兆個參數，布萊克威爾。龐⼤的電腦ᰁ，運算。第 三，我正在試著專⼼。做得好。第三，新電腦，新電腦創 造出新型軟體。新型軟體應該以⼀種新的⽅式分發，這樣 ⼀來，它可以⼀⽅⾯成為雲端的端點並易於使⽤，但仍然 允許您攜帶它，因為這是您的智慧。您的智慧應該被包裝 成⼀種讓您可以攜帶的⽅式。我們稱它們為 NIMs。

And 3rd, these NIMs are going to help you create a new type of application for the future. Not one that you wrote completely from scratch, but you're going to integrate them like Teams. Create these applications. We have a fantastic capability between NIMS, the AI technology, the tools, and the infrastructure DGX Cloud in our AI Foundry to help you create proprietary applications, proprietary chat bots. And then lastly, everything that moves in the future will be robotic. You're not going to be the only one. And these robotic systems, whether they are humanoid, AMRs, self driving cars, forklifts, manipulating arms, they will all need one thing. Giant stadiums, warehouses, factories. There can be factories that are robotic, orchestrating factories, manufacturing lines that are robotics, building cars that are robotics. These systems all need one thing.

第三，這些新興的智能裝置將幫助您打造未來的新型應⽤ 程式。不是完全從頭開始撰寫的那種，⽽是您將像 Teams ⼀樣整合它們。創建這些應⽤程式。我們在 AI Foundry 中擁有 NIMS、⼈⼯智慧技術、⼯具和基礎設施 DGX Cloud 之間的出⾊能⼒，可協助您創建專有應⽤程 式、專有聊天機器⼈。最後，未來所有的移動都將是機器 ⼈化的。您不會是唯⼀的⼀個。這些機器⼈系統，無論是 ⼈形機器⼈、AMRs、⾃駕⾞、堆⾼機、操作⼿臂，它們 都需要⼀個共同的東⻄。巨⼤的體育場、倉庫、⼯廠。可 以有全⾃動化的⼯廠，協調⼯廠，機器⼈製造線，製造汽 ⾞的機器⼈。這些系統都需要⼀個共同的東⻄。

They need a platform, a digital platform, a digital twin platform, and we call that Omniverse, the operating system of the robotics world. These are the 5 things that we talked about today. What does NVIDIA look like? What does NVIDIA look like? When we talk about GPUs, there's a very different image that I have when I when people ask me about GPUs. 1st, I see a bunch of software stacks and things like that. 2nd, I see this. This is what we announced to you today. This is Blackwell. This is the platform.

他們需要⼀個平台，⼀個數位平台，⼀個數位孿⽣平台， 我們稱之為Omniverse，這是機器⼈世界的作業系統。這 就是我們今天談到的五件事情。NVIDIA是什麼樣⼦？當 我們談到GPU時，我對於NVIDIA的印象完全不同。⾸ 先，我看到⼀堆軟體堆疊和類似的東⻄。其次，我看到這 個。這就是我們今天向你們宣布的。這是Blackwell。這 就是這個平台。

Amazing, amazing processors, NVLink switches, networking systems, and the system design is a miracle. This is Blackwell. This, to me, is what a GPU looks like in my mind. Listen, orange, green. I think we have one more treat for everybody. What do you think? Should we? Okay. We have one more thing to show you. Roll it.

令⼈驚嘆的處理器、NVLink交換器、網路系統，以及系 統設計真是奇蹟。這就是Blackwell。對我來說，這才是 我⼼⽬中的GPU樣貌。聽著，橘⾊、綠⾊。我想我們還 有⼀個驚喜要給⼤家。你們覺得呢？要不要？好的。我們 還有⼀樣東⻄要給你們看。播放吧。 Thank you. Thank you. Have a great have a great GTC. Thank you all for coming. Thank you. 謝謝⼤家。謝謝⼤家。祝⼤家參加台北電腦應⽤展有個美 好的⼀天。感謝⼤家的到來。謝謝。