China vs America: The Battle for Global Dominance Explained | Dan Wang interview

Summary

• Dan Wang’s core frame: China is an “engineering state” that treats society “as if it were just another building material” — the same machine that pours gleaming bridges and 500 GW of solar also produced the one-child policy and zero-COVID. America is its inverse, a lawyerly society whose pluralism guarantees debate but can’t execute; his prescription is symmetrical: “I want the US to be 20% more engineering and I want China to be 50% more lawyerly.”
• Innovation is not China’s binding constraint: “China’s major advantage in being a technological superpower is that it can just wait for the Americans to invent a lot of stuff… They’re the prairie fire that uses the American spark.” Solar PV was invented at Bell Labs in 1954; ~90% of the industry now sits in China. With 70M manufacturing workers versus 12M in the US, Wang thinks it’s harder for America to learn one-to-N scale-up than for China to learn zero-to-one — and the science gap is already closing.
• Wang frames the China equity discount as structural: ~9% average growth for 20 years against essentially flat Shanghai/Shenzhen exchanges; a party that “resembles the God of the Old Testament — very generous sometimes, but then randomly smite you” (ByteDance’s forced 2018 apology, Ant’s cancelled IPO); plus a Cayman Island entity with contractual relationships with ByteDance, capital controls, and a Taiwan-sanctions tail that is hard to lay to bed. If you’re a pension holding ByteDance and Treasury comes asking, “what are you doing?”
• AI is a genuine five-year toss-up despite America’s “decisive lead” today on compute: China adds ~500 GW of solar this year versus ~50 in the US, has 33 nuclear plants under construction versus zero, and 7 of Meta’s 11 publicly disclosed superintelligence hires went to Chinese universities. The deployment asymmetry is the kicker: “maybe America gets a much better McKenzie and China gets a much better Foxcon” — which then makes drones, munitions, and ships.
• The undervaluation question in one pairing: Apple (~$3–4T) spent 10 years on Project Titan and quit; Xiaomi, at maybe 1/20th the value, pledged $10B in 2021, has EVs on the road, and won one of the speed records at the “Norberg Ring” racecourse on its first attempt, beating Porsche and BMW. “Maybe what actually matters is a company saying that it’s going to do something and actually achieving it.”
• The sharpest warning is about horizon: even if top-down systems fail long-run (14,000 millionaires left China in 2023; censorship forecloses cultural and financial superpower status), “the future is made up of a series of short runs. If China is really able to do super well in the next 5 years, 15 years, [that] might be enough to subdue America in pretty substantial ways.”
• On war, Wang would still bet on peace — the Taiwan status quo currently works for Beijing, Taipei, and Washington — but the variables to watch are a perceived closing reunification window, Philippine ship-rammings in the South China Sea, and unpredictable border flare-ups like India in 2020. “If you bet on peace, it’s been a pretty good bet, and I expect that will continue.”

Deep dive

1. China is an engineering state — and society is just another building material

• The book’s original title was “Move Fast and Break People” (“publisher did not like that one — too much like Facebook… and they were right”). The coil of violence survives in “Breakneck”: physical dynamism — bridges, subways, hyperscalers — is “mostly pretty positive,” but “they’re not just physical engineers. They’re also social engineers.” The Three Gorges Dam displaced around a million people; add the one-child policy, zero-COVID, and repression in Tibet and likely Xinjiang.
• The lineage runs deeper than the fact that the entire Politburo Standing Committee held engineering degrees: the Great Wall and the Grand Canal, emperors who “barely hesitated to completely reorder a peasant’s relationship to her land,” people governed as “chess pieces to be moved around” — imperial times to the present.
• What the US has that Wang can’t imagine China adopting: pluralism. Six years in China (tech analyst at the firm rendered in the captions as Goffcow Trigonomics, 2017–2023, through the trade-war-turned-tech-war and all three years of zero-COVID) left him skeptical the party “has mastered some autocratic formula” — it cannot tolerate debate, useful dissent inside the system, or solve succession.

2. Innovation is not the bottleneck: the prairie fire uses the American spark

• Wang wants to dissolve the garage-genius definition of innovation. China became a technological superpower without “go into nature, take a lot of LSD, and Apple Computer comes out” — it imported managerial expertise (including Apple’s own standards), assembled components from the US, Japan, and Germany, and “learned a million and one things on the shop floor” until it could build things America no longer can.
• The compounding asset is process knowledge: 70 million manufacturing workers “solving three new problems a day before breakfast.” The worker on Shenzhen’s original 2008 iPhone lines might build a Huawei phone the next year, maybe start a drone company a few years later, and maybe start an EV-battery business after that — that career path is why Shenzhen is the hardware capital of the world.
• Nobel counting is the wrong scoreboard — prizes are “backwards looking by decades,” rate-limited, arbitrary. The Andy Grove essay (Bloomberg BusinessWeek, ~2010) has the right one: America sets ladders in place; its firms no longer climb them. And solar — invented in New Jersey, 1954, treated as a science project — is now ~90% a Chinese industry, polysilicon through final module.

3. The catch-up race is asymmetric, and it favors China

• Asked which is harder — the US learning scale-up or China learning zero-to-one — Wang is unequivocal: harder for the US. Chinese researchers take a rising share of top-1%-cited papers, China will likely reach the moon by the end of the 2020s while “America might not be able to get back,” and against some successes, including the TSMC Arizona fab “actually doing fairly well,” the apex manufacturers — Boeing, Intel, Detroit, Tesla — are “mostly going downhill.”
• His fix is deliberately not a statute list: get Americans to “treat technology as a political project as well as an aesthetic project.” The rot dates to a 1990s CEA chairman’s quip — “computer chips, potato chips, what’s the difference” — tolerable for socks and televisions, alarming when it reaches iPhones and munitions.
• America now has Sputnik moments without the action: Patrick catalogued a half dozen, from Obama on Chinese high-speed rail to senators on Huawei 5G to a Harvard professor in the NYT on Chinese universities. Meanwhile the Second Avenue subway cost ~$2B per mile and California HSR — approved by referendum in 2008 — has carried zero riders, with a first Bakersfield–Merced segment due 2030–2033. “Why is there no sense of crisis? Patrick, riddle me this.”
• Patrick’s theory — China compounding deliberately below the radar, per the hundred-year-marathon thesis — gets redirected: “it sounds like a problem with Americans rather than with the Chinese. They are really happy to broadcast all of their achievements… you only have to go over there.”

4. More alike than any other people — but China believes in the future harder

• The book’s first paragraph: Chinese and Americans are more alike than any other people — the “two great founts of entrepreneurial dynamism” (it’s Shenzhen and Silicon Valley inventing the future, “not so much Europe and Japan”), a shared technological sublime (Golden Gate, Apollo), and a shared instinct to muscle neighbors who don’t get in line, Philippines on one side, Canada and Denmark-over-Greenland on the other.
• Europe, by contrast, is “a mausoleum economy — things are very beautiful but things are also very dead,” with optimism “only about the past.”
• His cutting-room-floor material is Chinese futurism: no tech backlash (nothing in China is allowed to be partisan, so tech isn’t), e-commerce penetration roughly twice the US, AI regarded as something that could be a friend. Even the phone addiction he saw in Shanghai in late 2024 — whole dinner tables silently scrolling — reads two ways: social dysfunction, but people are not very concerned about what technology is doing to them.

5. Where to look: four cities, any factory, and the work-ethic gap

• The itinerary: Shanghai (“my favorite city in the world” — the bleeding edge of consumption, almost-all-electric DD taxis, subways that don’t screech), likely Hefei (two hours by rail, historically poor, now the center of China’s EV industry), Shenzhen (“probably the most boring city in China” but HQ to Huawei, Tencent, DJI), and Chongqing, the TikTok-famous city carved into cliffs with a subway running through an apartment building.
• For an envoy of US tech entrepreneurs, though: skip the cities and “anchor your trip around a few factories.” Even brake pads — he toured a “superbly interesting” brake-pad factory — are “always a ravishing experience.” European clean-tech investors are already making these trips; he knows of two this year and three more planned.
• At the top 0.01%, founders are peers who respect each other. At the median, Wang strongly suspects Chinese workers put in more hours: 996 versus Google clearing out by 3pm Friday. US, Japanese, and German automakers take ~6 years concept-to-road; Chinese makers ~18 months to 2 years — “at a first approximation, they’re working at least three times faster.” Mtoan survived a battle royale of 5,000 Groupon clones.
• The hedge, exactly as hedged: “the Chinese are just comprehensively outworking the Americans in some big ways. Maybe that doesn’t matter for technologies like AI — but for pretty much everything else, it pays to work at much faster cycles.”

6. The China discount is structural: an Old Testament God plus a Taiwan tail

• The valuation puzzle Patrick poses via ByteDance — massive free cash flow, equity at a fraction of a US multiple — Wang frames through three paradoxes. First: ~9% average GDP growth over 20 years against essentially flat Shanghai and Shenzhen exchanges — “nothing commensurate with 9% growth,” for structural reasons.
• Second, the party itself: “the Communist Party resembles something like the God of the Old Testament — very generous sometimes, but then randomly smite you.” ByteDance was humiliated in 2018 over jokes that “violated core socialist values”; its founder wrote a “totally self-flagellating letter” (“impossible to imagine any tech CEO in America writing something like that”). Jack Ma “shot off his mouth one too many times” in 2020 — Ant smashed, still not public, then scalps across online education, e-commerce, antitrust.
• Third, you may not own what you think you own: a Cayman Island entity with contractual relationships with ByteDance, capital controls deterring portfolio flows, and a Beijing that is “very clear that it intends to seize” Taiwan someday — with Russia’s treatment (frozen reserves, barred portfolio managers, even McDonald’s exiting) as the sanctions template. Not his expectation, “but none of us can be sure that they won’t.” Hence the endowment’s real question: do you want Treasury asking what you’re doing?
• On the hot-war variables themselves: Taiwan is the obvious flashpoint if Beijing decides its window is closing; the Philippines and China “really been ramming each other’s ships”; and China has ~25 neighbors and a history of eruptions like the fatal 2020 Himalayan clash with India. Still: “if you bet on peace, it’s been a pretty good bet, and I expect that will continue.”

7. Equilibrium, self-sufficiency, and which society is higher agency

• The 30-year US bet that “markets and the internet will change China” into a liberal democracy “could be firmly laid to bed.” His Hoover guru Steven Cotkin wants a new cold war — compete at the level of systems, deliver better for your people — but Wang is nervous about the analogy: the Cold War wasn’t cold for Vietnam, and unlike the Soviet case there’s enormous US–China trade. “We should come up with some new term.”
• American self-sufficiency is thinner than it looks: imports are ~12% of GDP, “but in a crisis it turns out the US actually really needs something like this 12%.” The COVID parable, as told by a Shanghai factory manager: American firms asked whether masks were their core competence; “Chinese companies say that making money is their core competence” — so Foxcon and JD.com retooled and made the masks.
• China is patching its remaining dependencies at scale: ~500 GW of solar added this year versus ~50 in the US, 33 nuclear plants under construction versus zero, one in every two cars sold electric by year-end, plus deliberate food and semiconductor self-sufficiency drives. Asked which society is higher agency, Wang doesn’t hesitate: China — citing Fukuyama’s definition of social trust as spontaneous coordination, which China displayed in early COVID “and the US in 2020 really did not.”

8. Lawyerly America: pluralism’s price is trains slower than 1914

• The lawyerly society guarantees his favorite American value — pluralism, robust debate, the Abundance-conference energy to self-diagnose — and you can’t build a $4 trillion Nvidia without substantial legal protections. But lawyers are “for the most part protectors of the rich,” and “society won’t work well if it is mostly about serving the rich.”
• Two radicalizing train anecdotes: the new Acela trains will be ~11 minutes slower than the old ones (“the foam seats are better”), and a 1914 timetable shows Grand Central–New Haven was faster a century ago — partly because Connecticut homeowners’ “very expensive lawyers” sued Amtrak’s route squiggly.
• His post-book realization cuts both ways: “maybe there can be too much state capacity.” Efficient engineers can drag society far off track — one child, zero COVID, “the number is right there in the name” — before any correction; the US “doesn’t get stupid ideas like zero COVID, but it also doesn’t get California high-speed rail.” The ideal sits between: “20% more engineering” for the US, “50% more lawyerly” for China.

9. Not the USSR, not Japan — but people are voting with their feet

• China studies collapse professionally: avoid Gorbachev’s simultaneous economic-plus-political reform; avoid Japan’s closed, all-domestic-value-add model (China threw open its doors to Apple and Tesla until the supply chain itself became Chinese). The resulting hybrid — “some of the control paranoia of the Soviet Union, some of the manufacturing excellence of Japan, and some of the American entrepreneurial hustle” — is why he expects a formidable power, not a collapse.
• But the exit data is loud, straight out of James C. Scott’s Zomia: young Chinese making art, dabbling in crypto, and smoking dope in the mountains around Chiang Mai rather than joining the “great rejuvenation of the Chinese people”; 14,000 millionaires departing in 2023; and at the 2024 peak, CBP apprehending 30–40,000 Chinese nationals a month at the Texas border after Darian Gap treks. Censoriousness (“they censor everything they can’t understand, which is a lot”) forecloses cultural superpower status; capital controls — R&B still ~4% of global trade volumes — foreclose financial superpower status.
• Don’t bet on a liberal successor to likely Xi Jinping either. Likely Deng Xiaoping, architect of reform and opening (“perhaps the most tremendous economic good ever in the history of humanity”), was also “half a Mao” — under whose rule the one-child policy was unleashed, and a “butcher of Beijing,” who fired two handpicked successors and never solved succession. “I’m skeptical that there are deep pluralist genes in China”; his bet is that Chinese politics looks much the same in the future.
• The narrow case he insists on is enough to matter: even if the engineering state fails in the long run, “the future is made up of a series of short runs” — 5 to 15 great Chinese years “might be enough to subdue America in pretty substantial ways.” His nightmare number: US manufacturing employment falling from 12 million to 6 million within a decade.

10. AI is a toss-up within five years — and Xiaomi says the scoreboard is wrong

• Today “the US still has a decisive lead” because it has the compute and China lacks leading Nvidia chips. Over five years it’s “much more of a toss-up”: China’s power buildout dwarfs America’s, and talent is fluid — nobody knows if the relevant pool is 1,000 researchers or 10,000, but 7 of Meta’s 11 publicly disclosed superintelligence hires went to Chinese universities (all, he believes, Chinese nationals), and repatriation risk is real — ByteDance-comparable pay, “maybe they just want the better noodles,” fear of anti-Asian populism. Meanwhile Trump “wants to make a deal”: a 15% export tax on Nvidia in exchange for more compute flowing to China.
• The deployment asymmetry may decide it: “maybe America gets a much better McKenzie and China gets a much better Foxcon — and then they’re going to use this much better AI Foxcon to produce a lot more drones and munitions and ships.” The US “simply doesn’t have the training data, the process knowledge” to AI its way into manufacturing.
• His favorite mispricing exhibit: Apple, worth $3–4 trillion, spent ~10 years on Project Titan and threw in the towel; Xiaomi — “probably not even 1/20th Apple’s valuation” — pledged $10B in 2021, has its first electric SUVs on Chinese streets, and won one of the speed records at the “Norberg Ring” racecourse on its first entry, beating Porsche and BMW. “Are we undervaluing China because we’re thinking too much about the market valuations of Nvidia and Apple?”
• Writing the book hardened both halves of his two-front war: Shenzhen convinced him China’s “engine of technological momentum” won’t be derailed by export controls — and researching one-child, “a campaign of rural terror meted out against overwhelmingly female bodies” (300 million abortions, 100 million sterilizations), convinced him not to give growth a moral pass. “Repression can grow worse while technological dynamism grows richer.” He wrote it outside the reach of Perry Link’s “anaconda in the chandelier” — his site is blocked in China — and will learn from his next visa application whether Beijing lets him back to Shanghai.