AI Bubble

Source:

https://hotwhitesnow.wordpress.com/2024/04/18/history-doesnt-repeat-itself-but-it-often-rhymes/

你也許聽過這句話：人類不會吸取教訓，歷史總是不斷重複。

美國文豪馬克吐溫 (Mark Twain, 1835-1910) 的說法：歷史不會重複，但是會押韻（History doesn’t repeat itself but it often rhymes）。他筆下有更細緻的寫法：歷史不會重複，而是像萬花筒，我們今日見到的圖像，是由過去的圖像的碎片所重組而成。（原文：History never repeats itself, but the Kaleidoscopic combinations of the pictured present often seem to be constructed out of the broken fragments of antique legends. Source: The Gilded Age: A Tale of Today） 

換言之，今日和過去有相似之處，但是也有不同之處。經歷過 2000 年 3 月科網股泡沫 (dot com bubble) 爆破的中年人看 AI 概念股 (Magnificent 7 Stocks)，也有相同的感覺。有些地方似曾相識，但是有些地方不同。

先講相似之處。

什麼是泡沫？就是物超所值，股價沒有盈利支持。股價抄上天高，散戶跟風入市，原因是害怕錯失機會（提示：Fear of Missing Out, FOMO + 養龍蝦），然後泡沫爆破，股價插水或破產收場，很多人損手。這種戲碼，每隔一段日子就會在股票市場上演。金融市場需要的，是炒作的藉口，時真時假，半真半假，也可以 100% 假，只要相信的傻瓜數目夠多，幕後玩家就可以撈一把，然後圈錢離場。是信者得救還是失救，時間會說明一切。對資深股民來說，這種玩法熟悉不過。 

分別在於：今次的 AI 泡沫是一種 Capital-intensive 的玩法，為了建設資料中心 (Data Centre) 以及購買晶片（提示：美國制裁中國的手段），AI 概念股（Magnificent 7 Stocks）債台高築，要運用創意財技（例如：連串的複雜交易或服務協議）把資金在某個圈子內傳來傳去（提示：Circular Financing），目的是托高估值 (Valuation)，才能繼續籌錢玩下去。這種玩法似音樂椅 (Music Chair) ，萬一音樂結束（資金鏈斷裂），泡沫就會爆破。英語傳媒的說法：AI 概念股（Magnificent 7 Stocks）= House of cards，即是：根基不穩，搖搖欲墜。

英倫銀行 (Bank of England) 去年10月底發表的一份研究報告指出：AI 概念令美股的估值重回 1999 年 dot com bubble 的高位，發展 AI 所需要的基礎建設 (Infrastructure) 例如數據中心 (Data Centre) 主要透過債務融資，但是 projected earnings growth of many AI-impacted companies is uncertain。換言之，泡沫已經形成，爆破風險高。欲知詳情，請參考《延伸閱讀》部份提供的資料。

網上也有外國人寫文章，重提股民熟悉的財經典故：被暗殺的美國總統 JFK 的父親（Joseph Kennedy Sr.）年輕時做過股票經紀，當他聽見擦鞋童跟他講最喜歡的股票，就知道股市已經偏高，是時候沽貨離場。時間是 1929 年，他沽出股票之後，美股大跌，實體經濟進入大蕭條（The Great Depression）時期。在歐洲，經濟大蕭條導致納粹主義興起，然後希特拉上台，發動戰爭，之後的事大家都懂。

也許我們應該把 AI 和 AI 概念股分開討論，事情會比較清楚明白。

AI 能否提升企業的生產力，最後取代那些工種，如何改變工作流程或商業模式，仍然是未知之數，企業和個人仍在摸索中。長遠來說，那是學院派的研究課題。

短期內，中文傳媒（舊媒體）繼續用 AI 販賣恐懼和製造焦慮，令打工仔、大學生和中產家長非常不安，然後向讀者或觀眾推銷收費的 AI 訓練班。在戰雲密佈的日子製造焦慮，只為討好所餘無幾的廣告客戶，舊媒體真是「功德無量」。

網上的 YouTuber 博流量，去得更盡：資本主義國家如何解決 AI 所導致的失業人口？答案是把失業者送上戰場！問題是：今日的戰爭模式已經改變。證據：烏克蘭用無人機轟炸俄羅斯的石油生產設備。換言之，人海戰術已經不管用了。

網民的留言：十九世紀的工業革命提升了企業的生產力，但是也導致共產主義的誕生，今次 AI 誕生又會搞出什麼東西？是加強資本主義（的剝削本質）還是迫使共產主義轉型（因為部份工種會被 AI 取代）？天曉得。

扯遠了，說回來。就算在可見未來，AI 真的帶來實質轉變，而那些轉變是普通人也感受得到，得益者也未必是今日的 AI 概念股（Magnificent 7 Stocks）。參考 2000 年 3 月科網股泡沫爆破的經驗，部份的 AI 概念股 (Magnificent 7 Stocks) 會止蝕離場甚至破產，只有極少數的 AI 概念股能夠把新科技轉化為盈利，得以持續發展，最後站穩腳跟，成為行業的龍頭。對，真相很殘酷。原因？

經濟學家 Paul Krugman的說法：amazing technology doesn't necessarily mean amazing profits.

South Park（港譯：衰仔樂園，又稱：南方四賤客）的黑色幽默：Phase 1: Stealing Underpants. Phase 2: ??? Phase 3: Profit.

巴菲特 (Warren Buffett) 的名句："Rule No.1: Never lose money. Rule No.2: Never forget rule No.1."

如果閣下跟 Auntie 年紀相若，也許還記得 Netscape, Yahoo, Tom.com 或 36.com（提示：TVB 陳國強）。1999-2000 年間，許多人相信互聯網會改變商業模式，也會為人類的生活帶來翻天覆地的改變，於是 Tom.com 的新股認購成為熱潮，股價曾經是認購價的 N 倍。如果你是中年人，想一想 2000 年至今，閣下的工作或生活改變了多少？如果你是生意人，互聯網怎樣改變你做生意的方法？有沒有帶來利潤或新的收入來源？說到這裡，又要借助巴菲特 (Warren Buffett) 的名句： “You never know who's swimming naked until the tide goes out.”（中譯：潮退了才知道誰人沒有穿內褲。）如果那條內褲是偷來的（提示：South Park），只不過是跟風或抄襲。撈一把就走，是股市的常態，美股和港股都一樣。 

讓我們回歸實體經濟。

香港的情況，是不少行業都因循守舊，被舊的商業模式或工作流程所困。在保守派掌權的組織推動 AI，會觸動很多人的飯碗和利益。在香港，玩弄程序的人多（尤其是政府部門以及公營機構），改變流程（否則難以應用 AI）等於失去自我保護的能力以及交出玩弄弱勢的武器，對既得利益者來說，是百般不情願的事情，他或她只會做點門面功夫當交差，應酬一下上司或老闆或官府。經常聽到的港式官腔：流程順暢=取得成功。你要公務員改變流程，等於取他或她的性命。由此觀之，香港的官商機構能否應用 AI，以及應用到那個程度，難以樂觀。而就算應用 AI 的確帶來好處，那些好處是否會轉化為金錢，中間需要多少時間，過程是怎樣，那些錢最後又落入誰人的口袋裡，是否會被轉走（而不是投資於這個城市的未來），又是另一堆問題，暫且不論。華人社會，政治殺死經濟，是常見的事。

結論：互聯網和AI 都是建基於電腦的東西，從玄學的角度看，屬火，美國的 AI 概念股（Magnificent 7 Stocks）成為炒作藉口，產生虛火，很正常。對，Auntie 扮神棍，我的志願是成為「龍婆」（提示：羅蘭）的接班人。以上所寫，信不信由你。萬一閣下自信爆棚，相信短炒可以致富，又或者眼光獨到，可以選中行業龍頭，當 Auntie 無講過。Thanks for reading.

What is DeepSeek and why did US tech stocks fall?

Why doubts have been raised about sustainability of US artificial intelligence boom

The Guardian

Jan 27 2025

https://www.theguardian.com/business/2025/jan/27/what-is-deepseek-and-why-did-us-tech-stocks-fall

Excerpt: What is DeepSeek? 

DeepSeek is a Chinese artificial intelligence (AI) company based in Hangzhou that emerged a couple of years ago from a university startup. Its stated goal is to make an artificial general intelligence – a term for a human-level intelligence that no technology firm has yet achieved. It’s not there yet, but this may be one reason why the computer scientists at DeepSeek have taken a different approach to building their AI model, with the result that it appears many times cheaper to operate than its US rivals.

Another reason it appears to have taken the low-cost approach could be the fact that Chinese computer scientists have long had to work around limits to the number of computer chips that are available to them, as result of US government restrictions.

But there are lots of AI models out there from OpenAI, Google, Meta and others. What’s the big deal?

This model uses a different kind of internal architecture that requires less memory use, thereby considerably reducing the computational costs of every search or interaction with the chatbot-style system. It has been praised by researchers for its ability to tackle complex reasoning tasks, particularly in mathematics and coding and it appears to be producing results comparable with rivals for a fraction of the computing power. DeepSeek has said it took two months and less than $6m (£4.8m) to develop the model, although some observers caution this is likely to be an underestimate. Nevertheless it is vastly less than the billions that the Silicon Valley tech companies are spending to develop AIs and is less expensive to operate.

Why did US tech stocks fall?

Hundreds of billions of dollars were wiped off big technology stocks after the news of the DeepSeek chatbot’s performance spread widely over the weekend. The timing was significant as in recent days US tech companies had pledged hundreds of billions of dollars more for investment in AI – much of which will go into building the computing infrastructure and energy sources needed, it was widely thought, to reach the goal of artificial general intelligence. DeepSeek’s performance seems to question, at least, that narrative.

What is the worry for Nvidia?

Nvidia is one of the companies that has gained most from the AI boom. It went from being a maker of graphics cards for video games to being the dominant maker of chips to the voraciously hungry AI industry. It has been compared to a modest trader in pickaxes and buckets in 19th-century California, which happened to be on the spot when the gold rush happened and so it became a massive supplier to the world’s richest industry. Tech companies looking sideways at DeepSeek are likely wondering whether they now need to buy as many of Nvidia’s tools. Its market value fell by $600bn on Monday.

What is DeepSeek not doing?

It hasn’t reached artificial general intelligence, the threshold at which AI starts to reason and which OpenAI and others in Silicon Valley are pursuing. Sam Altman, OpenAI’s chief executive, has cautioned that breakthrough is unlikely to be imminent. But it does seem to be doing what others can at a fraction of the cost.

（推介原因：DeepSeek 的發展模式令人想起中共的口號：多快好省地建設 XXXX。相關概念：Bandit version 山寨版、Pirate copy 海盗版、Reverse Engineering。）

Investopedia

Magnificent 7 Stocks: What You Need To Know

By Cedric Thompson Updated December 12, 2025

https://www.investopedia.com/magnificent-seven-stocks-8402262

Key Points:

• The Magnificent Seven stocks are a group of high-performing and influential companies in the U.S. stock market: Alphabet, Amazon, Apple, Tesla, Meta Platforms, Microsoft, and Nvidia.

• Bank of America analyst Michael Hartnett used the film name in 2023 when commenting on the seven highest-performing tech firms.

• The performance of the Magnificent Seven stocks is driven by technological innovation, market dominance, financial performance, brand equity, research and development, and global economic conditions.

• The FAANG stocks and Magnificent Seven stocks have some key differences.

In the world of finance, the Magnificent Seven are a group of seven high-performing, influential stocks in the technology sector. The name comes from a 1960 John Sturges Western film, “The Magnificent Seven,” （港譯：七俠蕩寇志）depicting a powerful group of seven gunmen.

What Are the Magnificent 7 Stocks? Definition & Key Players

The Magnificent Seven stocks are the most influential companies in the U.S. stock market. This term has been popularized to describe a set of dominant companies, particularly in the tech sector. The group currently includes Alphabet, Amazon, Apple, Tesla, Meta Platforms, Microsoft, and Nvidia, and spans four sectors: technology services, electronic technology, retail trade, and consumer durables. It operates across these industries: internet software/services, telecommunications equipment, internet retail, packaged software, semiconductors, and motor vehicles.

Bank of England

All chips in! Would a fall in AI-related asset valuations have financial stability consequences?

24 October 2025

https://www.bankofengland.co.uk/bank-overground/2025/all-chips-in-ai-related-asset-valuations-financial-stability-consequences

Key Points:

• “AI stocks” comprise a growing share of the market capitalisation of US stock indices. These companies frequently trade at valuation multiples that imply high expected future earnings growth. These two facts have pushed some valuation multiples of US stock indices close to levels seen at the peak of the dot com bubble. 

• While AI could have a transformational economic impact, which might justify these valuations, multiple factors make this outcome uncertain. Additionally, the physical infrastructure which underpins AI model training and inference is expected to require trillions of dollars of investment in the next five years, a significant share financed by debt.

• AI stocks have pushed some US stock valuation metrics to their highest level since the dot com bubble 25 years ago, though these metrics do not fully account for the high projected earnings growth of many AI-impacted companies. Whether these earnings projections will be realised – or even prove underestimates – is uncertain.

 

The Philosophical Economist

AI and The Shoeshine Boy

Erik Angner

Jun 12, 2024

https://erikangner.substack.com/p/ai-and-the-shoeshine-boy

Intro: “If shoe-shine boys are giving stock tips, then it’s time to get out of the market.” In 1929, the story goes, a shoeshine boy offered investment advice to Joe Kennedy. Instead of following the advice, Kennedy went back to the office and sold off his holdings. When everyone was talking about something, Kennedy assumed it was over. I think about this story every time somebody talks to me about the wonders of AI, which is daily. If there were shoeshine boys around, I have little doubt that they’d be talking about AI too.

 

PITZI FINANCIAL

An Ode to the Shoeshine Boy

Justin Gabriel

Feb 17, 2023

https://www.pitzlfinancial.com/blog/ode-shoeshine-boy

Excerpt: In 1929, at the height of an economic boom in America, Joseph Kennedy Sr. (father of JFK) was working as a stockbroker on Wall Street. As the story goes, Joseph was walking around when he decided to sit down for a shoeshine. While polishing his shoes, the young worker gave Joseph some of his favorite stock picks. When Joseph heard the shoeshine boy giving out stock tips, he figured the party was about to end, and it was time to get out of the market. Joseph proceeded to exit his positions in the market and bought short positions that bet on the market going down. 

Shortly after that, the stock market entered a free fall. On Monday, October 28, 1929, the market dropped about 13%. The next day it fell another 12%. These became better known as Black Monday and Black Tuesday, and ushered the United States into The Great Depression. Now did Joseph profit from this type of bet? Absolutely. It's estimated that he made somewhere north of $150 million during that period, which equates to roughly $3.5 billion in today’s dollars. Did he make these bets based on the shoeshine boy? Probably not, but it makes for a good narrative.

Warren Buffett>Quotes

https://www.goodreads.com/author/quotes/756.Warren_Buffett

Quote:“You never know who's swimming naked until the tide goes out.”

MIT Technology Review

ARTIFICAL INTELLIGENCE

The missing step between hype and profit

Coding aside, even the best AI systems struggle to be economically viable in the workplace. What happens then?

By Will Douglas Heaven

Apr 27 2026

https://www.technologyreview.com/2026/04/27/1136456/the-missing-step-between-hype-and-profit?trk=public_post_comment-text

Key Points:

• In the South Park episode “Gnomes,” which first aired in 1998, Kenny, Kyle, Cartman, and Stan discover a community of gnomes that sneak out at night to steal underpants from dressers. Why? The gnomes present their pitch deck. “Phase 1: Collect underpants. Phase 2: ? Phase 3: Profit.”

• The gnomes’ business plan has since become one of the greats among internet memes, used to satirize everything from startup strategies to policy proposals. Right now, it captures the state of AI. Companies have built the tech (Step 1) and promised transformation (Step 3). How they get there is still a big question mark. 

• AI boosters, on the other hand, are convinced that Step 3 is salvation and tend to glaze over the middle bit. They know where they want to go—more or less: It’s hazy up there and still some way off. But everyone’s taking a different route. Will they all make it? Will anyone?

• The tools aren’t just dropped into a cleanroom. They need to work in places contaminated with people and existing workflows. And sometimes adding AI will make things worse. Sure, maybe those workflows need to be torn up and refashioned around the new technology for it to achieve transformative status, but that will take time (and guts).

• That big hole? It’s right where Step 2 should be. The lack of agreement on exactly what’s about to happen—and how—creates an information vacuum that gets filled by the latest wild claim of the week, evidence be damned.

• We need fewer guesses and more evidence. The tech industry (and with it the world’s economy) rests on the held-out promise that AI really will be transformative. But that is not yet a sure bet. Next time you hear bold claims about the future, remember that most businesses are still figuring out what to do with their underpants.

YouTube

Make Profit by Stealing Underpants

SOUTH PARK (1:36 minutes)

https://youtu.be/a5ih_TQWqCA?si=jDKGgnB5C836WTQf

外國人的留言：This is literally most companies think about AI at the moment. Phase 1 has always been “Fire all the staff” and Phase 3 always “Profit” and Phase 4 is “Run for re-election.”

How Circular Deals Are Driving the AI Boom

Bloomberg Originals (10:02 minutes)

2026-01-23

https://www.youtube.com/watch?v=9yy_Wz0BbyU

Intro: The AI boom is everywhere, but much of it is illusory, with money flowing between a few players who have yet to profit. If it is a bubble and it pops, the consequences for everyone may be dire. 

If it turns out that AI is a bubble and it does indeed pop as many are warning, the consequences for everyone may be dire. Bloomberg Originals explains how circular deals between AI players figure into what’s being called the “wager to end them all.”

Is AI the next dot-com crash? | Business Beyond (19:42 minutes)

DW News

2026-04-25

https://www.youtube.com/watch?v=c5tdpOVrtdA

Intro: The AI industry is burning through trillions in data center investments before turning a real profit and the warning signs of a massive financial bubble hard to ignore. We checked every box of what historically causes a tech bubbles, from dot-com crash patterns to debt-fuelled infrastructure spending, and AI ticks all of them. One single breakthrough could be all it takes to bring the whole house of cards down. 

Key Points:

• Is AI the Next Big Bubble?

• Integrating AI Into Business Workflows

• The 4 Historical Factors That Create Tech Bubbles

• AI Reliability Crisis: Hallucinations, Slop & The Five Nines

• Debt-Fueled AI Boom: Big Tech Bets Big on Data Centers

• Are We in an AI Bubble?

Wikipedia - AI bubble

https://en.wikipedia.org/wiki/AI_bubble

Have We Been Partying Like It’s 1999?

This bubble may end with a tech-bro bailout

Paul Krugman

Feb 05, 2025

https://paulkrugman.substack.com/p/have-we-been-partying-like-its-1999

Key Points:

• One of the earliest columns I wrote for the New York Times was titled “The Ponzi Paradigm.” It mainly summarized Robert Shiller’s book “Irrational Exuberance,” which argued that bubbles, in which asset prices lose touch with reality, are common, and deeply rooted in human psychology.

• Both Shiller and I were, of course, thinking about the huge runup in stock prices, especially tech stocks, during the late 1990s. And in retrospect both of us had amazing timing (which in my case was sheer luck.) My column was published on March 12, 2000, almost perfectly coinciding with the tech-heavy Nasdaq’s peak:

• And it’s impossible not to wonder how much the current situation, with soaring valuations for a handful of technology stocks, resembles where we were 25 years ago, with the frenzy over AI now playing a role similar to that of the frenzy over the internet back then. I believe that there are strong similarities, but also some important and disturbing differences.

• Let’s talk first about tech bubble 1.0 in the light of history. At the time, people were extremely excited about the possibilities created by the internet and IT in general, and they weren’t wrong.

• True, the surge didn’t last. Back in 2000 the economist Robert Gordon argued that the IT revolution was far less significant than what he called the Second Industrial Revolution of the late 19th century, built around electricity, internal combustion engines, chemicals and — last but not least — indoor plumbing. (The great postwar boom was arguably about taking full advantage of this revolution.) So far the data have supported his skepticism.

• Still, the tech surge was real, and brought real benefits to the economy, adding perhaps 10 percent to real GDP. What it didn’t do was justify the high valuations temporarily placed on tech stocks.

• So why did tech stocks rise so high before crashing? Shiller argued that asset bubbles can act like natural Ponzi schemes: those who get in early make money, not from underlying asset returns, but from rising prices driven by later entrants; as people see the big payoffs others are receiving, they pile in too, driving the asset price even higher. Skeptics start to look foolish; eventually some of them join the party, driven by FOMO — fear of missing out. And then you run out of greater fools, the music stops, and investors are left with a terrible hangover.

• It’s impossible to read about the internet-based frenzy of the late 1990s without drawing parallels with the AI-based frenzy we’re experiencing now. In fact, even the numbers look similar. In 1999 the price-earnings ratio for the S&P 500 hit 33, which looked, and was, crazy. As I work on this post, the PE ratio is about 30.

• Now, like the internet — but unlike crypto, which still seems to have no use cases beyond money-laundering — AI clearly has significant real-world applications.

• As far as I can tell, it’s a reasonable guess that AI’s economic impact will look like that of the IT/internet boom of 1995-2005: a significant bump in productivity but not an enduring transformation of the growth rate. But that’s only a guess.

• But what about stock prices? To the extent that stock values in the 1990s reflected more than a natural Ponzi scheme, they reflected the belief that some of the tech players would eventually establish themselves as highly profitable quasi-monopolies along the lines of Microsoft, which to this day holds an entrenched position based on network externalities: everyone uses their products because everyone else uses their products.

• What’s different this time is that AI fever is concentrated on a handful of companies — the Magnificent 7 — most of which are already entrenched quasi-monopolies. I don’t know whether people realize how anomalous this is. Historically, major new technologies have tended to disrupt the existing market hierarchy; this time, investors are in effect expecting radical new technology to reinforce that hierarchy.

• And I guess I don’t understand why anyone expects AI to make highly profitable quasi-monopolies even more profitable. How much bigger can the market for Office or Google search get? I understand that these companies feel the need to invest in AI for defensive purposes, to fend off potential competitors. But this need should if anything make them less rather than more profitable.

• In any case, the concentrated nature of AI frenzy goes along with another big difference from the 1990s: political power. In the 1990s Silicon Valley types tended to consider government irrelevant to their libertarian techno-utopian dreams. These days Big Tech invests heavily in political influence, and the biggest players have either made their peace with or become active promoters of authoritarianism.

• All of which suggests that while AI fever bears a lot of resemblance to the dot com bubble, the end game may be quite different. Look at Donald Trump’s two big tech-related proposals — a strategic cryptocurrency reserve and a $500 billion AI infrastructure investment. From the point of view of the national interest, the first makes no sense at all; the second might conceivably have merit, but sits oddly with the Trump administration’s attempts to unravel all the Biden administration’s efforts on behalf of strategic technology, let alone green energy.

• But suppose that we think of these proposals not in terms of the national interest, but in terms of who would get the money. Who holds a lot of cryptocurrency that the government would buy up? Which companies do you think would receive the lion’s share of subsidies for AI?

• There are some clear similarities between the 90s tech bubble and recent AI fever. But this bubble may end, not with a pop, but with a giant tech-bro bailout.

（推介原因：經濟學家 Paul Krugman 分析目前的 AI 熱潮跟 2000 年科網股泡沫的異同，他的預測是以 Bailout 告終。）

Why AI Spending Reminds Jim Chanos of the Fracking Bubble

Some of what I learned from the market veteran

Paul Krugman

Feb 25 2025

https://paulkrugman.substack.com/p/why-ai-spending-reminds-jim-chanos

Excerpt: Last week I had a video conversation with Jim Chanos, the famous but now retired short-seller. Chanos is a very interesting guy, with an unusually strong sense of history, and of course a very different perspective from those of us who talk about markets but aren’t players. For now, the full interview and transcript are for paid subscribers only. But for today’s post I thought I would intersperse my own thoughts (and confusions) with some key extracts from Jim’s remarks. This is only part of what we talked about, and I’ll do another Chanos-based post soon. For now, however, two topics: Market complacency and the economics of AI. 

AI and the fracking analogy

Another topic we discussed was AI, which has played a huge role in driving up the valuations of a handful of technology companies. When I wrote about it recently, I made comparisons with the internet boom of the late 1990s, built around a technology that produced real economic gains, although not as large as some visionaries predicted. The thing about that boom was that it raised productivity but also cost some investors a lot of money, because they believed wrongly that the trendy companies of the moment would be able to establish long-term, highly profitable market dominance a la Microsoft.

Chanos wasn’t much into that analogy; he thinks this bubble may be worse, and made an analogy I really didn’t see coming:

Forget that. How about the capital being employed? There better be something new. I mean, we're talking now for the just a top handful of companies doing $300 to $500 billion in capex [capital expenditures] annually. I mean, AI isn’t like the internet, which made things more capital efficient and raised returns on capital. So far, AI is doing the opposite. It is a massively capital-intensive business. Someone joked that the top tech companies are now looking like the oil frackers did in 2014, 2015, where more and more capital is chasing arguably a variable return.

Translation: these days tech companies are spending hundreds of billions of dollars a year on equipment and buildings (the capex he’s talking about), so it’s not like the internet boom, which didn’t involve large-scale spending. And he’s doubtful about whether future returns will justify the current levels of AI spending.

Chanos pointed to the huge capital spending that big tech companies are now making on AI: The numbers are now getting so large from just even a couple years ago that the returns on invested capital are really now beginning to turn down pretty hard for these companies.

I've been a bear on the data centers, the old data center companies, because now the new guys are building bigger and better and faster ones and the old ones are obsolete. But the problem is that it's not so much the data centers that depreciate, they do because of the air conditioning and all the guts of them. It's the chips that you're paying $50,000 a piece for that are being leapfrogged by the same company. And so the question is how fast are you depreciating and are we gonna get into the realm of accounting chicanery? How fast are you depreciating these hundreds of billions of dollars if you have to keep re-upping newer and more expensive chips? So, you know, that's where the rubber hits the road and the numbers are getting big enough, that in a couple of years, those are gonna be uncomfortable questions.

Interestingly, yesterday’s news about Microsoft may very well confirm Chanos’ assessment: Microsoft is canceling some leases on data center capacity, possibly indicating that the company believes that it has been securing more AI capacity than it needs. Perchance, did Microsoft listen to my Saturday interview with Chanos?

（推介原因：經濟學家 Paul Krugman 跟沽空高手 Jim Chanos 討論 AI，後者認為 AI bubble 的殺傷力比 dot com bubble 更大。）

How Should We Think About the Economics of AI?

A conversation with Erik Brynjolfsson

Paul Krugman

Mar 22, 2025

https://paulkrugman.substack.com/p/how-should-we-think-about-the-economics

Excerpt:

Paul Krugman: I was about to say, as economists, we don't know anything about business, but you actually kind of got a foot in that world as well. Still, one big question we have here is, obviously, we've got something that is really, really impressive and novel and has startled a lot of people with how well it works. There are corporate valuations that are kind of based on how amazing the technology is. And one thing I thought we learned from the internet boom was that amazing technology doesn't necessarily mean amazing profits.

Erik Brynjolfsson: That's right. Well, let me underscore both parts of that. I do think that these technologies are really amazing and transformative, and they're probably going to create trillions of dollars worth of value. But much of that value, if we have a well-functioning market, will end up in the hands of consumers, not producers. Bill Nordhaus estimated that about 95% of the gains from innovations ultimately go to consumers, not to the people who create it. So there's no guarantee that any of these guys who invent the best new system are going to be able to cash in on it fully.

I think the internet's not a terrible analogy. You know, we saw lots of gains from that. But there was Pets.com which was a total joke, and there was Amazon that did pretty well, and there are lots of companies in between. I think this wave is going to be significantly bigger, both in terms of the consumer benefits and also in terms of the winners and losers. And lots of smart people are making bets that they're the company that's going to win. I think if you really do believe that the technology is as transformative as I do and as a lot of other people do, it's not implausible that some of these companies could be worth a trillion dollars or more, but probably not all of them. So, how to figure out which ones are on the winning side and which ones aren't? Well, we'll leave it to the market and others to sort that out over the next few years.

Paul Krugman: Yeah, I mean, back during the internet boom, the dot com and all of that, we used to say that investors seem to be betting on 10 or 15 different companies, that each one of them was going to be the next Microsoft and acquire this sort of spot, and they couldn't all do it.

（推介原因：經濟學家 Paul Krugman 跟研究資訊科技的學者 Erik Brynjolfsson 討論 AI。）

Technology Bubbles: Causes and Consequences

What history can teach us about AI frenzy

Paul Krugman

Oct 12, 2025

https://paulkrugman.substack.com/p/technology-bubbles-causes-and-consequences

Key Points: 

• It was a time of enormous optimism, based mainly on one specific technology that promised to have transformative effects on the economy. Businesses developing and implementing the new technology spent huge sums on construction and equipment. Individual investors piled into these businesses’ stocks, sending their prices soaring.

• And enthusiasts were right about the technology’s potential. It would eventually transform the economy, indeed society as a whole. But long-term transformation doesn’t necessarily translate into profits for businesses at the cutting edge. As it became clear that the financial returns wouldn’t live up to the hype, some companies went bust, while stocks tied to the technology lost most of their value. And plunging capital spending pushed the economy into a nasty recession.

• Whatever it is, few are denying that the technology is impressive. But warnings that there may be a huge AI bubble are getting louder. Worries about the financial underpinnings of all that capital spending are growing. And many people have noted that the AI boom is driving most, possibly all, of the economy’s recent growth. So what will happen if the boom goes bust?

• By the way, some early subscribers to this newsletter may remember that I interviewed Jim Chanos, the famous short-seller, back in February, and he made the case then for an AI bubble. But people weren’t yet ready to hear it.

• Today’s primer will not be about the long run economic and social implications of AI, which is an entirely different subject. I will write in a future primer about technology, growth and jobs in general, with some speculations about AI. For today, however, I’m going to focus on what history — especially, but not only, the telecommunications boom of the 1990s — can tell us about the AI boom and its consequences.

• Beyond the paywall I will address the following: 1. The AI boom in historical perspective 2. The logic of technology manias and bubbles 3. The “Winner Take All” nature of technology bubbles 4. What will happen if the AI boom goes bust?

（推介原因：Paul Krugman 分析 AI 所帶來的影響，討論是否會出現泡沫爆破。）

维基百科：TOM 集團

https://zh.wikipedia.org/zh-tw/TOM%E9%9B%86%E5%9C%98

節錄：TOM 集團有限公司（股份代號：2383）為香港聯合交易所主板上市的科技與媒體公司。 TOM 集團科技平台及投資相關的業務包括移動網際網路、電子商貿、金融科技和先進的大數據分析。此外，其媒體業務涵蓋出版和廣告經營。集團總部設於香港，並於北京及台北設立地區總部，TOM 集團為長江和記實業有限公司成員。公司在開曼群島註冊，主席為陸法蘭。TOM 集團（前稱 TOM.COM）2000 年 3 月 1 日在香港創業板上市，當時上市編號為 8001.HK。TOM.COM當時上市招股價為 1.78 元，上市首日收報 7.75 元，較招股價上升 3.35 倍。

（提示：2000 年 3 月科網股泡沫爆破。）

维基百科：36.com

https://zh.wikipedia.org/wiki/36.com

節錄：36.com 控股有限公司（除牌當時為 8036.HK），曾在香港交易所創業板上市，由鄭經翰和楊國猛創立。業務當時經營 36.com 網站和《茶杯》雜誌。後來因為科網股泡沫爆破，於是出售給東方魅力，更名由經營 M Channel 的流動廣告取代上市地位。

History Doesn’t Repeat Itself, But it Often Rhymes

Posted by davidjkentwriter in Abraham Lincoln, On Writing

https://hotwhitesnow.wordpress.com/2024/04/18/history-doesnt-repeat-itself-but-it-often-rhymes/

Excerpt: The famous quote in the title is almost universally attributed to Mark Twain. It’s often said that history repeats itself. Twain supposedly took a different tack, suggesting that while it doesn’t repeat itself, it often rhymes. 

As with many famous quotes, the person being quoted likely didn’t say it. Twain (nee Samuel Clemens) died in 1910 and apparently the first time he got credit for the saying was sixty years later, in 1970. That doesn’t mean he didn’t say something like it. Indeed, in the 1874 novel The Gilded Age: A Tale of Today, co-written with his neighbor, he did begin a more flowery sentence with a variation of the first clause: “History never repeats itself, but the Kaleidoscopic combinations of the pictured present often seem to be constructed out of the broken fragments of antique legends.”

This Time Is Different: Eight Centuries of Financial Folly

Carmen M. Reinhart and Kenneth S. Rogoff

The acclaimed New York Times bestselling history of financial crises

https://press.princeton.edu/books/paperback/9780691152646/this-time-is-different?srsltid=AfmBOopbXYNzIKRsq4XPyUMhft8p5z_ef-904CFZu2zcKrUrLhaDrYQs

Intro: Throughout history, rich and poor countries alike have been lending, borrowing, crashing, and recovering their way through an extraordinary range of financial crises. Each time, the experts have chimed, “this time is different”—claiming that the old rules of valuation no longer apply and that the new situation bears little similarity to past disasters. With this breakthrough study, leading economists Carmen Reinhart and Kenneth Rogoff definitively prove them wrong.

Covering sixty-six countries across five continents and eight centuries, This Time Is Different presents a comprehensive look at the varieties of financial crises—including government defaults, banking panics, and inflationary spikes—from medieval currency debasements to the subprime mortgage catastrophe. Reinhart and Rogoff provocatively argue that financial combustions are universal rites of passage for emerging and established market nations.

A remarkable history of financial folly, This Time Is Different will influence financial and economic thinking and policy for decades to come.

（推介原因：這本書的主題：各種金融危機看起來是「這次不一樣」，但其實在某一個時空或另一個國家或地區，都曾經發生過。）

Investopedia

Tulipmania: About the Dutch Tulip Bulb Market Bubble

By Adam Hayes

April 30, 2026

https://www.investopedia.com/terms/d/dutch_tulip_bulb_market_bubble.asp

Definition

The Dutch tulip market bubble was a 17th-century event in which tulip bulbs became highly overvalued due to speculative trading, ultimately leading to a dramatic market crash.

Key Points:

• Tulipmania, which took place in the Netherlands during the 1600s, is widely considered the first recorded financial bubble, where speculation led to tulip bulbs being priced at levels comparable to luxury goods.

• At its peak, the value of rare tulip bulbs soared to the cost of a grand canal-side mansion in Amsterdam, illustrating the extreme nature of the market frenzy before it collapsed.

• Despite popular belief, some modern scholars argue that the extent of the tulipmania might have been exaggerated and serves more as a lesson in the dangers of speculation and greed than a reflection of an actual economic crisis.

• The crash of the tulip bulb market did not devastate the Dutch economy but did lead to a cultural shock, impacting social relationships and credit reliability.

• Tulipmania acts as a model for the typical financial bubble cycle, seen in various modern examples like NFTs and dot-com stocks, where irrational exuberance drives prices unsustainably high before a crash ensues.

Tulipmania is a model for the general cycle of a financial bubble:

• Investors lose track of rational expectations.

• Psychological biases lead to a massive upswing in the price of an asset or sector.

• A positive feedback cycle continues to inflate prices.

• Investors realize that they’re holding an irrationally priced asset.

• Prices collapse due to a massive sell-off, and an overwhelming majority go bankrupt.

• Similar cycles have been observed in the prices of Beanie Babies, baseball cards, non-fungible tokens (NFTs), and shipping stocks. 

What Does Tulipmania Have to Do With Market Bubbles?

Tulipmania reflects the general cycle of a bubble, from the irrational biases and group mentalities that push up prices of an asset to an unsustainable level to the eventual collapse of those inflated prices. The example of tulipmania is used as a parable for other speculative assets, such as cryptocurrencies or dot com stocks.

How Does Tulipmania Relate to Bitcoin?

The bitcoin market is frequently compared with tulipmania. Both prompted highly speculative prices for a product with little clear utility. Bitcoin prices tend to crash after significant gains, exhibiting many signs of a classic bubble.

The Bottom Line

The Dutch tulip mania of the 1600s is a prime example of an asset bubble driven by irrational exuberance and group psychology. Some economists and historians question whether tulipmania was the widespread financial crisis that is referenced today in relation to other bubbles like dot com stocks before 2001, the subprime housing market before 2008. They suggest that the idea of tulipmania has been greatly exaggerated as a parable or lesson in taming greed and excess.

维基百科：鬱金香狂熱

https://zh.wikipedia.org/zh-tw/%E9%AC%B1%E9%87%91%E9%A6%99%E7%8B%82%E7%86%B1

鬱金香狂熱（荷蘭文：Tulpenmanie）1637 年發生在荷蘭，是世界上最早的泡沫經濟事件。當時由鄂圖曼土耳其引進的鬱金香球根異常地吸引大眾搶購，導致價格瘋狂飆高，然而在泡沫化過後，價格僅剩下高峰時的百分之一，讓荷蘭各大都市陷入混亂。這次事件和英國的南海泡沫事件以及法國的密西西比公司並稱為近代歐洲三大泡沫事件。2020 年代起隨著比特幣價格飆升，有人將比特幣熱潮類比作鬱金香狂熱。

相關的文章： 

萬寶路

2016 年 1 月 23 日

https://xiaoshousha.blogspot.com/2016/01/blog-post_23.html

節錄：這首充滿美國風格的音樂，有個非正式的名字叫 Marlboro theme song，大陸網民叫＜萬寶路進行曲＞。出自上世紀六十年代的美國西部電影 The Magnificent Seven（港譯：七俠蕩寇志），故事改編自黑澤明 (Akira Kurosawa) 於上世紀五十年代執導的日本電影＜七武士＞。美國版是得到日方正式授權改編的版本。演員名單：美國版的七俠，包括 Yul Brynner（港譯：尤伯連納）和 Steve McQueen（港譯：史提夫麥昆）。日本版的七俠，包括志村喬 (Takashi Shimura) 和三船敏郎 (Toshirō Mifune) ，這兩位都是黑澤明的愛將。

東尼（一）

2010 年 4 月 2 日

https://xiaoshousha.blogspot.com/2010/04/blog-post.html

節錄：他主張「價值投資法」(Value Investing)，發掘有潛力但是現階段被低估的股票，然後長期持有 (Buy and Hold) ，即是巴菲特 (Warren Buffett, 1930-) 的那一套。上一代阿媽教女兒揀老公，也是這樣的。東尼認為，「價值投資法」簡單易用，可以讓你安枕無憂。買入有價值的股票，千萬不要輕易放棄。他的揀股之道非常傳統，會叫你參考資產淨值 (Net Asset Value, NAV)，因為它是一條底線 (Bottom line) 或者安全網 (Safety net)，而盈利預測有可能出錯（因此影響到市盈率 P/E 的參考價值）。除了資產淨值 (NAV)，亦須注意股息的分派 (Dividend payout) 和增長潛力 (Growth potential)。

巫婆發功

2020 年 3 月 17 日

https://xiaoshousha.blogspot.com/search?q=%E5%B7%AB%E5%A9%86

節錄：上一次出現同樣況，是 Auntie 出動 The Hongkong & Shanghai Big Crab 插圖（那篇文章是＜前記者＞系列的第一篇），之後恒指插水式下瀉，也許我不應該再評論股市走勢，否則變女版秋官，被股民埋怨，甚至當女巫燒死，哈哈。