Last week, NVIDIA released its earnings report for the first quarter of 2024.

上週，英偉達發佈了 2024 年第一季度財報。

Once again, America's third-largest company massively outperformed expectations, with quarterly revenue soaring by 262% year-on-year, driven by massive demand for both its current-generation Hopper GPUs and its new Blackwell chips.

在當前一代 Hopper GPU 和新型 Blackwell 芯片的巨大需求推動下，美國第三大公司的業績再次大幅超出預期，季度收入同比飆升 262%。

Unsurprisingly, these results sent NVIDIA's stock soaring, and as of Thursday, shares are now trading about 20% higher than they were pre-earnings.

不出所料，這些業績使英偉達的股價一路飆升，截至本週四，英偉達的股價已比盈利前高出約 20%。

All in all, this means NVIDIA's stock has doubled since the beginning of this year, and multiplied tenfold since October 2022.

總而言之，這意味著英偉達的股價自今年年初以來已經翻了一番，自 2022 年 10 月以來更是翻了十倍。

Now, while NVIDIA's numbers are clearly great, the dizzying rate at which its stock has climbed, accompanied by similar claims among other AI highs, might in fact be a bit of a bubble.

現在，雖然英偉達的數據顯然非常出色，但其股價以令人眼花繚亂的速度攀升，以及其他人工智能公司的類似說法，實際上可能有點保麗龍。

So, in this video, we're going to have a look at why some people think the AI boom is really a bubble, why other people disagree, and why both sides could be sort of right at the same time.

是以，在這段視頻中，我們將看看為什麼有些人認為人工智能熱潮實際上是保麗龍，為什麼其他人不同意，以及為什麼雙方可能同時都是對的。

Before we start, if you haven't already, please consider subscribing and ringing the bell to stay in the loop and be notified when we release new videos.

在我們開始之前，如果您還沒有訂閱，請考慮訂閱並按鈴，以便在我們發佈新視頻時及時收到通知。

Now, before we get into it, the first thing to say is that this is not investment advice, and more generally, don't take investment advice from YouTubers.

現在，在我們開始討論之前，首先要說的是，這不是投資建議，更籠統地說，不要從 YouTubers 那裡獲得投資建議。

So, caveats aside, let's dive in.

所以，拋開這些注意事項，讓我們開始吧。

Is the AI boom actually a bubble?

人工智能熱潮實際上是一個泡沫嗎？

Well, the honest answer is that no one knows.

老實說，沒人知道。

So, we're going to present both sides of the argument, and you can make your mind up for yourself.

所以，我們將介紹爭論的正反兩方面，你可以自己做出決定。

Broadly speaking, there were two types of arguments on either side.

概括地說，雙方都有兩類論點。

There's a technological argument over whether or not AI is actually the revolutionary technology that its advocates claim it to be, and there's a more financial argument about whether or not the stock market is looking bubbly at the moment.

關於人工智能是否真的像其擁護者所說的那樣是一項革命性的技術，存在著技術上的爭論；而關於股市目前是否看起來有保麗龍，則存在著金融上的爭論。

So, let's start by looking at the arguments against AI being a bubble.

那麼，讓我們先來看看反對人工智能是保麗龍的理由。

The main arguments made on this side is a technological one, that generative AI is a genuinely unprecedented technology that merits the enormous investments in valuations that we've over the past couple of years.

這一方的主要論點是技術性的，認為人工智能是一項真正前所未有的技術，值得我們在過去幾年裡對其估值進行鉅額投資。

ChatGBT, for instance, was the fastest spreading tech platform in history, with an estimated 100 million monthly users only two months after launch.

例如，ChatGBT 是歷史上傳播速度最快的科技平臺，推出僅兩個月後，月用戶數量就達到了 1 億。

And AI's most bullish advocates argue that its ability to perform language-based cognitive tasks could transform our economy.

最看好人工智能的人認為，人工智能執行基於語言的認知任務的能力可以改變我們的經濟。

Goldman Sachs, for instance, expects AI to essentially double the rate of US productivity growth, while McKinsey reckons that AI and associated automation technologies could increase global GDP growth by an upper limit of 3.4% over the coming decade, more than doubling the global GDP growth rate.

例如，高盛預計，人工智能將使美國的生產率增長率基本上翻一番，而麥肯錫則認為，人工智能和相關自動化技術可在未來十年將全球 GDP 增長率的上限提高 3.4%，使全球 GDP 增長率翻一番以上。

The most bullish AI advocates, like these guys, expect AI to increase global GDP growth by more than 5%, which, assuming a baseline of about 3%, would imply a doubling of global GDP in the next 10 years.

最看好人工智能的人，比如這些人，預計人工智能將使全球 GDP 增長率提高 5%以上，假設基線為 3%左右，這意味著未來 10 年全球 GDP 將翻一番。

If these numbers are even close to the mark, then the extravagant valuations we've seen recently are easily justified.

如果這些數字與實際情況相差無幾，那麼我們最近看到的奢侈估值也就不難理解了。

The second argument is that we haven't seen a proliferation of so-called bullshit companies that you usually see during bubbles.

第二個論點是，我們還沒有看到保麗龍時期通常會出現的所謂 "狗屁公司 "激增的現象。

In the dot-com bubble, for instance, there were a whole load of companies that were basically just domain names rushing into IPOs and achieving eye-watering valuations.

例如，在網絡保麗龍時期，有一大批基本上只是域名的公司匆忙上市，並獲得了令人瞠目的估值。

Similarly, in the crypto bubble a couple of years ago, there were a whole load of clearly bullshit meme coins that had no real value apart from their ridiculous names.

同樣，在幾年前的加密貨幣保麗龍中，也出現了一大堆明顯是胡說八道的主題幣，它們除了可笑的名字外沒有任何實際價值。

Rather, most of the AI-related gains have been captured by the so-called Magnificent Seven, so Tesla, Meta, Alphabet, Apple, Amazon, Nvidia, and Microsoft, none of which could really be described as bullshit companies.

相反，與人工智能相關的大部分收益都被所謂的 "七巨頭 "所攫取，即特斯拉、Meta、Alphabet、蘋果、亞馬遜、Nvidia 和微軟。

The third argument is that even though the Magnificent Seven have seen steep rises in the market cap recently, they're not drastically overvalued.

第三個論點是，儘管 "華麗七俠 "最近的市值急劇上升，但它們的估值並沒有被大幅高估。

This is in part because some of their recent growth is, in a sense, a recovery from the in late 2022 and their price-to-earnings ratios haven't changed drastically in the last couple of years.

這部分是因為，從某種意義上說，它們最近的一些增長是從 2022 年末開始恢復的，而且它們的市盈率在過去幾年裡沒有發生大的變化。

A useful comparison can be made between Nvidia today and Cisco during the dot-com bubble.

我們可以將今天的 Nvidia 與互聯網保麗龍時期的思科公司做一個有益的比較。

During the 90s, Cisco was most focused on building routers and other internet hardware, in much the same way that Nvidia today is more about building the infrastructure behind the tech rather than the tech itself.

上世紀 90 年代，思科最專注於路由器和其他互聯網硬件的製造，這與今天的 Nvidia 更專注於製造技術背後的基礎設施而非技術本身的做法如出一轍。

But at the peak of the dot-com bubble, Cisco's two-year forward price-to-earnings ratio, that is, the ratio between its current market cap versus its forecast earnings in two years time, was over 100.

但在網絡保麗龍的頂峰時期，思科的兩年遠期市盈率（即當前市值與兩年後預測收益的比率）超過了 100。

Nvidia's two-year PE ratio, however, is only 27, high by normal standards but not absurdly so.

不過，Nvidia 的兩年市盈率僅為 27，按正常標準來看雖高，但並不離譜。

Alright, so those are the arguments against AI being a bubble.

好了，以上就是反對人工智能是泡沫的理由。

Let's take a look at the other side.

讓我們看看另一面。

The first argument is a technological one, that AI isn't all it's cracked up to be.

第一個論點是技術性的，即人工智能並不完全是想象中的那樣。

Proponents of this argument might point to AI's apparent plateau, the lack of obvious use cases, or like in AI, to previous innovations that just haven't come through.

這種觀點的支持者可能會指出，人工智能顯然已經到了高原期，缺乏明顯的使用案例，或者像人工智能一樣，以前的創新只是沒有實現而已。

But the second and perhaps more popular argument is that even if in the aggregate the Magnificent 7's PE ratio hasn't changed all that much, it's still gone up a bit and certain companies have seen bubble-esque increases.

但第二種可能更受歡迎的說法是，即使 "華麗 7 號 "的市盈率總體上沒有太大變化，但還是有所上升，某些公司的市盈率出現了泡沫式增長。

For instance, the combined market value of Alphabet, Amazon and Microsoft has jumped by $3 trillion during the AI boom, 150 times the $20 billion in revenue they expect to come from generative AI.

例如，在人工智能熱潮中，Alphabet、亞馬遜和微軟的總市值躍升了 3 萬億美元，是它們預計從生成式人工智能中獲得的 200 億美元收入的 150 倍。

Even Nvidia looks a bit bubbly.

就連 Nvidia 看起來也有點氣泡。

Its current PE ratio is nearly 70, more than double the other Magnificent 7 stocks, and while it justifies this with reference to future sales, this all assumes that Nvidia continues to dominate the chips market for the next couple of years and that demand for chips holds up.

它目前的市盈率接近 70，是其他 "華麗 7 號 "股票的兩倍多，雖然它以未來的銷售額為理由，但這一切都假定 Nvidia 在未來幾年繼續主導芯片市場，而且芯片需求保持穩定。

Which is possible, but not inevitable.

這是可能的，但並非不可避免。

The final argument is that even if we haven't seen that many bullshit companies, AI is clearly a bit of a corporate fad at the moment, as evidenced by the exponential rise in earnings call mentioning AI.

最後一個論點是，即使我們還沒有看到那麼多胡說八道的公司，但人工智能目前顯然是企業的一種時尚，提到人工智能的財報電話呈指數級增長就是證明。

And while the Magnificent 7 have captured most of the AI boom, AI startups are increasingly popular with venture capital, despite an otherwise difficult investment climate.

雖然 "華麗七巨頭 "佔據了人工智能熱潮的大部分份額，但人工智能初創企業也越來越受到風險投資的青睞，儘管投資環境並不樂觀。

Ultimately, no one knows, and we'll just have to wait to see if or when the bubble pops.

歸根結底，沒有人知道，我們只能靜觀保麗龍是否或何時破裂。

But perhaps it's possible that both sides are sort of right.

但也許雙方都有可能是對的。

Some bubbles, like the NFT bubble a couple of years ago, or the original tulip bubble in the 1630s, are wholly speculative.

有些保麗龍，如幾年前的 NFT 保麗龍，或 16 世紀 30 年代最初的鬱金香保麗龍，完全是投機性的。

There's no justifying the price, and people are only buying because they think they'll be able to sell on to someone else at a higher price.

價格沒有合理性可言，人們之所以購買，只是因為他們認為能夠以更高的價格賣給別人。

But some bubbles are based on a genuinely revolutionary technology.

但有些氣泡是基於真正革命性的技術。

It's just that market enthusiasm outruns the industry.

只是市場的熱情超過了行業的熱情。

The fibre optic bubble in the 80s and the dotcom bubble in the 90s, for instance, were predicated on the revolutionary potential of the internet, which turned out to be correct.

例如，上世紀 80 年代的光纖保麗龍和 90 年代的網絡保麗龍，都是以互聯網的革命性潛力為前提的，而事實證明這是對的。

It's just that everyone got a bit too excited a bit too early.

只是大家都有點興奮得太早了。

It's possible that AI will end up being the same sort of thing.

人工智能最終也有可能成為同樣的東西。

A revolutionary technology that just takes a bit longer to on tons of data analysis to interpret crucial economic data and understand political decisions.

一項革命性的技術，只是需要更長的時間來進行大量的數據分析，以解讀關鍵的經濟數據和理解政治決策。

But this kind of analysis isn't just crucial for our jobs.

但這種分析不僅對我們的工作至關重要。

As our world becomes more driven by AI and data, your job will likely require more analytical skills too.

隨著我們的世界越來越受人工智能和數據的驅動，你的工作也可能需要更多的分析技能。

That's why we've been using Brilliant.

這就是我們一直使用 Brilliant 的原因。

Brilliant is the STEM learning platform which can teach you all the skills you'll need in an increasingly digital workplace.

Brilliant 是一個 STEM 學習平臺，可以教給你在日益數字化的工作環境中所需的所有技能。

They've just introduced a whole collection of new programming and LLM courses, suitable for learners at any level, taking you through everything from learning Python to understanding the importance of training data.

他們剛剛推出了一整套新的編程和法學碩士課程，適合任何水準的學習者，從學習 Python 到了解培訓數據的重要性，應有盡有。

And Brilliant isn't just about memorization and lectures.

而 Brilliant 不僅僅是背誦和演講。

Brilliant teaches you by doing, using active learning techniques to teach you the principles behind otherwise quite complex topics, and ensuring that you actually understand what's going on with thousands of interactive lessons in maths, data analysis, programming, and AI.

Brilliant 採用主動學習技術，通過數學、數據分析、編程和人工智能方面的數千節互動課程，向你傳授原本相當複雜的課題背後的原理，確保你真正理解其中的內容。

All of Brilliant's courses are split into easy to digest chunks, meaning you can quickly and easily gain real knowledge in just a few minutes a day, with fun lessons you can do whenever you have time, replacing mindless scrolling on the bus or waiting for coffee with something stimulating and rewarding.

Brilliant 的所有課程都抽成易於消化的小塊，這意味著您每天只需花幾分鐘就能快速、輕鬆地獲得真正的知識，只要您有時間，就可以學習有趣的課程，在公車上或等咖啡的時候，用一些有刺激性和有益的東西來取代無意識的滾動。

To gain access to Brilliant for free for a full 30 days, head to brilliant.org forward slash tldr or click the link in the annual premium subscription and they'll know that you came from us, which really helps us out.

要免費訪問輝煌網站整整 30 天，請訪問 brilliant.org forward slash tldr 或點擊年度高級訂閱中的鏈接，他們就會知道您是從我們這裡來的，這對我們大有幫助。

Thanks to Brilliant for sponsoring this video and thank you for your support.

感謝 Brilliant 贊助本視頻，感謝您的支持。