Name: 埃隆-馬斯克為何豪賭超級計算機，以推動特斯拉和 xAI 的發展？ (Why Elon Musk Is Betting Big On Supercomputers To Boost Tesla And xAI)
Uploaded: 2024-09-25T06:24:28.000Z
Duration: 15 min 53 s

Tech titan Elon Musk is known for being a car guy, a rocket guy, a social media guy, and now he's also a supercomputer guy.

科技巨頭埃隆-馬斯克（Elon Musk）以汽車人、火箭人、社交媒體人著稱，現在他還是超級計算機人。

And Elon Musk says Tesla will spend well over $1 billion by the end of 2024 on building an in-house supercomputer known as Project Dojo.

埃隆-馬斯克（Elon Musk）表示，到 2024 年底，特斯拉將斥資超過 10 億美元，建造一個名為 "Project Dojo "的內部超級計算機。

Although supercomputers look a lot like data centers, they're designed to perform calculations and process data at extremely high speeds.

雖然超級計算機看起來很像數據中心，但其設計目的是以極高的速度進行計算和處理數據。

Both of them are about scaling up to very large amounts of computation.

However, like, you know, in a data center, you have a lot of small parallel tasks that are not necessarily connected to each other.

但是，在數據中心，有很多並行的小任務，它們之間並不一定有聯繫。

Whereas, for example, when you're training a very large AI model, those are not entirely independent computations.

舉例來說，當你訓練一個非常龐大的人工智能模型時，這些並不是完全獨立的計算。

So you do need tighter interconnection between those computations and the passing of data back and forth needs to be potentially at a much higher bandwidth and a much lower latency.

是以，這些計算之間需要更緊密的互聯，數據的來回傳輸需要更高的帶寬和更低的延遲。

Musk wants to use this supercomputing power to improve Tesla's autonomous driving capabilities and finally deliver on the company's years-long promise to bring robotaxis to market.

馬斯克希望利用這種超級計算能力來提高特斯拉的自動駕駛能力，並最終兌現公司多年來向市場推出機器人軸的承諾。

Supercomputers are also needed to train Tesla's humanoid robot, Optimus, which the company plans to use in its factories starting next year.

特斯拉還需要超級計算機來訓練仿人機器人 Optimus，該公司計劃從明年開始在工廠中使用該機器人。

All in all, Musk says that Tesla plans to spend $10 billion this year on AI.

總之，馬斯克表示，特斯拉計劃今年在人工智能上投入 100 億美元。

Musk's new AI venture, XAI, also needs powerful supercomputers to train XAI's chatbot, Grok, which directly competes with OpenAI's ChatGPT and Google's Gemini chatbots.

馬斯克的新人工智能企業XAI也需要強大的超級計算機來訓練XAI的哈拉機器人Grok，它與OpenAI的ChatGPT和谷歌的Gemini哈拉機器人直接競爭。

Several of Musk's supercomputer projects are already in development.

馬斯克的幾個超級計算機項目已經在開發中。

In August, Elon Musk teased Tesla's AI supercomputer cluster called Cortex-on-X.

今年 8 月，埃隆-馬斯克預告了特斯拉名為 Cortex-on-X 的人工智能超級計算機集群。

Cortex is being built at Tesla's Austin, Texas, headquarters.

Cortex 正在德克薩斯州奧斯汀的特斯拉總部製造。

Back in January, Tesla also announced that it planned to spend $500 million to build its Dojo supercomputer in Buffalo, New York.

今年 1 月，特斯拉還宣佈計劃斥資 5 億美元在紐約布法羅建造 Dojo 超級計算機。

Meanwhile, Musk just revealed that XAI's Colossus supercomputer in Memphis, Tennessee, was up and running.

與此同時，馬斯克剛剛透露，XAI 位於田納西州孟菲斯市的 Colossus 超級計算機已經開始運行。

CNBC wanted to learn more about what Musk's bet on supercomputers might mean for the future of his companies and the challenges he faces in the ultra-competitive world of AI development.

CNBC 想更多地瞭解馬斯克對超級計算機的押注對其公司未來的意義，以及他在競爭異常激烈的人工智能開發領域所面臨的挑戰。

You had supercomputers, if you go to any of the national labs, they're used for everything from simulation materials discovery to climate modeling to modeling nuclear reactions and so on and so forth.

你有超級計算機，如果你去任何一個國家實驗室，從模擬材料發現到氣候建模，再到核反應建模，等等等等，超級計算機無所不能。

However, what's unique about AI supercomputers is that they are entirely optimized for AI.

然而，人工智能超級計算機的獨特之處在於，它們完全針對人工智能進行了優化。

Musk launched XAI in 2023 to develop large language models and AI products like its chatbot, Grok, as an alternative to AI tools being created by OpenAI, Microsoft and Google.

馬斯克於 2023 年推出了 XAI，旨在開發大型語言模型和人工智能產品，如其哈拉機器人 Grok，以替代 OpenAI、微軟和谷歌正在開發的人工智能工具。

Despite being one of its original founders, Elon Musk left OpenAI in 2018 and has since become one of the company's harshest critics.

儘管埃隆-馬斯克是 OpenAI 的最初創始人之一，但他在 2018 年離開了 OpenAI，並從此成為該公司最嚴厲的批評者之一。

In June, it was announced that XAI would build a supercomputer in Memphis, Tennessee, to carry out the task of training Grok.

今年 6 月，XAI 宣佈將在田納西州孟菲斯市建造一臺超級計算機，以執行培訓 Grok 的任務。

It would represent the city's largest multibillion-dollar capital investment by a new-to-market company in Memphis history.

這將是孟菲斯歷史上新進入市場的公司最大的數十億美元資本投資。

The announcement came at the heels of XAI securing $6 billion in Series B funding, raising its valuation at the time from $18 to $24 billion.

這一消息是在 XAI 獲得 60 億美元 B 輪融資之後宣佈的，當時的估值從 180 億美元提高到 240 億美元。

By early September, Musk announced that his training supercluster in Memphis, called Colossus, was online.

9 月初，馬斯克宣佈，他在孟菲斯進行的名為 "巨像"（Colossus）的超級集群訓練已經上線。

The supercluster is powered by 100,000 NVIDIA H100 graphics processing units, or GPUs, making it the most powerful AI training system in the world, according to Musk.

據馬斯克稱，該超級集群由 10 萬個英偉達 H100 圖形處理器（GPU）驅動，是世界上最強大的人工智能訓練系統。

He went on to say that the cluster would double in size in the next few months.

他接著說，在接下來的幾個月裡，集群的規模將擴大一倍。

這些 GPU 已經存在了一段時間。

They started off in laptops, in desktops, to be able to offload graphics work from the core CPU.

它們最初用於筆記本電腦和臺式機，目的是將圖形處理工作從核心 CPU 上卸載下來。

If you go back sort of 10 years or so, 15 years ago, online gaming was blowing up and people wanted to game at speed.

追溯到 10 年左右，也就是 15 年前，在線遊戲風起雲湧，人們希望以最快的速度進行遊戲。

And then they realized that having graphics and the general purpose of the game on the same processor just led to constraints.

後來他們意識到，將圖形處理和遊戲的通用功能放在同一個處理器上只會造成限制。

So it's a very specific task to train a large language model.

是以，訓練大型語言模型是一項非常特殊的任務。

And doing that on a classic CPU, you can.

你可以在經典 CPU 上做到這一點。

But it's one of those examples where the particular architecture of a GPU plays well for that type of workload.

不過，這也是 GPU 的特定架構能很好地處理這類工作負載的例子之一。

In fact, GPUs became so popular that chipmakers like NVIDIA, for a time, had a hard time keeping up with demand.

事實上，GPU 變得如此流行，以至於英偉達（NVIDIA）為代表的芯片製造商一度難以滿足需求。

The fight for GPUs has even caused competition among Musk's own companies.

對 GPU 的爭奪甚至引發了馬斯克自己公司之間的競爭。

Musk's social media company X and XAI are closely intertwined, with X hosting XAI's Grok chatbot on its site and XAI using some capacity in X data centers to train the large language models that power Grok.

馬斯克的社交媒體公司X和XAI密切相關，X在其網站上託管了XAI的Grok哈拉機器人，而XAI則利用X數據中心的部分能力來訓練為Grok提供動力的大型語言模型。

In December 2023, Elon Musk ordered NVIDIA to ship 12,000 of these very coveted AI chips, H100 GPUs from NVIDIA, to X instead of to Tesla when they had been reserved for Tesla.

2023 年 12 月，埃隆-馬斯克命令英偉達（NVIDIA）向 X 公司而不是特斯拉公司運送 1.2 萬塊夢寐以求的人工智能芯片--英偉達（NVIDIA）的 H100 GPU，而這些芯片原本是留給特斯拉的。

So he effectively delayed Tesla's being able to continue building out data center and AI infrastructure by five, six months.

是以，他實際上將特斯拉繼續建設數據中心和人工智能基礎設施的時間延後了五、六個月。

The incident was one example that shareholders used in a lawsuit against Musk and Tesla's board of directors that accused them of fiduciary duty.

股東們在對馬斯克和特斯拉董事會提起的訴訟中，指控他們違反了信託責任，該事件就是其中一個例子。

They argued that after founding XAI, Musk began diverting scarce talent and resources from Tesla to his new company.

他們認為，在成立 XAI 之後，馬斯克開始將特斯拉稀缺的人才和資源轉移到他的新公司。

Musk defended his decision on X, saying that Tesla was not ready to utilize the chips and that they would have just sat in a warehouse had he not diverted them.

馬斯克為自己關於 X 的決定進行了辯護，他說特斯拉還沒有準備好利用這些芯片，如果不是他轉用了這些芯片，這些芯片就只能放在倉庫裡了。

Musk has gone as far as to suggest that Tesla should invest $5 billion into XAI.

馬斯克甚至建議特斯拉向 XAI 投資 50 億美元。

Still, Musk has big plans on how artificial intelligence can transform Tesla.

不過，對於人工智能如何改變特斯拉，馬斯克還是有很大的計劃。

In January, he wrote on X that Tesla should be Key to this transformation is Tesla's custom-built supercomputer called Dojo, details of which the company first publicly announced during Tesla's AI Day presentation in 2021.

今年 1 月，他在《X》雜誌上寫道："特斯拉轉型的關鍵在於特斯拉定製的超級計算機 Dojo，該公司在 2021 年特斯拉人工智能日的演講中首次公開宣佈了 Dojo 的詳細信息。

There's an insatiable demand for speed as well as capacity for neural network training.

人們對神經網絡訓練的速度和容量有著永不滿足的需求。

And a few years back, he asked us to design a super fast training computer.

幾年前，他要求我們設計一臺超快的訓練計算機。

就這樣，我們開始了 "道場計劃"。

During the company's Q2 earnings call last year, Musk told investors that Tesla would spend over a billion dollars on Dojo by the end of 2024.

在去年第二季度的財報電話會議上，馬斯克告訴投資者，到 2024 年底，特斯拉將在 Dojo 上花費超過 10 億美元。

A few months later, Morgan Stanley predicted that Dojo could boost Tesla's value by $500 billion.

幾個月後，摩根士丹利（Morgan Stanley）預測，Dojo 將使特斯拉的價值提升 5000 億美元。

Dojo's main job is to process and train AI models using the huge amounts of video and data captured by Tesla vehicles.

Dojo 的主要工作是利用特斯拉車輛捕獲的大量視頻和數據處理和訓練人工智能模型。

The goal is to improve Tesla's suite of driver assistance features, which the company calls Autopilot, as well as its more robust full self-driving or FSD system.

我們的目標是改進特斯拉的駕駛輔助功能套件（公司稱之為 Autopilot），以及更強大的完全自動駕駛系統或 FSD 系統。

They've sold, what is it, 5 million plus cars.

他們已經賣出了 500 多萬輛汽車。

Each one of those cars typically has eight cameras plus in it.

每輛車上一般都裝有八個攝像頭。

They're streaming all of that video back to Tesla.

他們正在將所有視頻流回特斯拉。

So what can they do with that training set?

那麼，他們能用訓練集做些什麼呢？

Obviously, they can develop full self-driving and they're getting close to that.

顯然，他們可以開發完全的自動駕駛技術，而且他們正在接近這一目標。

Despite their names, neither Autopilot nor FSD make Tesla vehicles autonomous and require active driver supervision, as Tesla states on its website.

儘管名稱相同，但正如特斯拉在其網站上聲明的那樣，Autopilot 和 FSD 都不能使特斯拉汽車自動駕駛，需要駕駛員主動監督。

The company has garnered scrutiny from regulators who say that Tesla falsely advertised the capabilities of its Autopilot and report by the National Highway Traffic Safety Administration also found that out of the 956 Tesla crashes the agency reviewed, 467 of those could be linked to Autopilot.

美國國家公路交通安全管理局的報告還發現，在該機構審查的 956 起特斯拉車禍中，有 467 起可能與 Autopilot 有關。

But reaching full autonomy is critical for Tesla, whose sky-high valuation is largely dependent on bringing robo-taxis to market, analysts say.

但分析師表示，實現完全自動駕駛對特斯拉來說至關重要，因為特斯拉的高估值在很大程度上取決於能否將機器人計程車推向市場。

The company reported lackluster results in its latest earnings report and has fallen behind other automakers working on autonomous vehicle technology.

該公司在最新財報中公佈的業績乏善可陳，在自動駕駛汽車技術方面已落後於其他汽車製造商。

These include Alphabet-owned Waymo, which is already operating fully autonomous vehicles in the U.S. cities, GM's Cruze and Amazon's Zoox.

其中包括 Alphabet 旗下的 Waymo（已在美國城市營運完全自動駕駛汽車）、通用汽車的 Cruze 和亞馬遜的 Zoox。

In China, competitors include Didi and Baidu.

在中國，競爭對手包括滴滴和百度。

特斯拉希望 Dojo 能改變這種狀況。

According to Musk, Dojo has been running tasks for Tesla since 2023.

據馬斯克稱，Dojo 從 2023 年起就開始為特斯拉執行任務。

And since Dojo has a very specific task, to train Tesla's self-driving systems, the company decided that it's best to design its own chip called the D1.

由於 Dojo 有一項非常特殊的任務，即訓練特斯拉的自動駕駛系統，是以公司決定最好自己設計一款名為 D1 的芯片。

This chip is manufactured in 7 nanometer technology.

It packs 50 billion transistors in a miserly 645 millimeter square.

它在 645 毫米見方的吝嗇空間裡裝入了 500 億個半導體。

One thing you'll notice, 100 percent of the area out here is going towards machine learning training and bandwidth.

你會注意到，這裡 100% 的區域都用於機器學習培訓和帶寬。

這是一臺純粹的機器學習機器。

For high-performance computing, it is very common to have supercomputers that have CPUs and GPUs.

在高性能計算方面，擁有 CPU 和 GPU 的超級計算機非常常見。

However, increasingly, AI supercomputers also contain specialized chips that are specially designed for AI workloads.

不過，越來越多的人工智能超級計算機還包含專門為人工智能工作負載設計的專用芯片。

One of the key things that came through when I was looking at D1 is latency.

我在研究 D1 時發現的一個關鍵問題是延遲。

It's training on a video feed that's coming from cameras in cars.

這是通過汽車上的攝像頭傳輸的視頻進行訓練。

So the big thing is kind of how do you move those big files around and how do you handle for latency?

是以，最重要的是如何移動這些大文件，以及如何處理延遲問題？

Aside from the D1, which is being manufactured by Taiwanese chip maker TSMC, Tesla is also designing the entire infrastructure of its supercomputer from the ground up.

除了由臺灣芯片製造商臺積電製造的 D1 外，特斯拉還從頭開始設計其超級計算機的整個基礎設施。

Designing a custom supercomputer gives them the opportunity to optimize the entire stack, right?

設計一臺定製超級計算機讓他們有機會優化整個堆棧，不是嗎？

Go from the algorithms and the hardware, make sure that they're perfectly in concert with each other.

從算法和硬件入手，確保它們相互完美配合。

It's not just Tesla, but if you see a lot of the hyperscalers, the Googles of the world, the Metas, the Microsofts, the Amazons, they all have their own custom chips and systems designed for AI.

不僅是特斯拉，如果你看到許多超級計算機、世界上的谷歌、美塔斯、微軟、亞馬遜，它們都有自己為人工智能設計的定製芯片和系統。

In the case of Dojo, the design looks something like this. 25 D1 chips make up what Tesla calls a training tile, with each tile containing its own hardware for cooling, data transfer and power and acting as a self-sufficient computer.

Dojo 的設計是這樣的。25 個 D1 芯片組成了特斯拉所謂的 "訓練瓦"，每個 "瓦 "都有自己的冷卻、數據傳輸和電源硬件，是一臺自給自足的計算機。

Six tiles make up a tray and two trays make up a cabinet.

六個瓷磚組成一個托盤，兩個托盤組成一個櫃子。

Finally, 10 cabinets make up an exopod, which Tesla says is capable of 1.1 exaflops of compute.

最後，10 個機櫃組成一個 exopod，特斯拉稱其可實現 1.1 exaflops 的計算能力。

To put that into context, one exaflop is equal to one quintillion calculations per second.

換句話說，1 個 exaflop 相當於每秒計算 1 萬億次。

This means that if each person on the planet completed one calculation per second, it would still take over four years to do what an exascale computer can do in one second.

這意味著，如果地球上的每個人每秒完成一次計算，仍需要四年多的時間才能完成超大規模計算機在一秒鐘內所能完成的計算。

But there are other supercomputers certainly that are performing that ballpark as well.

不過，其他超級計算機的性能肯定也在這個範圍內。

One of those supercomputers is located at the Department of Energy's Oak Ridge National Laboratory in Tennessee.

其中一臺超級計算機位於田納西州能源部橡樹嶺國家實驗室。

The system, called Frontier, operates at 1.2 exaflops and has a theoretical peak performance of 2 exaflops.

該系統名為 Frontier，運行速度為 1.2 exaflops，理論峰值性能為 2 exaflops。

The supercomputer is being used to simulate proteins to help develop new drugs, model turbulence to improve the engine designs of airplanes and create large language models.

這臺超級計算機被用於模擬蛋白質以幫助開發新藥物，建立湍流模型以改進飛機引擎設計，以及創建大型語言模型。

The next generation of zetascale supercomputers are already in development.

下一代 zetascale 超級計算機已經在開發中。

A zetaflop supercomputer has a computing capability equal to 1,000 exaflops.

zetaflop 超級計算機的計算能力相當於 1,000 exaflops。

As for Dojo, Dickens says its utility could go beyond turning Teslas into autonomous vehicles.

至於 Dojo，Dickens 說它的用途可能不僅僅是把特斯拉變成自動駕駛汽車。

If you wanted to train a robot on how to dig a hole, how many Tesla cars have driven past somebody digging a hole on the side of the road?

如果你想訓練機器人如何挖洞，有多少輛特斯拉汽車駛過路邊挖洞的人身邊？

And could you then point that at Optimus and say, hey, I've got hundreds of hours of how people dig holes.

然後你能不能指著擎天柱說，嘿，我有幾百個小時記錄人們是怎麼挖洞的。

I want to train you as a robot on how to dig holes.

我想把你訓練成一個會挖洞的機器人。

So I think you've got to think wider of Tesla than just a car company.

是以，我認為，特斯拉不僅僅是一家汽車公司，你必須把它想得更寬廣。

字幕列表影片播放

埃隆-馬斯克為何豪賭超級計算機，以推動特斯拉和 xAI 的發展？ (Why Elon Musk Is Betting Big On Supercomputers To Boost Tesla And xAI)

massive

improve

infrastructure

capacity