YouTube2Text

The AI Bubble Explained Like You're 5 — Transcript

by Crayon Capital · 1,689 words · 278 segments · language en · Watch on YouTube

Full transcript

  1. 0:00Picture this. The CEO of the most hyped
  2. 0:03tech company on Earth just admitted his
  3. 0:05$200 per month subscription loses money
  4. 0:08on every single customer. Not a little
  5. 0:10money, not break even with growth
  6. 0:12potential. Loses money. And Wall
  7. 0:16Street's response, they valued his
  8. 0:18company at $500 [music] billion. Anyway,
  9. 0:21welcome to the AI bubble, where we went
  10. 0:23from cool chatbot to $7 trillion gamble
  11. 0:26in 24 months, and why the people writing
  12. 0:28the checks might be in for the rudest
  13. 0:30awakening since 2008.
  14. 0:35November 30th, 2022.
  15. 0:38Just a regular Wednesday. Unremarkable
  16. 0:40cold brew flowing. Engineers debugging
  17. 0:43code. Then Open AAI dropped chat GPT.
  18. 0:47Within 48 hours, the internet had a new
  19. 0:49obsession. People were using it to write
  20. 0:52breakup texts, debug code, explain black
  21. 0:54holes, and generate bedtime stories.
  22. 0:57Dad, I'm talking to an AI that knows
  23. 1:00everything. It just explained quantum
  24. 1:02entanglement using Minecraft terms. And
  25. 1:04did I hear Minecraft? It has 1 million
  26. 1:07users in the first 5 days of launch. 2
  27. 1:10months later, 100 million users. Fastest
  28. 1:13growing app in history. Instagram took
  29. 1:172.5 years to hit 100 million. Facebook
  30. 1:20took 4.5 years. Chat GPT did it in 60
  31. 1:24days and Silicon Valley absolute panic.
  32. 1:27Inside Google's headquarters, executives
  33. 1:29were having a meltdown. We invented this
  34. 1:32technology. The transformer model that
  35. 1:35was us. The research paper has our names
  36. 1:37on it. Then why is my son using chat GPT
  37. 1:40and not Google AI? because we were too
  38. 1:43scared to release it. Well, release
  39. 1:46[music] something now. Call it barred,
  40. 1:48Gemini, whatever. Ship it. Google
  41. 1:51scrambled. They had the tech. They'd had
  42. 1:54it for years, but they were terrified of
  43. 1:56what it might say. Of lawsuits, of bad
  44. 1:59press. Open AI had no such fears. They
  45. 2:02shipped first. And in tech, first mover
  46. 2:05advantage is everything. Within weeks,
  47. 2:08Microsoft threw $10 billion at OpenAI
  48. 2:11with another $3 billion on standby.
  49. 2:13Google panic launched Bard, which
  50. 2:16immediately hallucinated in its first
  51. 2:17[music] demo. Amazon started dumping
  52. 2:20billions into Anthropic. We're a dog
  53. 2:23walking app, but the root optimization
  54. 2:24uses machine learning. Say no more.
  55. 2:27Here's $50 million. Don't you want to
  56. 2:30see our demo? Shut up and take my money.
  57. 2:33By mid 2023, if your slide deck
  58. 2:36mentioned neural networks or large
  59. 2:38language models, you were basically
  60. 2:39printing money. Companies with zero
  61. 2:42revenue, no product, just a team and a
  62. 2:45Google doc or raising $100 million
  63. 2:48rounds. The bubble gave us technological
  64. 2:50marvels like AI powered toothbrushes, AI
  65. 2:54recipe generators, AI dating coaches.
  66. 2:57One company raised $30 million to build
  67. 2:59an AI that summarizes potentially other
  68. 3:02AI outputs. We'd reached peak absurdity.
  69. 3:06But the real insanity, [music] the GPU
  70. 3:09shortage, these little computer chips,
  71. 3:11which look like harmless circuit boards,
  72. 3:13became more valuable than gold or oil.
  73. 3:16Nvidia became one of the most valuable
  74. 3:18companies on Earth. Their CEO Jensen
  75. 3:20Huang started wearing leather jackets to
  76. 3:22conferences like he was the Mcjagger of
  77. 3:25semiconductors and he kind of was
  78. 3:27[music]
  79. 3:28renting a single top tier age 100 GPU
  80. 3:31costs about $3 an hour. Sounds cheap,
  81. 3:34right? But startups don't need one, they
  82. 3:36need hundreds. A small startup trying to
  83. 3:39train [music] just one model might rent
  84. 3:41a 500 chip cluster. That's $1,500 per
  85. 3:44hour. $36,000 per day. over $1.1 million
  86. 3:49per month just for renting GPU. Yet,
  87. 3:52companies were still hoarding GPUs like
  88. 3:54doomsday [music] preppers stockpile
  89. 3:56canned beans. Nvidia couldn't
  90. 3:58manufacture them fast enough. The wait
  91. 4:00list was at one point up to 12 months
  92. 4:02long. Black markets [music] emerged.
  93. 4:05Seriously, people were smuggling $100
  94. 4:07bills across borders. By 2024,
  95. 4:10everything [music] was AI powered. Your
  96. 4:13toaster burns bread with 10% more
  97. 4:15efficiency. [music]
  98. 4:16Your calendar app suggests meeting times
  99. 4:19using algorithms. Your grocery store
  100. 4:21selfch [music] checkout. Just a camera,
  101. 4:23but now it's machine learning. The party
  102. 4:26was in full swing. Champagne flowing.
  103. 4:29Checks clearing. [music]
  104. 4:30Then somebody asked the uncomfortable
  105. 4:32question. How much does this actually
  106. 4:34cost?
  107. 4:36January 2025.
  108. 4:39Sam Alman, CEO of OpenAI, face of the AI
  109. 4:42revolution, tweeted something that made
  110. 4:44every investor's blood pressure spike.
  111. 4:47Insane thing. We are currently losing
  112. 4:49money on OpenAI Pro subscriptions.
  113. 4:52People use it much more than we
  114. 4:53expected. Yep, you heard it right. Open
  115. 4:56AAI charges $200 per month for their pro
  116. 4:59tier, and they lose money on every
  117. 5:01single subscriber, which means the more
  118. 5:04successful they are, the more money they
  119. 5:06lose at this rate. When they launched
  120. 5:08ChatGpt Plus at $20 per month, their
  121. 5:11entire pricing strategy was test two
  122. 5:14prices, $20 and $42. [music] See which
  123. 5:17people complain about less. Pick that
  124. 5:19one. No market research, no cost
  125. 5:22analysis, just Vibes. That's how a $500
  126. 5:25billion of company priced its flagship
  127. 5:27product, Vibes. But wait, it gets worse.
  128. 5:31Open AAI's 2024 finances revenue $3.7
  129. 5:36billion losses $5 billion. They didn't
  130. 5:40just fail to make a profit. They lost
  131. 5:42more money than they made by a lot. Now,
  132. 5:45here's where it gets truly insane.
  133. 5:47Industry forecasts show AI
  134. 5:49infrastructure will require $6.7
  135. 5:51trillion by 2030. Let me break that
  136. 5:54down. $5.2 $2 trillion directly for AI
  137. 5:58infrastructure, GPUs, accelerators,
  138. 6:00power upgrades, cooling, and data center
  139. 6:03construction. While the remaining $ 1.5
  140. 6:05trillion goes to traditional nonAI cloud
  141. 6:08infrastructure, the four largest tech
  142. 6:10companies alone are expected to spend
  143. 6:13$350 billion in 2025,
  144. 6:16one year. And here's the kicker. To
  145. 6:19justify these costs, the AI industry
  146. 6:21needs to generate $2 trillion in annual
  147. 6:23revenue by 2030. Current total AI
  148. 6:26revenues, somewhere around $200 billion.
  149. 6:30That's a 10 times gap.
  150. 6:33100 billion. Yep, [music] that's the
  151. 6:36number. Sam, our projections show us at
  152. 6:3920 billion in 2025. You're proposing we
  153. 6:42hit 100 billion by 2029. [music] It's
  154. 6:45the necessary growth trajectory.
  155. 6:48That's not a trajectory, Sam. That's a
  156. 6:50hockey stick drawn by someone who's
  157. 6:52never played hockey. Do you know what it
  158. 6:54costs to even attempt this?
  159. 6:57115 billion in cash burn by 2029.
  160. 7:00When do we become profitable?
  161. 7:032029, [music] maybe 2030 or 2031?
  162. 7:07And if revenue doesn't hit $100 billion,
  163. 7:11if this hockey stick of yours snaps,
  164. 7:13then we raise more money. From whom? Who
  165. 7:16is left? Microsoft? Saudi Arabia? A
  166. 7:20really optimistic pension fund? Plenty
  167. 7:22of fish in the sea, guys. Since 2021,
  168. 7:26private investors have pumped hundreds
  169. 7:27of billions into AI startups. Most of
  170. 7:30that went to [music] just 10 companies
  171. 7:32whose combined valuations swelled by
  172. 7:34nearly $1 trillion. Everyone's betting
  173. 7:37they'll become the next Google. Then the
  174. 7:39failures started rolling in. Chat GPT
  175. 7:42hallucinated. Tech speak for completely
  176. 7:44making things up and cited research
  177. 7:46papers [music] that don't exist, legal
  178. 7:48cases that never happened, and
  179. 7:50statistics from an alternate dimension.
  180. 7:52Google's AI told users to put glue on
  181. 7:55pizza to make cheese stick better. It
  182. 7:57learned this from a Reddit joke. Yes,
  183. 8:00really. The lawsuits arrived next.
  184. 8:02artists, writers, musicians,
  185. 8:05journalists, all asking the same
  186. 8:06question. [music] Did you train your AI
  187. 8:09on my copyrighted work? Answer: Um, yes.
  188. 8:14The New York Times sued Open AI. Getty
  189. 8:17Images sued Stability AI. Authors sued
  190. 8:21everyone. Musicians sued everyone else.
  191. 8:25[music] We didn't steal. We learned from
  192. 8:27publicly available data. You scraped
  193. 8:30[music] the entire internet without
  194. 8:31permission. and used it commercially.
  195. 8:34Well, when you say it like that, it
  196. 8:35sounds
  197. 8:36>> [music]
  198. 8:36>> illegal. By mid 2024, the cracks were
  199. 8:39showing. The vibe shifted from euphoria
  200. 8:41to something more dangerous. Doubt.
  201. 8:48The hype was loud. The math was ugly.
  202. 8:51And now everyone had the same
  203. 8:52uncomfortable question buzzing in their
  204. 8:54heads like a mosquito at 3:00 a.m. So,
  205. 8:58when does this whole [music] thing
  206. 8:59explode? This looks like a bubble. The
  207. 9:02losses are insane. The valuations are
  208. 9:04dulu. But this bubble isn't popping the
  209. 9:07same way the dot bubble did. Why?
  210. 9:09[music] Because the people inflating it
  211. 9:11are also the people with the biggest
  212. 9:12wallets on Earth. Unlike 1999 when
  213. 9:16startups had zero revenue and 17
  214. 9:18foosball tables, today's AI bubble is
  215. 9:20being funded by big tech companies that
  216. 9:22actually make money. Real money. Buy an
  217. 9:26entire country if we feel bored. Money.
  218. 9:28Microsoft prints billions from Azure.
  219. 9:31Google prints billions from ads. Meta
  220. 9:34prints billions from showing you memes
  221. 9:35your cousin already sent on WhatsApp.
  222. 9:38These giants aren't [music] investing in
  223. 9:39AI because it's profitable. They're
  224. 9:42investing because they're terrified of
  225. 9:44being the ones who didn't. If AI becomes
  226. 9:47the next [music] iPhone moment, the next
  227. 9:49Google moment, the next entire world
  228. 9:51changes overnight moment, then the
  229. 9:53company that slowed down becomes
  230. 9:54Blackberry. And nobody wants to be
  231. 9:57Blackberry. Think of big tech like poker
  232. 9:59players who all brought unlimited chips.
  233. 10:02If Microsoft folds, Google wins. If
  234. 10:06Google folds, Amazon wins. If Amazon
  235. 10:09folds, Meta [music] wins. No one blinks,
  236. 10:13no one folds, everyone keeps raising the
  237. 10:16stakes. And that brings [music] us to
  238. 10:18the real uncomfortable question. If
  239. 10:20everyone keeps spending, how do bubbles
  240. 10:22ever pop? Short answer is they burst
  241. 10:26when a real world event forces everyone
  242. 10:28to stop pretending. [music] And
  243. 10:30historically, that event usually comes
  244. 10:33in three forms. One, the money suddenly
  245. 10:36disappears. The dot bubble didn't
  246. 10:38collapse because websites were dumb. It
  247. 10:40collapsed because the cash spigot shut
  248. 10:42off. [music]
  249. 10:43Interest rates rose. Borrowing got
  250. 10:46expensive. Funding vanished. And
  251. 10:48thousands of [music] companies
  252. 10:48evaporated overnight. If interest rates
  253. 10:51rise anytime soon and big techs tighten
  254. 10:53their budgets, the AI bubble ends
  255. 10:56instantly.
  256. 10:58Two, a big [music] embarrassing meltdown
  257. 11:00happens. Every bubble has a moment. One
  258. 11:03spectacular failure that freaks everyone
  259. 11:05out. Do had earnings disasters. 2008 had
  260. 11:08layman. Crypto had FTX. AI's version
  261. 11:12could be a giant lab revealing massive
  262. 11:15losses, a botched model launch or a
  263. 11:18lawsuit that freezes everything. So,
  264. 11:21where are we [music] right now? Big tech
  265. 11:24is still spending. No meltdown earnings
  266. 11:26call yet. No FTX [music] level scandal.
  267. 11:29Reality is showing cracks, but not
  268. 11:31enough to break confidence. So, yes,
  269. 11:34this bubble is inflated. Yes, the
  270. 11:37numbers are insane, but it won't burst
  271. 11:39because of hype. It bursts when one of
  272. 11:42the giants finally says, "All right,
  273. 11:44that's enough." Until then, the bubble
  274. 11:48floats. We're not at the explosion.
  275. 11:51We're at the tightening. The inhale
  276. 11:53before the scream. The tension before
  277. 11:55the snap. The episode right before the
  278. 11:58season finale.

About this transcript

This page contains the full transcript of The AI Bubble Explained Like You're 5 by Crayon Capital, generated from the public captions YouTube serves with the video. The transcript has 1,689 words across 278 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.