1 00:00:00,200 --> 00:00:01,900 $2 trillion in real money 2 00:00:01,900 --> 00:00:03,800 Are rushing into the AI track crazily 3 00:00:03,900 --> 00:00:06,366 But there are only two endings for this money 4 00:00:06,400 --> 00:00:08,200 Or let the stock market completely skyrocket 5 00:00:08,533 --> 00:00:10,700 Or drag the entire market into the abyss 6 00:00:10,866 --> 00:00:11,933 The scariest thing is 7 00:00:11,933 --> 00:00:12,933 It seems now 8 00:00:12,933 --> 00:00:14,700 The latter is much more likely 9 00:00:14,933 --> 00:00:16,766 And this is definitely not Silicon Valley 10 00:00:16,766 --> 00:00:18,733 Big bosses fight behind closed doors with gods 11 00:00:18,800 --> 00:00:20,700 It will affect your wallet directly 12 00:00:20,866 --> 00:00:22,066 According to recent data 13 00:00:22,066 --> 00:00:23,533 Today in 2026 14 00:00:23,600 --> 00:00:26,166 62% of Americans own stocks 15 00:00:26,166 --> 00:00:28,366 This is the highest level in the past 20 years 16 00:00:28,666 --> 00:00:30,300 Do you think what you bought is stable? 17 00:00:30,300 --> 00:00:33,800 Pension 401K or S&P 500 Index Fund 18 00:00:33,933 --> 00:00:36,066 Do you think you will be passive if you throw money in? 19 00:00:36,066 --> 00:00:38,300 Yield is the safest long-term strategy 20 00:00:38,666 --> 00:00:40,066 After all, after a thousand years of 21 00:00:40,066 --> 00:00:42,400 The Internet Bubble and the 2008 Subprime Crisis 22 00:00:42,466 --> 00:00:44,300 The lesson most people learn is that 23 00:00:44,300 --> 00:00:46,300 Just keep holding the S&P 500 24 00:00:46,300 --> 00:00:47,300 You can always get your money back 25 00:00:47,333 --> 00:00:48,966 But the current market is no longer 26 00:00:48,966 --> 00:00:50,266 That's the market you thought it was 27 00:00:50,266 --> 00:00:51,866 You buy the S&P 500 28 00:00:51,900 --> 00:00:54,666 Essentially, you are betting on a very small number of companies. 29 00:00:54,666 --> 00:00:57,666 Companies that rely on AI to survive can continue to rise forever 30 00:00:57,933 --> 00:01:00,666 52% weighting in the S&P 500 31 00:01:00,733 --> 00:01:03,700 Now it is tightly concentrated in the hands of only 20 companies. 32 00:01:03,700 --> 00:01:06,466 Almost all of these companies are riding on the popularity of AI 33 00:01:06,733 --> 00:01:09,400 Every month your pension account is debited 34 00:01:09,400 --> 00:01:12,466 Most of them are actually sent directly to NVIDIA. 35 00:01:12,533 --> 00:01:15,600 Giants like Microsoft and Google are engaged in an AI arms race 36 00:01:15,600 --> 00:01:17,000 Whether you understand technology or not 37 00:01:17,000 --> 00:01:18,266 Whether you like it or not 38 00:01:18,333 --> 00:01:19,900 You have been forcibly tied up 39 00:01:19,900 --> 00:01:21,400 This sports car has no brakes 40 00:01:21,666 --> 00:01:22,533 Once the car rolls over 41 00:01:22,533 --> 00:01:23,800 The consequences are devastating 42 00:01:23,800 --> 00:01:25,333 It's not just hedge fund managers 43 00:01:25,333 --> 00:01:27,066 It's so easy to fall on Wall Street 44 00:01:27,200 --> 00:01:29,200 The biggest hit will be annual income 45 00:01:29,200 --> 00:01:31,333 Ordinary people with $30,000 to $80,000 46 00:01:31,333 --> 00:01:35,400 Be teachers nurses warehouse workers and couriers 47 00:01:35,666 --> 00:01:37,800 Those with civil service pensions 48 00:01:37,800 --> 00:01:40,133 Life-saving money for those local city employees 49 00:01:40,166 --> 00:01:43,066 Everyone is filling this bottomless pit dug by the AI company. 50 00:01:43,366 --> 00:01:45,266 If this set of business logic is so fragile 51 00:01:45,300 --> 00:01:46,966 Then what kind of power is suffering? 52 00:01:46,966 --> 00:01:49,333 Supporting this trillion-level AI carnival 53 00:01:49,500 --> 00:01:51,466 The answer is two words. 54 00:01:51,700 --> 00:01:52,766 More precisely 55 00:01:52,766 --> 00:01:54,300 called revolving financing 56 00:01:54,400 --> 00:01:56,366 Once you understand how to play 57 00:01:56,366 --> 00:01:57,166 you will think 58 00:01:57,166 --> 00:01:58,566 This is just like a legal 59 00:01:58,566 --> 00:02:00,100 Ponzi scheme is no different 60 00:02:00,200 --> 00:02:01,866 Let’s sort out the timeline first 61 00:02:02,066 --> 00:02:04,066 David camp, a partner at Sequoia Capital 62 00:02:04,066 --> 00:02:05,966 A question was thrown out before that made the entire technology circle 63 00:02:05,966 --> 00:02:07,866 The $600 Billion Problem with Fryers 64 00:02:07,900 --> 00:02:09,300 This account is actually very simple 65 00:02:09,300 --> 00:02:11,600 Look at how crazy Nvidia’s graphics cards are selling now. 66 00:02:11,700 --> 00:02:14,100 Then consider the cost of buying graphics cards to build a data center 67 00:02:14,100 --> 00:02:17,566 Include electricity, cooling fees and backup generator costs. 68 00:02:17,766 --> 00:02:19,200 Those base camps that buy computing power 69 00:02:19,300 --> 00:02:23,466 Such as Amazon AWS Microsoft Azure Google Cloud 70 00:02:23,566 --> 00:02:26,066 In order to recoup their capital and earn half of the gross profit 71 00:02:26,266 --> 00:02:28,766 The entire AI industry must have 6,000 employees per year 72 00:02:28,766 --> 00:02:30,566 Billions of dollars in real revenue 73 00:02:30,600 --> 00:02:32,566 But where is the revenue of 600 billion? 74 00:02:32,666 --> 00:02:34,133 In fact, the income of the entire industry is even 75 00:02:34,133 --> 00:02:35,533 Even a fraction of this number is enough. 76 00:02:35,700 --> 00:02:37,166 At present, it is only about 100 billion 77 00:02:37,466 --> 00:02:40,166 This huge gap in the middle is a super bubble 78 00:02:40,300 --> 00:02:41,666 In order to fill this gap 79 00:02:41,666 --> 00:02:43,200 Make the story not look so bad 80 00:02:43,466 --> 00:02:44,600 Silicon Valley giants invented 81 00:02:44,600 --> 00:02:46,600 A set of unique skills of left hand and right hand 82 00:02:46,700 --> 00:02:47,900 Let’s look at Nvidia first 83 00:02:48,066 --> 00:02:50,566 Just in this latest wave of operations in 2026 84 00:02:50,600 --> 00:02:52,066 NVIDIA almost smashed it 85 00:02:52,066 --> 00:02:54,900 $40 billion invested in AI infrastructure 86 00:02:55,133 --> 00:02:57,900 They invested $30 billion in Openai alone 87 00:02:58,000 --> 00:02:58,866 Not only that 88 00:02:58,866 --> 00:03:01,300 They also invested heavily in Anthropic 89 00:03:01,400 --> 00:03:02,866 Musk’s XAI 90 00:03:02,900 --> 00:03:05,533 Even those companies doing fiber optics and data centers 91 00:03:05,900 --> 00:03:06,800 On the surface 92 00:03:06,800 --> 00:03:09,133 This is the boss supporting the ecology, right? 93 00:03:09,133 --> 00:03:10,466 But you look deeper 94 00:03:10,500 --> 00:03:12,000 NVIDIA voted for open ai 95 00:03:12,000 --> 00:03:14,466 What did Open AI do with the 30 billion? 96 00:03:14,500 --> 00:03:17,266 I turned around and bought Nvidia’s next-generation AI chip. 97 00:03:17,566 --> 00:03:19,766 Money was withdrawn from Nvidia's account 98 00:03:19,766 --> 00:03:21,100 Walked around outside 99 00:03:21,133 --> 00:03:23,000 Finally, in the name of sales revenue 100 00:03:23,000 --> 00:03:24,766 Back to Nvidia’s financial report 101 00:03:24,866 --> 00:03:27,300 Nvidia's revenue numbers look to continue to explode 102 00:03:27,400 --> 00:03:29,966 Openai also seems to have the endorsement of top funders 103 00:03:30,166 --> 00:03:31,933 Valuations on both sides took off immediately 104 00:03:32,133 --> 00:03:33,333 in this whole process 105 00:03:33,333 --> 00:03:34,666 There is not even one real 106 00:03:34,666 --> 00:03:36,400 of end consumers pay for 107 00:03:36,400 --> 00:03:37,933 But the data is beautiful 108 00:03:37,966 --> 00:03:38,933 Let’s look at Microsoft again 109 00:03:39,100 --> 00:03:42,100 Microsoft has spent tens of billions on open AI over the years 110 00:03:42,100 --> 00:03:43,466 Even tens of billions of dollars 111 00:03:43,466 --> 00:03:45,100 But there's an open secret 112 00:03:45,200 --> 00:03:47,200 Most of this money isn't cash at all 113 00:03:47,266 --> 00:03:49,400 It’s a voucher for Azure cloud services 114 00:03:49,500 --> 00:03:50,866 This is equivalent to Microsoft giving 115 00:03:50,866 --> 00:03:52,766 Openai a huge gift card 116 00:03:52,766 --> 00:03:55,266 This card can only be used in Microsoft stores 117 00:03:55,566 --> 00:03:57,266 open AI can calculate this money 118 00:03:57,266 --> 00:03:58,700 into the capital you raise 119 00:03:58,700 --> 00:04:00,733 Bragging to the outside world that his valuation has increased again 120 00:04:00,733 --> 00:04:02,266 But you can’t use cloud services 121 00:04:02,266 --> 00:04:04,000 Vouchers are used to pay employees 122 00:04:04,200 --> 00:04:06,666 You Can't Pay Silicon Valley's Expensive Office Rent 123 00:04:06,666 --> 00:04:07,866 Can't afford lawyer's fees 124 00:04:08,066 --> 00:04:10,200 So the real cash flow is very little. 125 00:04:10,266 --> 00:04:11,666 And Microsoft is very thief 126 00:04:11,700 --> 00:04:13,300 In the Openai protocol 127 00:04:13,300 --> 00:04:15,166 Once Openai starts making money 128 00:04:15,166 --> 00:04:18,533 75% of the early profits must be returned to Microsoft first 129 00:04:18,566 --> 00:04:20,100 Until Microsoft recovers its capital 130 00:04:20,133 --> 00:04:22,300 This proportion dropped to 49% 131 00:04:22,533 --> 00:04:23,300 To put it bluntly 132 00:04:23,300 --> 00:04:25,966 Microsoft just lends its computing power as money 133 00:04:25,966 --> 00:04:27,400 Then treat this fake money as 134 00:04:27,400 --> 00:04:28,733 Own cloud service revenue 135 00:04:28,733 --> 00:04:29,800 Include it in the financial report 136 00:04:29,800 --> 00:04:31,400 This is not just revolving financing 137 00:04:31,400 --> 00:04:33,766 Even the quality of the revenue is watered down. 138 00:04:33,766 --> 00:04:36,166 If it only stays between two companies 139 00:04:36,200 --> 00:04:38,066 That's not considered systemic risk. 140 00:04:38,133 --> 00:04:40,300 But now this set of serial sets has 141 00:04:40,300 --> 00:04:42,133 Most of the tech world is tied into it. 142 00:04:42,266 --> 00:04:43,700 In addition to open AI and Microsoft 143 00:04:43,700 --> 00:04:45,200 Let’s take a look at what was just launched this year 144 00:04:45,300 --> 00:04:47,800 Corwave’s valuation has soared past $50 billion 145 00:04:47,900 --> 00:04:50,600 This is a company that specializes in AI cloud computing power 146 00:04:50,966 --> 00:04:52,933 Nvidia is not only a shareholder of Coralwave 147 00:04:52,966 --> 00:04:55,100 Recently, it was added in early 2026 148 00:04:55,100 --> 00:04:56,700 $2 billion investment 149 00:04:56,800 --> 00:04:59,866 Then NVIDIA has already signed an agreement with Coralwave 150 00:04:59,900 --> 00:05:02,000 Buy them before 2032 151 00:05:02,000 --> 00:05:03,500 More than $6 billion in services 152 00:05:03,666 --> 00:05:04,466 Not only that 153 00:05:04,566 --> 00:05:06,933 Who are the many big customers that Corwave has? 154 00:05:06,933 --> 00:05:09,700 Is Microsoft Microsoft taking over Corwave lately? 155 00:05:09,700 --> 00:05:11,700 2/3 of sales in a quarter 156 00:05:11,866 --> 00:05:13,133 And Microsoft itself is 157 00:05:13,133 --> 00:05:14,666 One of NVIDIA's largest customers 158 00:05:14,900 --> 00:05:16,100 As for open a i 159 00:05:16,133 --> 00:05:18,300 They still hold shares in AMD 160 00:05:18,466 --> 00:05:20,766 Oracle also signed a contract with open ai 161 00:05:20,766 --> 00:05:22,666 Billions of dollars in computing power 162 00:05:22,666 --> 00:05:24,300 Do you understand this endless cycle? 163 00:05:24,300 --> 00:05:28,766 Technology giant cloud service provider chip factory AI startup 164 00:05:29,066 --> 00:05:30,400 They form a closed loop 165 00:05:30,700 --> 00:05:32,200 a invests money into b 166 00:05:32,300 --> 00:05:34,166 b uses money to buy c’s server 167 00:05:34,300 --> 00:05:36,166 c then buy a’s chip 168 00:05:36,366 --> 00:05:38,733 Everyone is buying each other's services 169 00:05:38,866 --> 00:05:40,766 Boost each other’s valuations 170 00:05:40,933 --> 00:05:43,066 As long as the money in this circle is still circulating 171 00:05:43,100 --> 00:05:45,000 Everyone’s revenue will double 172 00:05:45,133 --> 00:05:47,266 Super unicorns with valuations exceeding one trillion 173 00:05:47,566 --> 00:05:50,333 Morningstar analysts recently broke the news 174 00:05:50,400 --> 00:05:53,066 As the AI industry prepares for IPO 175 00:05:53,133 --> 00:05:55,466 This kind of circular trading is the most deadly minefield 176 00:05:55,733 --> 00:05:57,600 Because this is not an ordinary debt relationship 177 00:05:57,666 --> 00:05:58,866 General business loans 178 00:05:58,866 --> 00:06:00,066 you owe the bank money 179 00:06:00,066 --> 00:06:01,300 The bank is just a creditor 180 00:06:01,300 --> 00:06:02,333 But in the AI circle 181 00:06:02,333 --> 00:06:03,933 The investor who writes you the check 182 00:06:03,933 --> 00:06:06,766 At the same time, it is also the supplier who asks you for computing power fees. 183 00:06:06,866 --> 00:06:09,400 Once Openai or other major 184 00:06:09,400 --> 00:06:11,066 Model company’s revenue is not up to standard 185 00:06:11,166 --> 00:06:13,700 Can’t come up with money to pay for future computing power contracts 186 00:06:13,733 --> 00:06:15,600 No one among these giants can escape 187 00:06:15,766 --> 00:06:17,200 when the bubble bursts 188 00:06:17,200 --> 00:06:19,566 They can't even find a partner to share the risk with. 189 00:06:19,566 --> 00:06:21,366 Because everyone is on the same rope 190 00:06:21,466 --> 00:06:22,466 What's even crazier is 191 00:06:22,466 --> 00:06:23,900 In order to hide the true debt 192 00:06:24,000 --> 00:06:25,700 The methods of the giants are extremely ruthless 193 00:06:25,866 --> 00:06:28,133 For example, Meta’s operation at the end of 2025 194 00:06:28,133 --> 00:06:30,100 Directly issued US$30 billion in corporate bonds 195 00:06:30,400 --> 00:06:32,766 At the same time, through a structured joint venture with Morgan Stanley 196 00:06:32,766 --> 00:06:36,100 Companies took on another $30 billion in off-balance sheet debt 197 00:06:36,200 --> 00:06:37,366 What is off-balance sheet debt 198 00:06:37,366 --> 00:06:39,766 It is an account that is not written on the balance sheet. 199 00:06:39,966 --> 00:06:41,366 why are they hiding 200 00:06:41,366 --> 00:06:43,333 Because AI infrastructure is a bottomless pit 201 00:06:43,500 --> 00:06:45,300 If they were all spread out in the sun 202 00:06:45,300 --> 00:06:47,000 Investors will be scared away immediately 203 00:06:47,000 --> 00:06:48,666 The nature of this circular financing 204 00:06:48,933 --> 00:06:51,200 It’s just that there’s not much real money on the table. 205 00:06:51,466 --> 00:06:53,000 It's all a lot I owe you 206 00:06:53,000 --> 00:06:54,366 You owe me a piece of paper 207 00:06:54,500 --> 00:06:56,200 Several giants are making excuses for each other 208 00:06:56,300 --> 00:06:57,600 The bet is that one day in the future 209 00:06:57,666 --> 00:06:59,566 Real consumers or businesses can 210 00:06:59,566 --> 00:07:01,500 Enter the market with tens of trillions of real money 211 00:07:01,700 --> 00:07:03,300 Cashed in all these IOUs 212 00:07:03,300 --> 00:07:04,900 before this illusion is shattered 213 00:07:04,900 --> 00:07:07,166 The revenue on the financial report will look amazing 214 00:07:07,200 --> 00:07:09,100 But as long as one company can't stand it 215 00:07:09,100 --> 00:07:11,666 Or maybe everyone suddenly wants to convert their white notes into cash 216 00:07:11,700 --> 00:07:15,166 This game of drumming and passing flowers will collapse in an instant 217 00:07:15,533 --> 00:07:18,000 Morgan Stanley’s forecast has already made it clear 218 00:07:18,000 --> 00:07:20,066 From 2025 to 2028 219 00:07:20,066 --> 00:07:21,666 Global spending on data centers 220 00:07:21,666 --> 00:07:23,666 The money will be as high as 3 trillion US dollars 221 00:07:23,700 --> 00:07:25,133 800 billion of this is due to 222 00:07:25,133 --> 00:07:26,666 private credit financing 223 00:07:26,666 --> 00:07:28,066 This money is not from tech companies 224 00:07:28,066 --> 00:07:29,733 Profit earned from selling products 225 00:07:29,733 --> 00:07:30,800 But a borrowed debt 226 00:07:31,100 --> 00:07:34,933 Using investors’ money to buy equipment from other companies 227 00:07:35,133 --> 00:07:38,333 So if there are so many Big Macs supporting us, 228 00:07:38,333 --> 00:07:40,733 You may think that an industry like Openai 229 00:07:40,733 --> 00:07:43,800 You must be so poor that all you have left is money, right? 230 00:07:43,900 --> 00:07:45,133 The truth is just the opposite 231 00:07:45,133 --> 00:07:46,866 Take a look at the latest financial data 232 00:07:46,933 --> 00:07:48,400 It just makes my scalp numb. 233 00:07:48,700 --> 00:07:52,166 Open AI revenue will reach US$3.7 billion in 2024 234 00:07:52,266 --> 00:07:53,533 Loss of $5 billion 235 00:07:53,600 --> 00:07:55,100 By the end of 2025 236 00:07:55,100 --> 00:07:56,666 They brag about their annual 237 00:07:56,666 --> 00:07:58,600 Revenue reached $20 billion 238 00:07:58,666 --> 00:07:59,800 sounds scary 239 00:07:59,800 --> 00:08:01,166 increased more than 3 times 240 00:08:01,166 --> 00:08:02,466 but at what cost 241 00:08:02,600 --> 00:08:04,600 The price is that the black hole of loss is getting bigger and bigger. 242 00:08:04,666 --> 00:08:07,133 Microsoft's SEC filings reveal the truth 243 00:08:07,133 --> 00:08:09,933 Only in the quarter from July to September 2025 244 00:08:10,000 --> 00:08:12,200 Openai lost $12 billion 245 00:08:12,333 --> 00:08:14,533 Their own internal financial forecasts are even more outrageous 246 00:08:14,700 --> 00:08:18,133 A net loss of US$14 billion is expected in 2026. 247 00:08:18,400 --> 00:08:20,366 From 2023 to 2028 248 00:08:20,400 --> 00:08:23,266 Cumulative losses will reach US$44 billion 249 00:08:23,300 --> 00:08:25,066 Some investment banks are even more ruthless 250 00:08:25,133 --> 00:08:26,333 Deutsche Bank thinks 251 00:08:26,333 --> 00:08:27,933 before they actually make a profit 252 00:08:28,000 --> 00:08:31,266 Cash flow hole will hit $143 billion 253 00:08:31,566 --> 00:08:32,766 human business history 254 00:08:32,866 --> 00:08:34,966 No startup has ever dared 255 00:08:34,966 --> 00:08:37,066 Without a mature business model 256 00:08:37,133 --> 00:08:38,200 Losing like this 257 00:08:38,366 --> 00:08:39,966 Uber burned money to grab the market 258 00:08:39,966 --> 00:08:42,200 After all, what we are grabbing is a real share of taxi rides. 259 00:08:42,200 --> 00:08:45,200 But Openai is currently burning money doing fill-in-the-blank questions on computing power. 260 00:08:45,466 --> 00:08:46,333 What's more terrible is 261 00:08:46,333 --> 00:08:48,266 Although chatgpt has 800 million users 262 00:08:48,266 --> 00:08:51,500 But only 5% are really willing to pay. 263 00:08:51,566 --> 00:08:54,700 The remaining 95% is all spent on expensive computing power. 264 00:08:54,733 --> 00:08:58,100 And it’s different from Google, Meta, Amazon 265 00:08:58,200 --> 00:08:59,800 Openai has no retreat 266 00:09:00,000 --> 00:09:01,066 If Google doesn’t do AI 267 00:09:01,066 --> 00:09:02,400 There are also search ads to cover up 268 00:09:02,533 --> 00:09:04,400 Amazon also has e-commerce and AWS 269 00:09:04,400 --> 00:09:06,366 Meta has an Instagram advertising empire 270 00:09:06,366 --> 00:09:07,933 And Openai only has one life 271 00:09:07,933 --> 00:09:08,866 Just a big model 272 00:09:08,900 --> 00:09:10,500 That's why Sam Altman 273 00:09:10,500 --> 00:09:12,200 Looking for money outside all day long 274 00:09:12,200 --> 00:09:14,166 Even looking for sovereign funds in the Middle East 275 00:09:14,466 --> 00:09:16,266 Because as soon as this machine stops 276 00:09:16,266 --> 00:09:17,266 No more money 277 00:09:17,266 --> 00:09:18,300 Will die soon 278 00:09:18,300 --> 00:09:19,666 They have even been forced to start 279 00:09:19,666 --> 00:09:21,366 Are you considering adding ads to your products? 280 00:09:21,666 --> 00:09:24,166 And this was the worst option Ultraman himself ever said. 281 00:09:24,333 --> 00:09:26,333 So what is now in front of the market 282 00:09:26,333 --> 00:09:28,100 It's an extremely deformed monster 283 00:09:28,100 --> 00:09:30,500 On one side are the giants with their left hand and their right hand 284 00:09:30,500 --> 00:09:33,600 Use staggering capital expenditures to support gorgeous financial reports 285 00:09:33,866 --> 00:09:36,000 On the other side, real losses are occurring in the form of 286 00:09:36,000 --> 00:09:38,100 Losing blood at a rate of tens of millions of dollars every day 287 00:09:38,366 --> 00:09:40,466 Since these giants are spending money like crazy 288 00:09:40,466 --> 00:09:42,333 Even the financial reports are so magical 289 00:09:42,400 --> 00:09:43,366 That stands to reason 290 00:09:43,366 --> 00:09:45,900 Thousands of billions of dollars are spent on this 291 00:09:45,900 --> 00:09:48,000 You have to listen to it on the enterprise side. 292 00:09:48,066 --> 00:09:49,666 For example, productivity has been greatly improved 293 00:09:49,666 --> 00:09:51,566 Or directly replace half of the employees 294 00:09:51,566 --> 00:09:53,600 Saves companies huge costs 295 00:09:53,800 --> 00:09:54,700 You think too much 296 00:09:54,700 --> 00:09:57,500 This leads to the most terrible point in the entire AI carnival 297 00:09:57,500 --> 00:09:59,266 It is also the most shameful data 298 00:09:59,300 --> 00:10:01,500 Back to August 2025 299 00:10:01,533 --> 00:10:03,533 There is one that should have directly dominated the list 300 00:10:03,533 --> 00:10:05,766 Global financial headlines report released 301 00:10:05,766 --> 00:10:06,866 But the strange thing is 302 00:10:06,866 --> 00:10:08,900 The mainstream media said nothing about it 303 00:10:09,133 --> 00:10:12,100 This report tracks hundreds of large companies in depth 304 00:10:12,133 --> 00:10:14,800 These companies have spent a total of money on generative AI 305 00:10:14,800 --> 00:10:17,566 $30-40 billion in real money 306 00:10:17,700 --> 00:10:18,933 Guess what the result will be? 307 00:10:19,133 --> 00:10:21,200 95% of the surveyed companies stated that 308 00:10:21,266 --> 00:10:23,566 Their investment in AI is not seen 309 00:10:23,566 --> 00:10:25,400 Any measurable return on investment 310 00:10:26,100 --> 00:10:27,166 Don't rush to refute 311 00:10:27,300 --> 00:10:29,666 I know you usually use a large model to write a weekly report 312 00:10:29,733 --> 00:10:31,266 It feels great to translate an email 313 00:10:31,333 --> 00:10:33,166 But from the perspective of a business owner 314 00:10:33,400 --> 00:10:34,966 They are calculating another account 315 00:10:35,133 --> 00:10:37,666 Spend a lot of money to buy the enterprise version of AI service 316 00:10:37,700 --> 00:10:39,166 It turns out that this thing is basically 317 00:10:39,166 --> 00:10:40,700 Cannot completely replace labor 318 00:10:40,966 --> 00:10:43,100 It always talks serious nonsense 319 00:10:43,100 --> 00:10:44,800 This is the so-called AI illusion. 320 00:10:45,066 --> 00:10:46,900 What is the actual usage of the enterprise? 321 00:10:46,900 --> 00:10:49,366 An employee uses AI to generate a draft 322 00:10:49,500 --> 00:10:51,100 Then you have to arrange for someone to go there 323 00:10:51,100 --> 00:10:54,133 Manual review and modification line by line 324 00:10:54,166 --> 00:10:55,566 Finally apply it manually 325 00:10:55,666 --> 00:10:57,366 It does speed things up a little bit 326 00:10:57,366 --> 00:10:58,733 But in the end it works 327 00:10:58,800 --> 00:11:00,000 People are still responsible 328 00:11:00,166 --> 00:11:01,600 The boss paid for the AI 329 00:11:01,600 --> 00:11:03,400 But the corresponding employee costs cannot be cut 330 00:11:03,400 --> 00:11:04,500 Can't come down at all 331 00:11:04,500 --> 00:11:05,800 U.S. Census Bureau data 332 00:11:05,800 --> 00:11:07,400 He even tore off the fig leaf directly. 333 00:11:07,466 --> 00:11:09,400 AI adoption rates among large enterprises 334 00:11:09,400 --> 00:11:11,600 In fact, it will already peak in 2025 335 00:11:11,766 --> 00:11:13,733 There has even begun to show a downward trend 336 00:11:13,966 --> 00:11:15,466 Everyone found out after trying it 337 00:11:15,466 --> 00:11:17,333 This thing is a money-burning gold-eating beast. 338 00:11:17,400 --> 00:11:19,400 It would be better to recruit a few interns 339 00:11:19,566 --> 00:11:22,300 This forms an extremely ridiculous business trap 340 00:11:22,600 --> 00:11:23,800 traditional software industry 341 00:11:23,800 --> 00:11:25,166 For example, Microsoft sells office 342 00:11:25,400 --> 00:11:27,200 Or Adobe sells photo editing software 343 00:11:27,333 --> 00:11:28,733 Development costs are fixed 344 00:11:28,733 --> 00:11:30,100 Sell one more license 345 00:11:30,200 --> 00:11:31,466 Cost is almost zero 346 00:11:31,466 --> 00:11:32,500 It's all pure profit 347 00:11:32,733 --> 00:11:34,266 But the large AI model is different 348 00:11:34,333 --> 00:11:36,500 It is not a one-time development and done 349 00:11:36,500 --> 00:11:39,133 As long as the user presses the Enter key in the dialog box 350 00:11:39,166 --> 00:11:41,900 The back-end GPU pool has to be counted once. 351 00:11:42,100 --> 00:11:43,400 Computing power is the cost of electricity 352 00:11:43,533 --> 00:11:45,000 Computing power is the depreciation of the chip 353 00:11:45,133 --> 00:11:47,000 This leads to a fatal knot 354 00:11:47,300 --> 00:11:48,800 AI company products become more popular 355 00:11:48,800 --> 00:11:49,733 The more people use it 356 00:11:49,733 --> 00:11:51,166 The more money the company loses 357 00:11:51,466 --> 00:11:53,566 Faced with this situation where the more you sell, the more you lose. 358 00:11:53,566 --> 00:11:56,066 How do these big AI companies write their scripts? 359 00:11:56,100 --> 00:11:57,766 It’s very easy to draw a pie 360 00:11:57,933 --> 00:11:59,533 They deceived investors and the entire 361 00:11:59,533 --> 00:12:00,866 The logic of the market is like this 362 00:12:00,866 --> 00:12:01,766 Don't panic, everyone 363 00:12:01,900 --> 00:12:04,000 We are losing money now because computing power is too expensive 364 00:12:04,266 --> 00:12:05,966 As long as we keep throwing money at chips 365 00:12:06,100 --> 00:12:08,966 In the future, the cost of computing power will definitely decrease exponentially. 366 00:12:09,066 --> 00:12:09,966 until then 367 00:12:10,000 --> 00:12:12,600 Our income will exceed the rate of burning money 368 00:12:12,666 --> 00:12:13,533 achieve profitability 369 00:12:13,866 --> 00:12:16,000 After all, when the dot-com bubble burst, 370 00:12:16,066 --> 00:12:18,166 Isn’t Amazon also surviving despite losses? 371 00:12:18,166 --> 00:12:19,766 Did it finally become a trillion-dollar empire? 372 00:12:19,766 --> 00:12:21,100 That sounds fine 373 00:12:21,100 --> 00:12:22,600 But this rhetoric conceals a 374 00:12:22,600 --> 00:12:24,533 The extremely scary macroeconomic truth 375 00:12:24,733 --> 00:12:26,000 Economist at Harvard University 376 00:12:26,000 --> 00:12:28,900 Jason Foreman made a terrible calculation 377 00:12:29,166 --> 00:12:30,966 In the first half of 2025 378 00:12:31,166 --> 00:12:33,366 The growth of the entire U.S. GDP 379 00:12:33,366 --> 00:12:37,200 About 92% are directly related to AI 380 00:12:37,200 --> 00:12:39,133 Supported by infrastructure investment 381 00:12:39,300 --> 00:12:41,300 You heard it right 92% 382 00:12:41,333 --> 00:12:44,066 If you build a data center with AI and buy a graphics card 383 00:12:44,100 --> 00:12:45,966 All the money for pulling up the power grid will be deducted. 384 00:12:46,000 --> 00:12:47,700 The actual performance of the U.S. economy in that half year 385 00:12:47,700 --> 00:12:50,333 The growth rate is only a pitiful 0.1% 386 00:12:50,366 --> 00:12:51,366 what does this mean 387 00:12:51,400 --> 00:12:53,666 Explain the economic growth of the entire superpower 388 00:12:53,666 --> 00:12:56,466 The engine has relied heavily on a race without 389 00:12:56,466 --> 00:12:58,566 Purchasing sprees proven to be profitable 390 00:12:58,800 --> 00:13:00,966 We are using the resources of the whole society to 391 00:13:00,966 --> 00:13:03,066 For a company that doesn’t even have a business model 392 00:13:03,066 --> 00:13:05,100 Paying for the future of run-through startups 393 00:13:05,266 --> 00:13:07,400 Once this huge lie is exposed 394 00:13:07,466 --> 00:13:08,800 What will happen 395 00:13:08,900 --> 00:13:11,366 This is definitely not as simple as a few stocks falling to their limit. 396 00:13:11,566 --> 00:13:13,900 Oliver Wyman, a leading investment advisory firm 397 00:13:14,100 --> 00:13:16,933 Oliver Wyman has done extremely detailed model deductions 398 00:13:17,200 --> 00:13:18,666 If the AI bubble bursts 399 00:13:18,733 --> 00:13:21,000 The market will face two crash scenarios 400 00:13:21,100 --> 00:13:22,600 And each one is more tragic than the other 401 00:13:22,700 --> 00:13:25,533 The first is a relatively mild stock market carnage 402 00:13:25,666 --> 00:13:27,533 This is the so-called great shrinkage of rights and interests. 403 00:13:27,766 --> 00:13:28,766 The logic is very simple 404 00:13:28,866 --> 00:13:30,566 Capital’s patience is limited 405 00:13:30,566 --> 00:13:32,466 When Wall Street realizes AI is 406 00:13:32,466 --> 00:13:34,166 It won’t be possible in the short term. 407 00:13:34,266 --> 00:13:36,200 Or those cycles in financial reports 408 00:13:36,200 --> 00:13:38,266 When financing finally fails 409 00:13:38,333 --> 00:13:40,000 Investors will start selling 410 00:13:40,300 --> 00:13:41,166 We said before 411 00:13:41,166 --> 00:13:43,500 The S&P 500's returns over the past few years have been 412 00:13:43,500 --> 00:13:46,966 About 75% is contributed by AI-related concept stocks 413 00:13:46,966 --> 00:13:49,766 As long as the leading brothers like Nvidia and Microsoft fall, 414 00:13:49,933 --> 00:13:52,066 The entire market will be brought down in an instant 415 00:13:52,333 --> 00:13:54,300 What if the Internet happened again in 2,000 years? 416 00:13:54,300 --> 00:13:55,700 Bubble level sell-off 417 00:13:55,766 --> 00:13:56,766 what will happen 418 00:13:57,000 --> 00:13:58,666 American household wealth will directly 419 00:13:58,666 --> 00:14:00,400 More than $20 trillion evaporated 420 00:14:00,533 --> 00:14:01,866 on the books of overseas investors 421 00:14:01,866 --> 00:14:03,666 $15 trillion will be wiped out 422 00:14:04,000 --> 00:14:05,333 Based on current valuation 423 00:14:05,333 --> 00:14:06,866 The wealth created by this avalanche 424 00:14:06,866 --> 00:14:09,466 Losses could exceed $33 trillion 425 00:14:09,500 --> 00:14:10,800 This is more than the United States in a whole year 426 00:14:10,800 --> 00:14:12,666 GDP is even bigger 427 00:14:12,800 --> 00:14:16,866 Countless ordinary people will have their 401K pension accounts cut in half 428 00:14:17,000 --> 00:14:18,200 You worked hard 429 00:14:18,200 --> 00:14:19,933 A pension saved for half a lifetime 430 00:14:19,933 --> 00:14:22,866 Just a few big negative lines on Wall Street were wiped out. 431 00:14:23,166 --> 00:14:24,666 But that's not the scariest thing 432 00:14:24,666 --> 00:14:27,200 The second scenario is the real hell mode 433 00:14:27,333 --> 00:14:28,933 Debt-induced financial tsunami 434 00:14:29,100 --> 00:14:29,933 you need to know 435 00:14:29,966 --> 00:14:32,266 The Internet bubble of 2000 436 00:14:32,333 --> 00:14:35,066 The people who built the website mainly burned venture capital money. 437 00:14:35,066 --> 00:14:36,333 That is equity financing 438 00:14:36,733 --> 00:14:37,866 The company went bankrupt 439 00:14:37,866 --> 00:14:38,866 Investors admit defeat 440 00:14:39,100 --> 00:14:41,100 It would be miserable to just pat your butt and leave. 441 00:14:41,200 --> 00:14:43,166 But it is unlikely to be transmitted to the banking system 442 00:14:43,200 --> 00:14:45,466 But this time the AI bubble is completely different 443 00:14:45,500 --> 00:14:47,533 It was piled up with huge amounts of debt 444 00:14:47,666 --> 00:14:49,066 Remember Morgan Stanley earlier? 445 00:14:49,066 --> 00:14:51,133 That $800 billion in private credit mentioned? 446 00:14:51,200 --> 00:14:53,800 Except for those Wall Street giants who can be named 447 00:14:53,900 --> 00:14:55,666 There are also a large number of small and medium-sized banks 448 00:14:55,666 --> 00:14:58,133 Local banks also came into trouble. 449 00:14:58,166 --> 00:15:01,666 They lend AI to various start-ups to rent computing power 450 00:15:01,666 --> 00:15:03,200 Or take these high-risk technologies 451 00:15:03,200 --> 00:15:05,300 The company regards it as its core depositor 452 00:15:05,600 --> 00:15:06,700 if the bubble bursts 453 00:15:06,733 --> 00:15:09,066 Once these AI companies stop burning money 454 00:15:09,066 --> 00:15:10,800 Can’t afford expensive computing power rent 455 00:15:10,900 --> 00:15:12,566 Those who loaned them money to buy GPUs 456 00:15:12,566 --> 00:15:14,200 Financial institutions will explode instantly 457 00:15:14,500 --> 00:15:16,933 Since the asset side is highly depreciated second-hand chips, 458 00:15:17,000 --> 00:15:18,933 These banks simply cannot recoup their costs 459 00:15:19,066 --> 00:15:20,266 As soon as bad debts come out 460 00:15:20,333 --> 00:15:22,666 The script of Silicon Valley bank collapse will repeat itself 461 00:15:22,900 --> 00:15:25,066 And don’t forget this is the age of social media 462 00:15:25,300 --> 00:15:26,733 Panic spreads online 463 00:15:26,733 --> 00:15:28,400 The speed is measured in seconds 464 00:15:28,400 --> 00:15:30,766 In 2008 you had to queue up to withdraw money from the bank 465 00:15:30,766 --> 00:15:32,100 Now just click on your phone 466 00:15:32,100 --> 00:15:34,133 Deposits can be transferred in two clicks 467 00:15:34,333 --> 00:15:36,733 Bank runs that used to take months to develop 468 00:15:36,866 --> 00:15:38,400 It only takes a few hours now 469 00:15:38,566 --> 00:15:40,133 Wall Street's Famous Big Short 470 00:15:40,266 --> 00:15:42,766 The original character from the big short movie 471 00:15:42,766 --> 00:15:44,900 Michael Burry Michael Burry 472 00:15:45,000 --> 00:15:47,500 That is, the one who was involved in the subprime mortgage crisis in 2008 473 00:15:47,500 --> 00:15:50,000 The ruthless person who shorted the real estate market before the outbreak 474 00:15:50,200 --> 00:15:52,400 He recently issued a very stern warning 475 00:15:52,600 --> 00:15:53,466 Bury thinks 476 00:15:53,500 --> 00:15:55,200 If the AI bubble starts to collapse 477 00:15:55,400 --> 00:15:57,466 The U.S. government will definitely act like a bailout 478 00:15:57,466 --> 00:15:59,200 Introduce various policies to support 479 00:15:59,266 --> 00:16:00,966 Desperately trying to save this bubble 480 00:16:01,166 --> 00:16:02,100 But the problem is 481 00:16:02,100 --> 00:16:04,000 This time is different from 2008 482 00:16:04,066 --> 00:16:06,133 This time the financial hole is being filled by technology giants 483 00:16:06,133 --> 00:16:08,933 Revolving financing and huge amounts of debt are too big 484 00:16:09,066 --> 00:16:11,566 So big that even the Federal Reserve’s money printing presses can’t be filled. 485 00:16:11,666 --> 00:16:13,533 When 3 trillion or even tens of trillions 486 00:16:13,533 --> 00:16:15,333 The wealth myth collapsed in a matter of months 487 00:16:15,466 --> 00:16:17,266 Those made with the left hand instead of the right hand 488 00:16:17,266 --> 00:16:18,966 The revenue ledger was completely shredded 489 00:16:19,400 --> 00:16:20,966 That's never what really bleeds 490 00:16:20,966 --> 00:16:22,800 Several technology tycoons who cashed out in advance 491 00:16:22,866 --> 00:16:24,500 Behind these astronomical figures 492 00:16:24,500 --> 00:16:26,666 are more than 2.5 million native Americans 493 00:16:26,666 --> 00:16:28,533 Jobs under direct threat 494 00:16:28,866 --> 00:16:30,566 The money disappeared into ashes 495 00:16:30,566 --> 00:16:32,766 But who will clean up the mess? 496 00:16:33,133 --> 00:16:35,133 To satisfy the greedy appetite of AI 497 00:16:35,133 --> 00:16:36,866 And huge infrastructure buildings are rising from the ground 498 00:16:36,900 --> 00:16:38,666 It is the core of the next crisis 499 00:16:38,700 --> 00:16:40,000 This involves bursting the bubble 500 00:16:40,000 --> 00:16:41,600 The real economy is in disaster after 501 00:16:41,600 --> 00:16:43,600 Those who are crazy about AI 502 00:16:43,600 --> 00:16:45,900 Nuclear power plants, natural gas power plants and 503 00:16:45,900 --> 00:16:48,400 A data center that covers an area larger than a town 504 00:16:48,700 --> 00:16:49,933 Not only will it not disappear 505 00:16:50,000 --> 00:16:51,866 Instead, it will become a set of ordinary 506 00:16:51,866 --> 00:16:53,800 The heavy shackles on people's necks 507 00:16:54,066 --> 00:16:55,466 Many people naively think 508 00:16:55,466 --> 00:16:57,733 The big model is just a few lines of code on the Internet 509 00:16:57,800 --> 00:17:00,066 When the bubble bursts, the server shuts down. 510 00:17:00,166 --> 00:17:01,966 Programmers change projects and continue working on them 511 00:17:02,000 --> 00:17:03,733 But the reality is not like that at all 512 00:17:04,000 --> 00:17:05,200 Although technology giants are 513 00:17:05,200 --> 00:17:07,366 On the book, the left hand is playing with the right hand 514 00:17:07,466 --> 00:17:09,333 A game of selling computing power to each other 515 00:17:09,500 --> 00:17:10,733 But the chip in the physical world 516 00:17:10,733 --> 00:17:12,200 You can’t fake the electricity bill. 517 00:17:12,400 --> 00:17:13,800 You have to have real money 518 00:17:13,800 --> 00:17:15,666 Hardware and constant power 519 00:17:15,800 --> 00:17:17,966 Only in this way can these gold-eating beasts be able to move. 520 00:17:17,966 --> 00:17:20,733 To maintain the myth that AI can always evolve 521 00:17:20,733 --> 00:17:23,133 These giants have completely become red-eyed 522 00:17:23,200 --> 00:17:25,066 It even started to compete with basic energy sources in the United States. 523 00:17:25,066 --> 00:17:26,800 Lifeline signed a contract of betrayal for decades 524 00:17:27,133 --> 00:17:29,200 Let’s take a look at what the giants are doing recently 525 00:17:29,200 --> 00:17:30,200 crazy move 526 00:17:30,366 --> 00:17:32,333 Microsoft to power its data centers 527 00:17:32,366 --> 00:17:34,766 Directly signed a long-term power purchase agreement 528 00:17:34,966 --> 00:17:36,166 Actually need to restart 529 00:17:36,166 --> 00:17:38,533 Three Mile Island Nuclear Power Plant in Pennsylvania 530 00:17:38,900 --> 00:17:40,566 Anyone who knows a little bit about history knows that 531 00:17:40,566 --> 00:17:42,166 This place happened in the United States back then 532 00:17:42,166 --> 00:17:44,100 The worst nuclear accident in history 533 00:17:44,100 --> 00:17:45,600 But now in order to engage in AI 534 00:17:45,733 --> 00:17:47,066 Microsoft doesn't even have this level of old 535 00:17:47,066 --> 00:17:48,766 Nuclear power plants have to be re-ignited 536 00:17:49,100 --> 00:17:49,900 Not just Microsoft 537 00:17:49,933 --> 00:17:51,300 Google also signed an agreement 538 00:17:51,300 --> 00:17:54,766 To restart Iowa's Dwaine Arnold nuclear power plant 539 00:17:54,933 --> 00:17:56,800 open AI and Oracle are even more exaggerated 540 00:17:56,800 --> 00:17:58,933 They went straight to Abilene, Texas 541 00:17:59,000 --> 00:18:00,333 Installing like crazy 542 00:18:00,333 --> 00:18:02,400 1 gigawatt natural gas generating set 543 00:18:02,500 --> 00:18:04,000 A stargate just for them. 544 00:18:04,000 --> 00:18:05,600 Super computer park power supply 545 00:18:05,700 --> 00:18:07,266 The most outrageous thing is Meta Facebook 546 00:18:07,600 --> 00:18:08,933 they built it in ohio 547 00:18:08,933 --> 00:18:10,500 Prometheus Data Center 548 00:18:10,566 --> 00:18:12,366 Expected to go online in 2026 549 00:18:12,533 --> 00:18:14,900 It will directly consume 1 gigawatt of electricity. 550 00:18:15,100 --> 00:18:16,500 And this is just the appetizer 551 00:18:16,666 --> 00:18:18,733 The one Meta planned in Louisiana 552 00:18:18,733 --> 00:18:20,933 Super data center called Hyperion 553 00:18:20,966 --> 00:18:23,333 By the time it is fully operational in 2030 554 00:18:23,333 --> 00:18:26,133 Electricity consumption will reach an extremely terrifying 5 gigawatts 555 00:18:26,400 --> 00:18:27,933 What is the concept of 5 gigawatts? 556 00:18:27,966 --> 00:18:29,866 That's enough to sustain two cities of New Orleans 557 00:18:29,866 --> 00:18:31,700 All daily electricity use by residents and businesses 558 00:18:31,733 --> 00:18:33,400 A computer room of a technology company 559 00:18:33,533 --> 00:18:35,300 The power consumption is equivalent to that of two seats 560 00:18:35,300 --> 00:18:36,766 American cities with a population of one million 561 00:18:37,000 --> 00:18:39,066 In order to attract these giants to build factories 562 00:18:39,133 --> 00:18:40,266 U.S. local governments 563 00:18:40,266 --> 00:18:41,766 It’s also a tip-off that cost one’s life. 564 00:18:42,066 --> 00:18:43,600 There are currently 42 states in the United States 565 00:18:43,600 --> 00:18:45,933 You either exempt data centers from sales tax 566 00:18:45,933 --> 00:18:47,933 Either provide massive tax breaks 567 00:18:48,200 --> 00:18:50,400 Just 16 states that currently have public data 568 00:18:50,466 --> 00:18:51,533 in the past 5 years 569 00:18:51,533 --> 00:18:53,000 To please the tech giants 570 00:18:53,100 --> 00:18:55,933 $6 billion in forgone tax revenue alone 571 00:18:56,166 --> 00:18:57,466 The US$6 billion was originally 572 00:18:57,466 --> 00:18:58,933 Can be used to build public schools 573 00:18:59,100 --> 00:19:00,866 Repairing dilapidated highways 574 00:19:00,933 --> 00:19:03,566 Or paid to firefighters and police officers as wages 575 00:19:03,766 --> 00:19:06,733 But now all the money has been poured into that 576 00:19:06,733 --> 00:19:08,966 In data centers that look like air-raid shelters 577 00:19:08,966 --> 00:19:10,733 Go for an illusory one 578 00:19:10,733 --> 00:19:12,200 I don’t know when it will be realized 579 00:19:12,200 --> 00:19:13,933 of general artificial intelligence to pay for 580 00:19:13,933 --> 00:19:16,000 Today, when the bubble has not yet burst, 581 00:19:16,066 --> 00:19:17,333 Ordinary people have actually 582 00:19:17,333 --> 00:19:19,266 Start paying for this carnival 583 00:19:19,400 --> 00:19:21,400 What if the AI bubble bursts one day? 584 00:19:21,400 --> 00:19:24,366 Data centers are closing because they can't pay their electricity bills 585 00:19:24,466 --> 00:19:26,000 Did you think it was over? 586 00:19:26,000 --> 00:19:26,900 Not at all 587 00:19:27,200 --> 00:19:28,866 those that are rekindled 588 00:19:28,866 --> 00:19:30,566 Nuclear power plants will not disappear in a vacuum 589 00:19:30,733 --> 00:19:32,966 signed by tech giants and power companies 590 00:19:32,966 --> 00:19:35,333 Contracts written in black and white for decades will not be invalidated 591 00:19:35,700 --> 00:19:36,566 the only thing that disappeared 592 00:19:36,566 --> 00:19:38,466 It was originally built to fool everyone. 593 00:19:38,466 --> 00:19:40,500 Huge AI revenue from these things 594 00:19:40,500 --> 00:19:41,366 until then 595 00:19:41,466 --> 00:19:42,933 Risk-taking private technology 596 00:19:42,933 --> 00:19:44,800 Company files for bankruptcy 597 00:19:44,800 --> 00:19:47,400 Or be bailed out by Wall Street and the government 598 00:19:47,466 --> 00:19:49,733 And the sky-high infrastructure bills left behind 599 00:19:49,766 --> 00:19:52,200 Will pass the skyrocketing electricity bills across the board 600 00:19:52,333 --> 00:19:54,800 Forced distribution to every ordinary family 601 00:19:55,133 --> 00:19:57,733 Local residents may have to spend the next twenty or thirty years 602 00:19:57,733 --> 00:20:00,600 Pay dozens or even hundreds of dollars more in electricity bills every month 603 00:20:00,600 --> 00:20:03,133 Just to replace Meta or open AI back then 604 00:20:03,133 --> 00:20:05,133 The Bawang Power Grid contract signed to pay off debts 605 00:20:05,300 --> 00:20:06,100 At this time 606 00:20:06,100 --> 00:20:07,333 There must be die-hard Silicon Valley 607 00:20:07,333 --> 00:20:08,866 Fans are going to jump out and refute 608 00:20:08,900 --> 00:20:11,200 When the dot-com bubble burst in 2000 609 00:20:11,200 --> 00:20:12,766 Indeed, a large number of companies died 610 00:20:12,800 --> 00:20:14,666 But it also left a precious legacy 611 00:20:14,900 --> 00:20:16,966 Those bankrupt telecommunications companies that year laid 612 00:20:16,966 --> 00:20:19,066 Thousands of kilometers of fiber optic cable 613 00:20:19,166 --> 00:20:21,766 It is precisely because of these extremely cheap unfinished optical fibers 614 00:20:21,866 --> 00:20:24,566 Later, Youtube Netflix and 615 00:20:24,566 --> 00:20:26,600 Today’s extremely prosperous Internet ecosystem 616 00:20:26,933 --> 00:20:28,100 The AI bubble has burst 617 00:20:28,100 --> 00:20:28,933 Computing power price reduction 618 00:20:28,933 --> 00:20:30,700 Isn’t it just good enough to benefit future generations? 619 00:20:30,700 --> 00:20:32,333 This set of rhetoric can be said to be the entire 620 00:20:32,333 --> 00:20:34,333 The biggest lie in tech right now 621 00:20:34,466 --> 00:20:35,466 None 622 00:20:35,666 --> 00:20:38,200 AI hardware and fiber optics in the Internet era are inextricably linked 623 00:20:38,200 --> 00:20:40,133 There is a fatal difference in physical properties 624 00:20:40,400 --> 00:20:41,766 Optical fiber buried underground 625 00:20:41,766 --> 00:20:42,766 As long as you don't dig it up 626 00:20:42,766 --> 00:20:44,966 It will last for twenty or thirty years without any problems. 627 00:20:45,066 --> 00:20:46,766 Its value is extremely stable 628 00:20:46,766 --> 00:20:48,133 But what about AI chips? 629 00:20:48,133 --> 00:20:51,200 You spend tens of thousands of dollars to buy Nvidia’s most powerful GPU 630 00:20:51,333 --> 00:20:53,533 Just go through 1- 631 00:20:53,533 --> 00:20:55,666 3 years of high-intensity full load operation 632 00:20:55,733 --> 00:20:58,266 Its physical properties will undergo irreversible decay 633 00:20:58,566 --> 00:20:59,566 What's more terrible is 634 00:20:59,566 --> 00:21:01,933 The iteration speed of AI hardware is too fast 635 00:21:01,933 --> 00:21:04,300 When the next-generation architecture chip is released, 636 00:21:04,366 --> 00:21:06,800 Not only do older generation chips not have enough computing power to keep up, 637 00:21:06,800 --> 00:21:09,400 The energy consumption ratio was completely pushed to the ground and rubbed. 638 00:21:09,566 --> 00:21:11,066 A full 3 years ago top 639 00:21:11,066 --> 00:21:12,966 Luxurious data center with NVIDIA chips 640 00:21:12,966 --> 00:21:15,400 It is simply impossible to run the latest large models today 641 00:21:15,500 --> 00:21:17,366 No one wants to rent it out even at a reduced price 642 00:21:17,566 --> 00:21:18,933 Because I use it to run programs 643 00:21:18,933 --> 00:21:21,133 The electricity bill is more expensive than the chip itself 644 00:21:21,133 --> 00:21:22,766 Once the AI bubble bursts 645 00:21:22,800 --> 00:21:25,133 These cost tens of billions of dollars 646 00:21:25,133 --> 00:21:27,733 What’s left in the data center is nothing at all 647 00:21:27,733 --> 00:21:29,800 Digital highways for the next generation 648 00:21:29,866 --> 00:21:31,700 But a bunch of people with severe fever 649 00:21:31,733 --> 00:21:33,533 Rapidly Aging Electronic Waste 650 00:21:33,800 --> 00:21:36,166 Their economic value will instantly drop to zero. 651 00:21:36,166 --> 00:21:37,933 That's just a pile of scrap metal. 652 00:21:38,000 --> 00:21:40,266 Then this crash is destined to come 653 00:21:40,300 --> 00:21:42,366 How will it end? 654 00:21:42,533 --> 00:21:44,300 There’s a good chance it won’t be like 2008 655 00:21:44,300 --> 00:21:45,566 The collapse of Lehman Brothers 656 00:21:45,566 --> 00:21:47,266 The sky collapsed overnight 657 00:21:47,500 --> 00:21:51,166 Rather, it will be like a long and painful slow bloodletting. 658 00:21:51,566 --> 00:21:53,500 Referring to the experience of the Internet bubble that year 659 00:21:53,566 --> 00:21:55,166 The entire collapse may 660 00:21:55,166 --> 00:21:56,733 Lasts up to two years 661 00:21:56,800 --> 00:21:57,966 over time 662 00:21:58,200 --> 00:21:59,766 More and more venture capital institutions and 663 00:21:59,766 --> 00:22:01,800 Wall Street tycoons will suddenly wake up 664 00:22:01,900 --> 00:22:03,933 It turns out that AI won’t be able to win in the short term. 665 00:22:03,933 --> 00:22:05,700 Develop a decent business implementation model 666 00:22:05,966 --> 00:22:07,500 Once the donor fathers are unwilling to 667 00:22:07,500 --> 00:22:09,300 Paying for endless losses 668 00:22:09,366 --> 00:22:12,066 The break in the capital chain will spread from the bottom 669 00:22:12,366 --> 00:22:14,900 We can actually see signs of this now. 670 00:22:15,133 --> 00:22:17,766 Those AI application companies on the market that provide single functions 671 00:22:17,766 --> 00:22:19,200 Working on AI to write articles 672 00:22:19,200 --> 00:22:21,400 A start-up that makes AI schedules 673 00:22:21,466 --> 00:22:24,100 Why are they going bankrupt in droves? 674 00:22:24,400 --> 00:22:26,800 Because as long as they prove that there is a market for a certain feature 675 00:22:26,933 --> 00:22:28,600 Microsoft Google Apple 676 00:22:28,600 --> 00:22:31,066 These underlying large model providers will directly 677 00:22:31,066 --> 00:22:33,400 Integrate this function into your own system for free 678 00:22:33,400 --> 00:22:35,966 Crush startups to death overnight 679 00:22:36,133 --> 00:22:38,300 Even the current industry leader Openai 680 00:22:38,300 --> 00:22:40,800 Life only exists in the fantasy of PPT 681 00:22:40,933 --> 00:22:43,466 Their own internal financial forecast reports show 682 00:22:43,533 --> 00:22:46,400 The company will continue to suffer huge losses until 2028 683 00:22:46,533 --> 00:22:49,766 Theoretically, profitability will barely be achieved until 2030 684 00:22:50,000 --> 00:22:50,866 but please note 685 00:22:50,933 --> 00:22:53,133 The prerequisite for profit is extremely stringent 686 00:22:53,366 --> 00:22:55,700 Must be on schedule before 2030 687 00:22:55,700 --> 00:22:57,733 Be as smart as humans 688 00:22:57,766 --> 00:22:59,666 AGI that even surpasses humans 689 00:22:59,733 --> 00:23:01,533 This is not a business plan at all 690 00:23:01,766 --> 00:23:04,200 This is simply betting your life and fortune in a casino. 691 00:23:04,266 --> 00:23:06,900 Go bet on a world where even basic science 692 00:23:06,900 --> 00:23:09,366 A technological miracle whose principles are not fully understood yet 693 00:23:09,766 --> 00:23:11,733 If before this technological miracle comes 694 00:23:11,766 --> 00:23:13,666 What should I do if all the money in my account is burned first? 695 00:23:13,733 --> 00:23:16,666 The most realistic outcome is that the big fish eats the small fish 696 00:23:16,666 --> 00:23:17,600 winner takes all 697 00:23:17,600 --> 00:23:19,000 But ordinary people pay the bill 698 00:23:19,133 --> 00:23:21,466 If the funds of small and medium-sized AI companies are broken 699 00:23:21,566 --> 00:23:23,533 The ones who have the most confidence to survive are actually 700 00:23:23,533 --> 00:23:26,000 Microsoft, Google and Amazon 701 00:23:26,200 --> 00:23:28,733 Because they didn’t rely on AI at all to start their business. 702 00:23:28,866 --> 00:23:30,166 They have cloud computing in their hands 703 00:23:30,166 --> 00:23:31,266 There is a search engine 704 00:23:31,300 --> 00:23:33,000 There are e-commerce empires like these every year 705 00:23:33,000 --> 00:23:34,600 Stable cash flow for money printing 706 00:23:34,766 --> 00:23:36,166 They have hundreds of ways to escape 707 00:23:36,166 --> 00:23:38,266 And have a strong enough balance sheet 708 00:23:38,266 --> 00:23:40,166 You can do it when mourning floods the land 709 00:23:40,400 --> 00:23:43,333 Bring those desperate AIs to the market at extremely low prices 710 00:23:43,333 --> 00:23:46,466 All star teams and patents are purchased at low prices 711 00:23:46,533 --> 00:23:48,300 A vigorous AI revolution 712 00:23:48,300 --> 00:23:49,400 In the end it may just become 713 00:23:49,400 --> 00:23:50,900 A very boring ending 714 00:23:51,066 --> 00:23:52,733 Thousands of AI tools around the world 715 00:23:52,733 --> 00:23:54,366 and start-up projects all died 716 00:23:54,766 --> 00:23:57,466 All technology and data are eventually fully incorporated 717 00:23:57,600 --> 00:24:00,366 Become a division within these three giant companies 718 00:24:00,400 --> 00:24:01,933 or an ancillary service 719 00:24:02,300 --> 00:24:04,266 Giants successfully merged the market 720 00:24:04,266 --> 00:24:07,100 But those costs that were spread out during the carnival stage 721 00:24:07,166 --> 00:24:09,000 Most of it cannot be taken back. 722 00:24:09,000 --> 00:24:10,366 Those who have already signed 723 00:24:10,400 --> 00:24:12,933 Half-built power plants and transmission grids at sky-high prices 724 00:24:13,000 --> 00:24:14,333 The giants won't care 725 00:24:14,366 --> 00:24:17,266 The bill will be fixed on the electricity bill of local residents. 726 00:24:17,400 --> 00:24:19,533 Those investors who were fooled into entering the market at a high position 727 00:24:19,700 --> 00:24:22,000 Those who watched the S&P 500 surge and 728 00:24:22,000 --> 00:24:24,733 The complacent 401K pension account holder 729 00:24:24,933 --> 00:24:26,566 The wealth on your books has shrunk 730 00:24:26,566 --> 00:24:27,933 in Wall Street reports 731 00:24:27,933 --> 00:24:29,266 will only be treated lightly 732 00:24:29,266 --> 00:24:31,466 Recorded as systematic investment losses 733 00:24:31,566 --> 00:24:32,966 A multi-trillion dollar hole 734 00:24:32,966 --> 00:24:34,166 will be extremely accurate 735 00:24:34,166 --> 00:24:35,666 Allocated silently 736 00:24:35,666 --> 00:24:37,500 On the heads of tens of millions of ordinary Americans 737 00:24:37,733 --> 00:24:39,300 And most of you 738 00:24:39,333 --> 00:24:40,733 From beginning to end, there is a real 739 00:24:40,733 --> 00:24:42,566 I have never used any useful AI applications. 740 00:24:42,600 --> 00:24:45,300 Not to mention participating in this shocking gamble 741 00:24:45,500 --> 00:24:48,766 AI technology will definitely survive this crash 742 00:24:48,800 --> 00:24:50,400 and someday in the future 743 00:24:50,400 --> 00:24:52,066 it will be better funded 744 00:24:52,166 --> 00:24:53,666 The surviving giants 745 00:24:53,700 --> 00:24:57,166 In a slower, pragmatic, less bragging way 746 00:24:57,166 --> 00:24:58,466 Re-market 747 00:24:58,466 --> 00:24:59,600 But the problem is 748 00:24:59,733 --> 00:25:01,700 What Silicon Valley promises to all mankind today 749 00:25:01,700 --> 00:25:03,400 That wonderful world of science fiction 750 00:25:03,700 --> 00:25:05,266 and the one that will actually be implemented in the future 751 00:25:05,266 --> 00:25:07,133 Between the real world full of compromises 752 00:25:07,266 --> 00:25:08,800 across a huge abyss 753 00:25:08,966 --> 00:25:11,600 And the real money needed to fill this abyss 754 00:25:12,100 --> 00:25:14,333 In the end, everything is caused by the most powerful person in this society. 755 00:25:14,333 --> 00:25:16,933 Ordinary people who can’t afford the losses will foot the bill. 756 00:25:17,066 --> 00:25:18,766 So the current problem has long been 757 00:25:18,766 --> 00:25:20,933 Will the AI bubble burst? 758 00:25:20,933 --> 00:25:22,000 the real problem is 759 00:25:22,000 --> 00:25:23,266 When the banquet is over 760 00:25:23,266 --> 00:25:24,700 When there is nothing on the ground 761 00:25:24,800 --> 00:25:27,266 Who will be forced to the table? 762 00:25:27,333 --> 00:25:29,300 Pay that astronomical bill 763 00:25:29,600 --> 00:25:31,966 Who can only pick up among the ruins? 764 00:25:31,966 --> 00:25:34,200 Those pieces that can't be put together at all 765 00:25:34,466 --> 00:25:36,800 Openai is still losing blood like crazy 766 00:25:36,866 --> 00:25:39,300 But the reason why this dream machine stalled 767 00:25:39,366 --> 00:25:41,366 It's probably not what you think 768 00:25:41,400 --> 00:25:42,466 That’s it for today 769 00:25:42,466 --> 00:25:43,766 Friends, don’t forget to like it 770 00:25:43,766 --> 00:25:44,900 thank you for watching 771 00:25:44,900 --> 00:25:46,166 See you next time