Transcriptly

China Doesn't Need Better Al to Beat America | Stanford China Researcher, Dan Wang

EO16:56Added Aug 31, 2026

Dan Wang, author of Breakneck and research fellow at Stanford's Hoover Institution, explains why China's engineering-driven system is outbuilding America at ...

Watch on YouTube →
Contributed by 刘嘉琪

Transcript

Transcript format

  1. Intro

  2. 00:00Chinese are, I think, far ahead of the
  3. 00:02Americans. Last year, the United States
  4. 00:05built around five ships. China built
  5. 00:07about 1,500 ships. It takes on average
  6. 00:10five or 6 [music] years for an American
  7. 00:12automaker to think of a new model and
  8. 00:15then release that model onto the
  9. 00:17streets. In China, the speed of
  10. 00:18iteration is much closer to 18 [music]
  11. 00:20months, moving three times faster than
  12. 00:22the Americans. Last year, China built
  13. 00:24about 300 gawatt [music] of solar power
  14. 00:26alone. US built about 30 gawatt. China
  15. 00:29right now has 40 nuclear power plants
  16. 00:31under construction. The United States
  17. 00:33has zero. Who is going to be a clear
  18. 00:35winner? Right now, it looks like the US
  19. 00:38and China are both very intent on
  20. 00:41losing. Both these countries making very
  21. 00:43big mistakes, self- beatings, and try to
  22. 00:46hobble themselves. My name is Dan Wong.
  23. 00:49I am the author of Breakneck, China's
  24. 00:51[music] Quest to Engineer the Future.
  25. 00:53I'm also a research fellow at Stanford
  26. 00:55Tuver Institution. For six years in
  27. 00:57China, I was a technology analyst at an
  28. 01:00[music] economic research firm called
  29. 01:02Goffcow Dragons. And what I was really
  30. 01:05trained to [music] do was to uh think
  31. 01:07about China through the lens of
  32. 01:09political economy um everything that it
  33. 01:11wants to achieve especially [music] in
  34. 01:13the realm of technology.
  35. 01:25Open AI just launched the most powerful

  36. Does the Future Belong to China? - America's Biggest Mistake

  37. 01:27AI model.
  38. 01:28Elon Musk announces Gro is the best AI.
  39. 01:30Google claims their new model says
  40. 01:32they've built the smartest AI ever made.
  41. 01:34The United States can build and maintain
  42. 01:36the largest, most powerful and most
  43. 01:39advanced AI infrastructure.
  44. 01:40I think that there's no doubt that the
  45. 01:42US is ahead of China on artificial
  46. 01:45intelligence and the only question is
  47. 01:47whether the US has a very substantial
  48. 01:49lead or only a moderate lead. And my
  49. 01:52view is that the US has at best a
  50. 01:55moderate lead over China and it is not
  51. 01:57so much of a decisive lead. We're
  52. 01:59sitting here in Silicon Valley which is
  53. 02:02very capable of magical thinking. But I
  54. 02:04think that the Silicon Valley sort of
  55. 02:06magical thinking does not translate very
  56. 02:08well to Washington DC which engages in
  57. 02:10all sorts of magical thinking about how
  58. 02:12it could outco compete China if it could
  59. 02:15only have somewhat better artificial
  60. 02:17intelligence which could really promote
  61. 02:19a decisive strategic advantage because
  62. 02:22the US will be able to invent super
  63. 02:24intelligence and the Chinese will not.
  64. 02:26That is a category of magical thinking
  65. 02:28which does not seem very compelling to
  66. 02:30me because they assume that uh China
  67. 02:32will not be able to catch up pretty
  68. 02:34quickly.
  69. 02:36Every country is complex and a country
  70. 02:38as big as China with about 1.4 billion
  71. 02:41people is especially complex. One of the
  72. 02:43things I really appreciated after
  73. 02:45spending years living in China was just
  74. 02:47how different its three main economic
  75. 02:50hubs feel. Beijing was a city that uh we
  76. 02:54in [music] Shanghai called western
  77. 02:56Pyongyang because it is just so
  78. 02:58centralized. It is so Stalinist defining
  79. 03:01the political system. See so much of
  80. 03:03these concrete buildings with these dust
  81. 03:05storms. Shanghai which is much more of a
  82. 03:07commercial hub. In Shanghai we refer to
  83. 03:09ourselves as [music] the Paris of the
  84. 03:11east much more designed by the French
  85. 03:13with these leafy boulevards, cafe
  86. 03:15[music] line streets, much better
  87. 03:17weather. the southwest which is much
  88. 03:19more defined by regional beauty, a lot
  89. 03:22of different mountains, very distinct
  90. 03:24personalities, [music]
  91. 03:25amazing ethnic diversity and that is
  92. 03:27something um in which China is uh
  93. 03:29somewhat [music] similar to the United
  94. 03:30States where we're speaking from the San
  95. 03:32Francisco Bay area. It is very very
  96. 03:35different from New York City which is
  97. 03:36still very very different from Texas
  98. 03:38which is different from Washington DC.
  99. 03:41That's something that really brings to
  100. 03:42mind the differences uh between China as
  101. 03:45well. I think that the Chinese economy
  102. 03:46is incredibly competitive and my view is
  103. 03:49that if you're selling any products,
  104. 03:51China is the most competitive economic
  105. 03:53system uh in the world. By most
  106. 03:55indications, [music] it looks like China
  107. 03:57is in fact far ahead of the United
  108. 03:59States because what we have seen with uh
  109. 04:02the Chinese [music] companies is that
  110. 04:03they have actually produced a lot of
  111. 04:06excellent models in spite of pretty
  112. 04:09substantial export [music] controls that
  113. 04:11the US government has asserted by now
  114. 04:13close to a decade on some of China's
  115. 04:15leading firms. [music]
  116. 04:16And China has produced a lot of models
  117. 04:18which are not substantially behind a lot
  118. 04:20of the American alternatives. [music]
  119. 04:22the Chinese companies are producing
  120. 04:24open-source uh models. They seem to be
  121. 04:26more cheaply run. [music] If we're
  122. 04:28taking a look at some of these
  123. 04:29particular video generation models
  124. 04:32coming out of China, they seem to be
  125. 04:33maybe even ahead of the [music] United
  126. 04:35States. If we look a little bit beyond
  127. 04:38just who is making the best models right
  128. 04:40now, well, it doesn't look [music] to me
  129. 04:42like China is substantially behind at
  130. 04:44all. What is far more important perhaps
  131. 04:47in the next few years as we start to
  132. 04:50deploy AI is that China is just vastly
  133. 04:52outb buildinging the United States in
  134. 04:54electrical power. So China is building
  135. 04:56so much more coal, so much more nuclear,
  136. 04:58so much more solar and wind than the
  137. 05:00United States. Last year, China built
  138. 05:02about 300 gawatt of solar power alone.
  139. 05:05US built about [music] 30 gawatt. China
  140. 05:07right now has 40 nuclear power plants
  141. 05:09under construction. The United States
  142. 05:11has zero. China is made up of some of
  143. 05:13the most advanced factories in the world
  144. 05:16today. What they have are these things
  145. 05:17that they refer to as dark factories.
  146. 05:20Essentially, it is dark because most of
  147. 05:22the lights are off because you hardly
  148. 05:24need any human workers and which
  149. 05:26everything is very well automated. But
  150. 05:28it is not just these dark factories that
  151. 05:30matter so much. I think that we don't
  152. 05:31have to take a look at these glistening
  153. 05:33most advanced factories. What really
  154. 05:35defines the backbone of Chinese
  155. 05:37manufacturing are the broader masses of
  156. 05:40factories that are able to make iPhones,
  157. 05:43consumer electronics, electric vehicles,
  158. 05:45and their batteries in which the Chinese
  159. 05:48are, I think, far ahead of the Americans
  160. 05:50in terms of all sorts of manufactured
  161. 05:53products. China has these very
  162. 05:55well-developed production ecosystems. So
  163. 05:58for any part that you care to name, all
  164. 06:00of these factories, whether it's
  165. 06:02batteries, whether it's [music] sensors,
  166. 06:03they're all right next to each other in
  167. 06:05which the factory managers can just go
  168. 06:07next door and have another part. China
  169. 06:09has these dense networks of uh workers.
  170. 06:12It's about 70 million workers in
  171. 06:14manufacturing. So you [music] have mass
  172. 06:17workers, you have mediumskilled workers,
  173. 06:19you have very highskilled engineers or
  174. 06:21probably the most skilled engineers in
  175. 06:23the world today. these sort of
  176. 06:24disparities um can become a little bit
  177. 06:27scary when you combine all of these sort
  178. 06:28of things. I'm not sure if this is much
  179. 06:31less valuable than the US which is able
  180. 06:33to build more sophisticated models by
  181. 06:36every measure. But how much is that
  182. 06:38greater sophistication uh really really
  183. 06:40worth? I'm not sure I would trade it for
  184. 06:42a much more vibrant industrial
  185. 06:44manufacturing ecosystem.

  186. The Dark Side of the Engineering State

  187. 06:49China's leadership is made up of a lot
  188. 06:51of engineers and for the most part I
  189. 06:54think that the engineers have done a
  190. 06:56good job in terms of promoting physical
  191. 06:58dynamism in the country. If you are
  192. 07:00growing up in China probably uh 20 years
  193. 07:03ago your city had no subway lines and
  194. 07:06now you can have maybe a dozen maybe two
  195. 07:08dozen subway lines to get around. China
  196. 07:10built its very first highspeed rail
  197. 07:12system about 15 years ago and now it has
  198. 07:16about twice as many highspeed rail lines
  199. 07:18as the rest of the world combined. And
  200. 07:20so having the sort of really tall
  201. 07:22bridges that connect villages, having
  202. 07:24subways, [music]
  203. 07:25rail, electrical power, a manufacturing
  204. 07:28base. These are all good things that the
  205. 07:30engineering state has been able to
  206. 07:31produce. But there is a problem with the
  207. 07:33engineering [music] state. I think the
  208. 07:35fundamental problem with country made up
  209. 07:37of engineers is that they cannot stop
  210. 07:40themselves from entering social
  211. 07:42engineering, from entering population
  212. 07:45engineering, from stopping themselves
  213. 07:46from treating the population as yet
  214. 07:49another building material to be torn
  215. 07:52down and remolded as they wish.
  216. 07:54especially uh something like the one
  217. 07:56child policy in which the Chinese
  218. 07:58leadership really believed in the
  219. 08:00western dumerism of the 1970s that there
  220. 08:03were these mathematical formulas that
  221. 08:05determine the optimal population. All
  222. 08:07countries had to reduce the population
  223. 08:10as much as possible so that the
  224. 08:11environmental carrying capacity could be
  225. 08:14maintained and all of these calculations
  226. 08:16were bunkked. But the Chinese took it a
  227. 08:18little bit too seriously. They involved
  228. 08:20a missile scientist [music] by the name
  229. 08:22of Sunien to really get into the ears of
  230. 08:25the leadership and offer the
  231. 08:26mathematical justification for something
  232. 08:28as brutal as the one child policy which
  233. 08:31eventually involved over 35 years around
  234. 08:35300 million abortions as well as 100
  235. 08:37million sterilizations [music] of women.
  236. 08:40This was something that I think of as a
  237. 08:42campaign of rural terror that ran
  238. 08:44throughout the countryside. This is the
  239. 08:46problem [music] when you have a
  240. 08:47leadership that does not respect living
  241. 08:50in China produces all sorts of new
  242. 08:53adventures. One of the unexpected things
  243. 08:56in terms of the geopolitical competition
  244. 08:58was that my personal website danwong.co
  245. 09:01where I published my annual letters was
  246. 09:04blocked one day in the spring of 2022.
  247. 09:07This is a very small casualty of 2022,
  248. 09:09but I was a little bit surprised to see
  249. 09:11that my rinky dink website, which is
  250. 09:14hardly as big as Facebook or Wikipedia
  251. 09:16or the Wall Street Journal, um, ought to
  252. 09:18be blocked as well. But this is the
  253. 09:20[music] uh sort of information
  254. 09:22environment that people in China simply
  255. 09:24have to deal with in which if one sends
  256. 09:26a message on WeChat to uh someone else
  257. 09:29in China, they may not have received it
  258. 09:31because the state and the sensors inside
  259. 09:34the Tencent have already decided to
  260. 09:36block it already. That the state has
  261. 09:39already shuttered most journalism, most
  262. 09:41independent journalism in China. There
  263. 09:44are some very strict controls in the
  264. 09:46information ecosystem in China. Many
  265. 09:48people are aware of the censorship
  266. 09:51apparatus and there has been this uh
  267. 09:53wonderful analogy produced by a
  268. 09:55professor named Perry Link and this
  269. 09:58metaphor of an anaconda in a chandelier.
  270. 10:01So imagine if all of us are sitting
  271. 10:03around a dinner table with a a
  272. 10:05chandelier above us and inside the
  273. 10:08chandelier is this giant anaconda that
  274. 10:10is just coiled there. That is a lot of
  275. 10:12what the censorship apparatus feels
  276. 10:14like. None of us need to fear for the
  277. 10:17anaconda to emerge and strangle one
  278. 10:20[music] of us. All we just have to do is
  279. 10:22to know that there's an anaconda hanging
  280. 10:24there that might come down and might
  281. 10:26not. And that is enough to intimidate
  282. 10:28[music] and introduce fear in a lot of
  283. 10:30people. And I think that this is one of
  284. 10:31these things in China's information
  285. 10:33ecosystem where you never really know
  286. 10:35when the anaconda might come down before
  287. 10:38people automatically self-censor. And I
  288. 10:40think that sort of attitude is corrosive
  289. 10:42to aspects of creative thinking in
  290. 10:45China. That one of my determinations as
  291. 10:47a writer has been not to be intimidated
  292. 10:50by whatever are the minations in China
  293. 10:54and to just tell the truth as best as I
  294. 10:56can within China. talking about what I
  295. 10:59saw were some of these spectacular
  296. 11:01achievements around manufacturing as
  297. 11:03well as public works at the same time as
  298. 11:05acknowledging that the communist party
  299. 11:07made these pretty grave [music] mistakes
  300. 11:08as I saw it during the zerocoid program.
  301. 11:11I think it was really important for us
  302. 11:13[music] to just be able to see both of
  303. 11:15these things. And I think that it is
  304. 11:17really important for all of us who value
  305. 11:19[music]
  306. 11:20independent thinking, critical thinking,
  307. 11:22not to be too given into that and not to
  308. 11:25allow censorship uh inflict our
  309. 11:28consciousness too much.

  310. Winning by Losing Less

  311. 11:34Who is going to be a clear winner? Right
  312. 11:36now it looks like the US and China are
  313. 11:38both very intent [music] on losing and
  314. 11:41both of these countries had this
  315. 11:43remarkable skill in delivering
  316. 11:46self-beatings and try to hobble
  317. 11:48themselves in this uh very big race. Now
  318. 11:50my model of the US China competition is
  319. 11:53that whichever country is ahead is going
  320. 11:56to make mistakes out of overconfidence
  321. 11:59and hubris and whichever country is
  322. 12:00behind is going to really feel the crack
  323. 12:02of the whip in order to really catch up.
  324. 12:04And I am suspicious of any argument says
  325. 12:07that China must win because of
  326. 12:09manufacturing or China must lose because
  327. 12:12of demographics. I think that both of
  328. 12:14these countries have a lot of strengths
  329. 12:16and that this is a dynamic process in
  330. 12:19which both countries are going to try to
  331. 12:21run faster. I was living in China in
  332. 12:232021 when top leader Sidin Ping decided
  333. 12:26to have a controlled demolition of the
  334. 12:29property sector as well as smack around
  335. 12:31a lot of tech entrepreneurs most notably
  336. 12:33Jack Ma because he felt that he was in a
  337. 12:36very good spot after managing co very
  338. 12:39very well and actually what we've seen
  339. 12:41is that this property slowdown this
  340. 12:44smacking round of tech entrepreneurs has
  341. 12:46been exactly the problem of the cause of
  342. 12:49a lot of China's economic problems
  343. 12:50today. So, the US felt like it was ahead
  344. 12:53for a little while. And uh what did it
  345. 12:55do? President Trump is now doing what
  346. 12:57President Trump is doing. Namely,
  347. 12:59eroding a lot of alliances that the US
  348. 13:01has had with these tariffs. Uh
  349. 13:02de-industrializing America. Under uh
  350. 13:05Trump's first term, the US has lost
  351. 13:07about 80,000 manufacturing jobs. And
  352. 13:10President Donald Trump has done these
  353. 13:12pretty hurtful things to US
  354. 13:14manufacturing. One of the most hurtful
  355. 13:16was last year when the Department of
  356. 13:19Homeland Security deported around 300
  357. 13:22South Korean engineers from the state of
  358. 13:24Georgia who were working at a Hyundai
  359. 13:26plant trying to build electric vehicle
  360. 13:28batteries. And I think this was not a
  361. 13:30pleasant experience for the engineers. I
  362. 13:33think that this did not look good to
  363. 13:34much of the rest of the world such that
  364. 13:36if I were an engineer in France or
  365. 13:39Taiwan or South Korea, maybe I will have
  366. 13:41some doubts about whether I want to take
  367. 13:43a job in Georgia in the future or
  368. 13:45Alabama in the future or Texas in the
  369. 13:47future. And so what we see are both
  370. 13:49these countries making very big mistakes
  371. 13:52and I think the country will go to the
  372. 13:54one that stops making so many of these
  373. 13:56mistakes. I think one of these
  374. 13:58challenges with China is that it could
  375. 14:00do much better if the Communist Party
  376. 14:02could just learn to be a little bit more
  377. 14:03restrained and do a little bit less
  378. 14:05[music] and try less to be these intense
  379. 14:08social engineers, to be engineers of the
  380. 14:10soul. At the same time, the United
  381. 14:12States could do much much better if it
  382. 14:13were not so [music] intent on trying to
  383. 14:15have these very strange political
  384. 14:17figures that have also been beating up
  385. 14:19on America's allies and the home front
  386. 14:21[music] as well. So my message to both
  387. 14:23of these countries is do less, please.
  388. 14:27I think a decade from now [music] the
  389. 14:29label of made in China is going to be
  390. 14:32viewed as excellent quality and the way
  391. 14:34that made in Germany, made in Japan has
  392. 14:36been viewed [music] as excellent
  393. 14:37quality. I think over the next 10 years
  394. 14:39the future of manufacturing is going to
  395. 14:41move more and more towards China.
  396. 14:43Silicon Valley founders have [music]
  397. 14:44been some of the most inspirational,
  398. 14:47hardest working, most ambitious people
  399. 14:50ever in history [music] and they have
  400. 14:51been rewarded for it by driving a lot of
  401. 14:54economic value. But I would say that
  402. 14:56[music] Silicon Valley has not
  403. 14:57necessarily delivered value throughout
  404. 14:59the rest of the country in which feels
  405. 15:01that [music]
  406. 15:02uh the entrepreneurs in California are
  407. 15:04very very different from that. That's
  408. 15:06not usually Silicon Valley's fault.
  409. 15:08[music] That is the fault of the San
  410. 15:10Francisco government, the California
  411. 15:11government, the American government,
  412. 15:13which have not been able to deliver
  413. 15:15[music] some of these basic services
  414. 15:16that people want like orderly streets
  415. 15:19that are clean, that [music] do not have
  416. 15:20too much crime, in which people feel
  417. 15:22that they can have affordable housing,
  418. 15:24in which they feel that they are living
  419. 15:26in a a growing [music] thriving country.
  420. 15:29Instead, the California government has
  421. 15:31driven a lot of people away from the
  422. 15:33state of California. Net out migration
  423. 15:35from California, [music]
  424. 15:36especially to states like Texas and
  425. 15:38Arizona, have been pretty considerable.
  426. 15:40So, there are all sorts of uh
  427. 15:42deficiencies with [music] the US. And I
  428. 15:44hope that the Silicon Valley founders
  429. 15:46can also think a little bit more broadly
  430. 15:49um and be supported by the politicians
  431. 15:51[music] and deliver these benefits to
  432. 15:52ordinary Americans who don't always feel
  433. 15:54that they are very well connected to
  434. 15:56some of these uh strange people here in
  435. 15:58Silicon Valley.

  436. Next Episode

  437. 16:01If we look at within companies and we
  438. 16:02compare jobs that are more exposed to AI
  439. 16:05versus less exposed to AI, we're seeing
  440. 16:06that the jobs that are more exposed to
  441. 16:08AI, the young workers in those jobs are
  442. 16:10seeing 16% slower employment growth. So
  443. 16:13that's pretty large. It's actually
  444. 16:15high-income jobs because they do the
  445. 16:18kind of knowledge work that AI [music]
  446. 16:19is more capable of doing right now.
  447. 16:21We're not going to be in a world where
  448. 16:22there's no AI. That we're just not going
  449. 16:24back to that. That's not going to be
  450. 16:26what's happening going forward. But I
  451. 16:27also think that's an enormous
  452. 16:28opportunity. So whether you're more
  453. 16:30automated or augmented really depends on
  454. 16:33what are the tasks that you're focusing
  455. 16:34on. Are you increasing the scope of
  456. 16:36tasks that you can do or are your tasks
  457. 16:38getting [music] shrunk by the
  458. 16:39introduction of this technology?
  459. 16:46[music]
  460. 16:46Hey.
  461. 16:54Hey.