YouTube2Text

Is the AI Bubble About to Pop? (Explained Like You're 5) — Transcript

by PENNY · 2,457 words · 356 segments · language en · Watch on YouTube

Full transcript

  1. 0:00July 2026. Leopold Aschenbrenner, a
  2. 0:03young former OpenAI researcher, runs a
  3. 0:05fund that bets billions on AI. He's
  4. 0:08right. AI keeps growing. And still, in a
  5. 0:10single month, his fund shrinks from
  6. 0:12about $45 billion
  7. 0:14to about $10 billion. How can you be
  8. 0:16right about the future and still lose
  9. 0:18that much money? Then, last week, AI and
  10. 0:21chip stocks lost more than $1 trillion
  11. 0:24in value in less than a week. So, is the
  12. 0:27AI bubble about to pop?
  13. 0:28>> Pop? I use AI every day. It's amazing.
  14. 0:31>> The internet was amazing, too. Its
  15. 0:34bubble still popped because bubbles
  16. 0:36don't pop when the technology stops
  17. 0:37working. They pop when the money runs
  18. 0:39out. And the money behind AI is doing
  19. 0:41strange things. Some of it goes in a
  20. 0:44circle. And one of the richest companies
  21. 0:46on Earth just borrowed money for 100
  22. 0:48years. If it pops, it's not just
  23. 0:50billionaires who pay. If you save for
  24. 0:52retirement, it might be you. So, let's
  25. 0:55follow the money step by step. And by
  26. 0:57the end, you'll know three warning signs
  27. 0:59to watch for yourself. First, what is a
  28. 1:02bubble? Let's make AI tiny. Imagine a
  29. 1:05small village by a river. One morning,
  30. 1:07someone finds gold in the water.
  31. 1:09>> Gold? If I find enough, I'm rich.
  32. 1:12>> To dig, Mia needs a shovel. And the only
  33. 1:14one who makes shovels is Hank. Soon, the
  34. 1:17whole village wants to dig. So, everyone
  35. 1:19wants a shovel.
  36. 1:20>> Shovels! Get your shovels! Gold not
  37. 1:22included.
  38. 1:23>> So, who gets rich in a gold rush? Take a
  39. 1:26guess.
  40. 1:27Often, not the diggers. The ones selling
  41. 1:29the shovels. Today, AI is the gold. The
  42. 1:32diggers are companies like Google,
  43. 1:34Microsoft, Meta, and OpenAI. And the
  44. 1:36biggest shovel seller is Nvidia, which
  45. 1:38makes the chips AI runs on. So, when
  46. 1:41does it become a bubble? The famous
  47. 1:43investor George Soros said, "Every
  48. 1:45bubble starts with something real. There
  49. 1:47really is gold in the river. But then,
  50. 1:49people make a mistake. They think
  51. 1:51there's more gold or that it comes
  52. 1:52faster than it really does. And that
  53. 1:54mistake is what pushes prices way too
  54. 1:57high. So, AI is useful doesn't answer
  55. 1:59the question. The gold can be real and
  56. 2:02it can still be a bubble. That's exactly
  57. 2:04what happened with the internet. Let's
  58. 2:05go back to the late 1990s. Companies
  59. 2:08laid glass cables all over America
  60. 2:10because a famous number said internet
  61. 2:12traffic doubled every 100 days. That's
  62. 2:14about 12 times more every year. In
  63. 2:17reality, it doubled about once a year.
  64. 2:19Sounds like a small mistake, but after 3
  65. 2:21years, it's the difference between 2,000
  66. 2:24times more and eight. So, they built way
  67. 2:27too much. After the crash, only a few
  68. 2:29percent of those cables were even
  69. 2:30switched on. Cisco, which sold the
  70. 2:32shovels of the internet, was briefly the
  71. 2:34most valuable company in the world in
  72. 2:36March 2000. Two years later, it had lost
  73. 2:39about 80% of its value. But the internet
  74. 2:42was real. It just came slower and paid
  75. 2:44less than people had bet. And it wasn't
  76. 2:46the first time. In the 1920s,
  77. 2:49electricity, cars, and radio were brand
  78. 2:51new and all of them were real. The stock
  79. 2:54market still crashed in '29. So, the
  80. 2:56real question isn't is AI useful? It's
  81. 2:59do the numbers add up? And to check
  82. 3:01that, we have to look at how much money
  83. 3:03is going in. This year, the four biggest
  84. 3:05diggers, Microsoft, Google, Amazon, and
  85. 3:07Meta, plan to spend over 700 billion
  86. 3:10dollars, mostly on AI data centers. A
  87. 3:13data center is a giant building full of
  88. 3:15computer chips. It's where AI does its
  89. 3:17thinking. 700 billion is hard to even
  90. 3:20imagine. So, picture this. You spend a
  91. 3:22thousand dollars every single second day
  92. 3:25and night, no sleep. To spend 700
  93. 3:27billion, you would need about 22 years.
  94. 3:30Why so much? Because nobody wants to be
  95. 3:32too late. Meta's boss, Mark Zuckerberg,
  96. 3:35said he'd rather waste a couple hundred
  97. 3:36billion dollars than be late. These are
  98. 3:38the richest companies on Earth. Surely
  99. 3:41they can pay that from their own pocket?
  100. 3:42Mostly, yes. But between April and June,
  101. 3:45Google spent more on building than its
  102. 3:47whole business brought in. For the first
  103. 3:49time since it joined the stock market in
  104. 3:512004. And when your own money isn't
  105. 3:53enough, you borrow. So, Google borrowed
  106. 3:55money from investors. It pays them
  107. 3:57interest every year and pays the loan
  108. 3:59back at the end. And the end is 100
  109. 4:02years from now. 100 years. The last tech
  110. 4:05company that did that was Motorola in
  111. 4:071997, just before the internet bubble
  112. 4:09popped. That doesn't mean it will happen
  113. 4:11again, but it's a strange repeat, and
  114. 4:14some borrowing is harder to see. Meta
  115. 4:17needed a giant data center in Louisiana
  116. 4:19for around $27 billion.
  117. 4:21Instead of borrowing that itself, Meta
  118. 4:23set up a separate company with a
  119. 4:25partner. That company borrowed the
  120. 4:26money, built the data center, and rents
  121. 4:28it to Meta. So, the loan isn't in Meta's
  122. 4:31own books. But Meta promised to pay
  123. 4:33extra if the building ends up worth
  124. 4:35less. In the village, it would sound
  125. 4:37like this.
  126. 4:37>> I don't have any debt. I just rent this
  127. 4:39barn.
  128. 4:40>> And if the barn loses its value?
  129. 4:42>> Then I pay. It says so right here.
  130. 4:44>> That's allowed, and it's common. But it
  131. 4:46means there's more borrowed money behind
  132. 4:48AI than you see at first. So, a mountain
  133. 4:51of money is going in, some of it
  134. 4:52borrowed. That only works if even more
  135. 4:55money comes back out. So, does it? Let's
  136. 4:58go back to the village. Hank wants his
  137. 5:00shovel sales to look amazing, so he has
  138. 5:02an idea.
  139. 5:03>> Mia, I'll give you 100 coins, but you
  140. 5:05have to buy my shovels with them.
  141. 5:07>> Deal.
  142. 5:08>> So, Mia buys 100 coins worth of shovels
  143. 5:11from Hank with Hank's own money.
  144. 5:13>> Look, I sold 100 coins of shovels.
  145. 5:16>> But how much new money came into Hank's
  146. 5:18shop? Zero. It's his own money walking
  147. 5:20in a circle. Real life is more
  148. 5:22complicated, but parts of AI look a bit
  149. 5:24like this. In 2025, Nvidia announced it
  150. 5:28would invest up to $100 billion
  151. 5:30in OpenAI. And OpenAI uses money like
  152. 5:33that to buy Nvidia chips. Nvidia also
  153. 5:36bought $2 billion of shares in
  154. 5:38CoreWeave, a company that rents out AI
  155. 5:40computers. And CoreWeave buys Nvidia
  156. 5:43chips. And it's not just them. The BIS,
  157. 5:46a kind of bank for the world's central
  158. 5:47banks, found that more than half of the
  159. 5:49money put into AI companies comes from
  160. 5:51other AI companies. To be fair, the
  161. 5:54chips are real and they really work. But
  162. 5:56the circle makes sales look bigger than
  163. 5:58the new money coming in. So, who brings
  164. 6:00new money in? Only the people at the
  165. 6:02very end. The ones who pay for AI. You,
  166. 6:05me, and businesses. In the village,
  167. 6:07that's the gold buyer. He's the only one
  168. 6:09who pays with money from outside. So,
  169. 6:11how much are customers paying? OpenAI's
  170. 6:14sales are close to 50 billion dollars a
  171. 6:16year. Anthropic's are about 67 billion a
  172. 6:18year going by its last 3 months. That's
  173. 6:21growing incredibly fast. But together,
  174. 6:23it's still far less than the 700 billion
  175. 6:25being spent this year. Now, a data
  176. 6:27center lasts for years, so it doesn't
  177. 6:30have to pay for itself in one. But until
  178. 6:32the customers catch up, someone has to
  179. 6:34cover the gap. And more and more, that
  180. 6:36someone is borrowed money. That's where
  181. 6:39it gets dangerous. Back in the village,
  182. 6:41meet Rita. She doesn't dig and she
  183. 6:43doesn't make shovels. She rents them
  184. 6:45out.
  185. 6:45>> Can't afford a shovel? Rent one from me.
  186. 6:48>> To buy all those shovels, Rita borrowed
  187. 6:50from the village banker.
  188. 6:51>> 100 coins. You pay me eight coins of
  189. 6:53interest every year.
  190. 6:55>> Easy. Mia rents my shovels for 12 coins
  191. 6:57a year, fixed for 5 years. You only want
  192. 7:00eight.
  193. 7:01>> In real life, Rita is a company called
  194. 7:03CoreWeave. It has borrowed around
  195. 7:05[clears throat] 52 billion dollars to
  196. 7:06buy AI chips and rent them out. Like
  197. 7:09Mia, its customers rent those chips on
  198. 7:11long contracts. And like Rita, it pays
  199. 7:13interest on everything it borrowed.
  200. 7:15Then, last week, the people who lent
  201. 7:17CoreWeave money got nervous and started
  202. 7:19selling those loans to others. Here's
  203. 7:21why that matters. When people are afraid
  204. 7:23you can't pay them back, they'll only
  205. 7:25lend you more if you pay more interest.
  206. 7:27One of CoreWeave's loans pays about 9.6%
  207. 7:30On the market, buyers suddenly wanted
  208. 7:32about 15. Its old loans don't change,
  209. 7:35but every new loan could get that
  210. 7:37expensive. And CoreWeave still needs
  211. 7:39billions more for data centers it
  212. 7:40already promised. Think of it like rent.
  213. 7:43It's as if your rent jumped from $960 to
  214. 7:461,500 a month. Same home, just way more
  215. 7:49expensive. Its stock fell about 15% in
  216. 7:51one day. Its boss called that irrational
  217. 7:54and pointed to contracts worth around a
  218. 7:56hundred billion dollars. Maybe he's
  219. 7:58right, but there's a second problem and
  220. 8:00it's not just Coreweave's. Prices in
  221. 8:02America went up faster than expected.
  222. 8:04Oil got expensive. So many now expect
  223. 8:07America's central bank to raise interest
  224. 8:09rates. And then everything bought with
  225. 8:11borrowed money gets more expensive. Back
  226. 8:13in the village.
  227. 8:14>> Rita, times have changed. Any new loan
  228. 8:17costs 15 coins a year.
  229. 8:18>> 15? But Mia only pays me 12.
  230. 8:21>> That's the trap. Mia still pays Rita 12
  231. 8:24coins a year, but every new loan now
  232. 8:27costs Rita 15. She pays out more than
  233. 8:29she earns. And remember Leopold from the
  234. 8:31start? This is exactly what happened to
  235. 8:33his fund. He was right about AI, but the
  236. 8:36fund had bought with borrowed money.
  237. 8:38When prices dipped, the people who lent
  238. 8:40him money wanted cash, right now. So he
  239. 8:43had to sell. You can be right about the
  240. 8:45future and still go broke before it
  241. 8:47arrives. And that's how bubbles usually
  242. 8:49pop. Not because the technology stops
  243. 8:51working. The famous investor Ray Dalio
  244. 8:53put it simply, "Wealth isn't the same as
  245. 8:56money. You can't spend wealth. To get
  246. 8:58money, you have to sell it." So when
  247. 9:00interest goes up and the lenders want
  248. 9:02their money back, everyone who borrowed
  249. 9:04has to sell, all at the same time.
  250. 9:06Prices fall. That forces even more
  251. 9:08people to sell. Like dominoes. So is AI
  252. 9:12doomed? Not so fast. The other side has
  253. 9:15strong arguments. Back in 2000, the
  254. 9:17cables lay in the ground and nobody used
  255. 9:19them. Today, big investors say nobody
  256. 9:22can get enough AI chips. Even old chips
  257. 9:24still rent for high prices. In the
  258. 9:26village, back then, shovels piled up in
  259. 9:29the shop. Today, people wait in line for
  260. 9:31them. And the money is real. Anthropic
  261. 9:33sales grew about 45% in just three
  262. 9:36months because businesses pay for AI
  263. 9:38that saves them time. Also, the biggest
  264. 9:40spenders aren't tiny startups like in
  265. 9:422000. Microsoft, Google, Amazon, and
  266. 9:45Meta make huge profits from other
  267. 9:47things. They won't disappear overnight.
  268. 9:49And even when a bubble pops, the
  269. 9:50technology usually wins. Those empty
  270. 9:53cables from 2000, today they carry your
  271. 9:56videos. But there's one catch that
  272. 9:57worries both sides. Even if AI changes
  273. 10:00the world, AI companies might not keep
  274. 10:02the money. Think about airplanes. They
  275. 10:04changed the world, but for decades
  276. 10:07airlines barely made money. Warren
  277. 10:09Buffett once joked that added up, they'd
  278. 10:11earned about zero since the very first
  279. 10:13flight. Why? Every airline sells the
  280. 10:15same seat from A to B. So, you pick the
  281. 10:18cheapest. AI could go the same way.
  282. 10:20People switch between AI apps all the
  283. 10:22time. One study found OpenAI's share of
  284. 10:25AI sold to businesses fell from about
  285. 10:27half to about a quarter in 2 years.
  286. 10:29Great for you. Not so great for the
  287. 10:31companies that spent billions. So, who's
  288. 10:33right? Ray Dalio, one of the few
  289. 10:35investors who saw the 2008 crash coming,
  290. 10:38says he sees clear signs of a bubble.
  291. 10:40But he also says it's not yes or no,
  292. 10:43it's how big. And anyone who tells you
  293. 10:45exactly when it pops is guessing. But if
  294. 10:47it does pop, remember the question from
  295. 10:49the start. Who pays? First, the owners
  296. 10:52of AI companies. And that's not just
  297. 10:54billionaires. Nvidia alone is about 8%
  298. 10:57of America's 500 biggest companies all
  299. 11:00added together. So, if your retirement
  300. 11:02savings follow those 500 companies, like
  301. 11:04many do, you probably own a small piece
  302. 11:06of the AI bet, often without knowing it.
  303. 11:09Second, the people who lent the money.
  304. 11:12Many AI loans are bought by pension
  305. 11:13funds and insurance companies. And
  306. 11:15third, workers. Builders, electricians,
  307. 11:18and engineers. The International
  308. 11:20Monetary Fund, which watches the world
  309. 11:21economy, just said the economy is
  310. 11:23standing on one leg. The AI leg. In one
  311. 11:26of its what if tests, a sharp AI crash
  312. 11:30shrinks the whole world economy by about
  313. 11:321% by 2028. Not a prediction, but it
  314. 11:35shows how much now depends on AI. So, is
  315. 11:38AI a bubble? And if it is, will it pop?
  316. 11:41And how hard? Nobody knows for sure, but
  317. 11:44you don't have to guess. You can watch
  318. 11:46three warning signs yourself, and you
  319. 11:48already know all of them. Sign one, real
  320. 11:50customers. Do people and businesses pay
  321. 11:53more for AI every few months? If yes,
  322. 11:56the money is real and the spending can
  323. 11:57pay off. If sales stall, or it's mostly
  324. 12:00the same money going in circles, the AI
  325. 12:02companies have spent far more than comes
  326. 12:04back. Sign two, the cost of borrowing.
  327. 12:07Remember Rita? Watch what lenders ask AI
  328. 12:10companies to pay. If borrowing keeps
  329. 12:12getting more expensive, companies with
  330. 12:13lots of debt get squeezed first. Like
  331. 12:16Rita, they may have to sell, and prices
  332. 12:18can fall like dominoes. Sign three,
  333. 12:20time. How long until a data center earns
  334. 12:23back what it cost? Some investors say
  335. 12:25the best ones do it in one to three
  336. 12:27years. If that keeps getting longer, the
  337. 12:29money comes back too slowly. Then,
  338. 12:31spending gets cut, and the shovel
  339. 12:33sellers, like Nvidia, feel it first.
  340. 12:35None of these signs tells you the exact
  341. 12:37day, but if all three turn red at the
  342. 12:39same time, the bubble gets more
  343. 12:41dangerous. If they stay green, AI may
  344. 12:44grow into its price. So, is it popping?
  345. 12:47Nobody knows. But Leopold's story shows
  346. 12:49how it usually ends. Not when the dream
  347. 12:51dies, when the borrowed money runs out.
  348. 12:54But the biggest borrower in the world
  349. 12:55isn't Google or Coreweave. It's a
  350. 12:58country, and it owes more than 40
  351. 13:00trillion dollars. Who lent it all that
  352. 13:02money? What happens when interest goes
  353. 13:04up? And can it ever be paid back?
  354. 13:06In the next video, we look at America's
  355. 13:0840 trillion dollars of debt. See you
  356. 13:10there.

About this transcript

This page contains the full transcript of Is the AI Bubble About to Pop? (Explained Like You're 5) by PENNY, generated from the public captions YouTube serves with the video. The transcript has 2,457 words across 356 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to choose a caption track available on the video.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, caption language selection, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.