China, China, China. Breaking Down China’s Tech Surge | BG2 w/ Bill Gurley and Brad Gerstner

Summary

• Bill Gurley returned from his first post-COVID China trip with one framing quote that carries the episode: “every founder and every VC in China studies the West at a nauseating level… The West doesn’t do that of China.” His core argument — if China is the most consequential relationship for America, “I don’t understand how you’d want to know less” — and he notes that I think all but four of the 24 members of the Select Committee on the CCP have never been to China.
• The auto data is the tradeable shock: likely Xiaomi builds 1,000 cars a day with 2,000 employees (vs. ~6 employees per car/day in the US), is sold out with a 30-40 week backlog, and its ~$40K car made Ford CEO Farley say “if we lose this we do not have a future at Ford.” Implication Gurley draws: reshoring auto jobs may find “there may not be any jobs” — global car manufacturing could total ~400K in 5-10 years.
• The US is only ~14% of China’s exports and ~3% of its GDP — Brad’s point that America overestimates its leverage. BYD’s $10-20K EVs illustrate the comparative-advantage argument: why would Mexico buy the $50K one from America?
• Hidden-value call: likely Baidu’s Apollo robotaxi rival to Waymo could cost ~$30K per vehicle, if that figure is right, vs. reportedly $150K+ for Waymo (partly China’s solid-state LiDAR edge), while people value Waymo at $170B inside Google — yet “you can buy shares of Baidu for zero enterprise value,” a $30B market cap with $30B cash.
• On AI: China’s 14th five-year plan mentioned open source; likely Qwen matters because Gurley thinks Alibaba is roughly a 70% player in China cloud, ByteDance is the company to watch on consumer AI, and open models helping each other create hyper-competition “maybe more” intense than EVs. Entrepreneurs there aren’t particularly concerned about a model monopolist.
• Gurley’s debunk of the “they only win by stealing and subsidies” narrative: BYD’s Stella Li — “come show me the money I’m getting from the government” — plus the thought experiment that free Tesla patents + subsidies still wouldn’t make Ford/GM competitive. Brad’s warning: the cheating narrative “allows us to delude ourselves into believing we don’t need to get better ourselves.”
• Policy stance (both in the “tech pragmatist” camp with Jensen and Cook): compete, don’t decouple — invite Chinese JVs in EVs/solar, including some 49/51 ownership arrangements of the kind used in China, let Korea build US nuclear at 1/4 the cost, reserve tariffs for genuine national-security industries (rare earths, chips, pharma), and strip the regulatory “mud.” A pragmatic Trump-China deal “could easily get done” if the rhetoric isn’t derogatory.
• Watch items: China’s new K visa may allow people studying technology to come without a job requirement, just as Gurley heard stories of groups of 50-100 admitted PhD students being told they could not attend in the US; China’s VC market is in a lull (Sequoia/GGV split, provinces investing on “non-starter” terms, “don’t be the tallest tree”) — yet entrepreneurial energy in Shenzhen is undiminished.

Deep dive

1. Study China to understand it

• Gurley’s motivation, delivered as a challenge: China is “the most consequential relationship for our country,” so “I don’t understand how you’d want to know less.” He’d been 4-5 times pre-COVID, took his Mandarin-speaking daughter, and read an early copy of Dan Wang’s Breakneck on the flight over — anticipating the “agent for the CCP” accusations before they come.
• The episode’s signature exchange, from an unnamed lunch companion: “every founder and every VC in China studies the West at a nauseating level” — podcasts, speeches, financials — “and the West doesn’t do that of China.” Gurley: “maybe we should be studying the best and the brightest over there.”
• His receipts on Washington’s ignorance: a ChatGPT deep-research run on the 24 members of the Select Committee on the CCP found I think all but four have never been to China, and those who went, went seven years ago. “Wouldn’t you want the most educated people on that committee?”

2. Engineers build, lawyers protect

• Dan Wang’s (Hoover Institution) core lens: the Politburo is dominated by former engineers, DC by former lawyers — which may explain why China is great at building and weaker on social things. Gurley stresses the book is balanced: Wang gives lawyers the edge on free speech and personal rights, hated the abrupt Shanghai lockdown, and is scathing on the one-child policy (now flipped to encouraging three children, unsuccessfully).
• The under-appreciated mechanism behind high-speed rail, overnight cities and EV/solar dominance: provincial leaders compete like divisions of a company — run your province well and “you might get the CEO job” in the federal government. Brad’s pushback that this sounds like Newsom vs. DeSantis; Gurley’s distinction: a singular government can make the choices, and its every-five-years plan sets priorities for the provinces.
• The flip side Gurley doesn’t smooth over: the same competition produces ghost cities and overbuilding — world-leading EV and solar sectors where “some of these companies need to go bankrupt and a province may not want them to because of employment.”

3. China’s EV innovation accelerates

• BYD, the world’s largest EV maker at ~4M vehicles, started in batteries, competes with Foxconn in phones (bought what sounds like Jabil Circuit), and spans $10-15K entry cars to a high-end sports car and an SUV Gurley literally drove into the water. On the “China can’t innovate” trope: “TikTok was there first… then Reels copied TikTok,” ByteDance’s founder and Lei Jun are “remarkably unique,” and even Pop Mart — a $40-50B public toy company — counts.
• The Xiaomi story as told: likely Xiaomi’s Lei Jun (whom Gurley met back in ‘03-‘04 running e-commerce firm Joyo) built I think the world’s #3 handset maker, then decided around 2021 to build a car — around the time Apple said it would build one. His method, from his 2024 State of the Union talk Gurley urges everyone to watch: having had a driver for 10 years, he switched seats, left notes on every unfamiliar car in his company lot, and claims he drove 170 cars to learn the product.
• The numbers that stunned Gurley on the factory tour: 1,000 cars/day with 2,000 employees, highly automated, sold out with a 30-40 week backlog and $5K just to join the waitlist. Versus ~6 employees per car per day in the US — so “if you want to bring the jobs back, by the time you get the jobs back there may not be any jobs”; global car manufacturing could total ~400K in 5-10 years.
• The Western CEO chorus Gurley cites as cover: Farley shipped a Xiaomi to Chicago, called it “the most humbling thing I’ve ever seen” and said “if we lose this we do not have a future at Ford”; Mercedes’ CEO wants “a reality check”; Stellantis’ Carlos Tavares called Chinese EVs “possibly the biggest risk” and EU tariffs on them “a major trap for automakers.”

4. America’s leverage is limited

• Brad’s framing numbers: the US is ~14% of China’s exports and ~3% of China’s GDP — “we’re just not that important to China.” BYD has a big presence in Hungary and is building—or already has—a factory in Mexico; “why, if you’re Mexico, would you not buy the $10-20K EV? Why would you buy the $50K one from America?”
• Gurley’s comparative-advantage warning, stated categorically: close the import border to the world’s most competitive products and “your people are forced to buy a product that is overpriced… from a standard of living perspective, they’re worse off.” Protection also risks entrenching “the regulatory grift” (Brad’s concession, with a carve-out for genuine national-security industries).
• The buried-asset observation: Apollo, likely inside Baidu — a $30B market cap with $30B cash, “zero enterprise value” — is a robotaxi rival to Waymo, while Waymo gets a $170B mark inside Google. If it costs ~$30K a vehicle vs. reportedly $150K+ for Waymo (thanks partly to China’s solid-state LiDAR advantage), “I don’t know why you’d want to deploy Waymos” for rest-of-world, where Gurley suspects governments don’t share Washington’s hawkishness.

5. Compete through JVs and reform

• Gurley’s advice to Trump: be open to JVs coming toward us — for 50 years Western automakers opened facilities in China, and some were forced into 49/51 ownership arrangements; now let Korea (which builds nuclear at 1/4 US cost) build plants here, let Ford or GM buy or JV with a struggling Chinese EV maker (NIO’s stock “not doing all that well”), same for solar. And attack regulation: “an autocratic country that has specific goals can move so much faster… we’ve created so many people whose jobs it is to block things.” Credit to Governor Shapiro on Three Mile Island and I-95.
• On the deal odds — Gurley’s explicit hedge: “I have zero insight… this is pure speculation,” having met no government officials. But his read: China will be driven more by what’s “fair and face-saving” than numeric, so “a pragmatic deal could easily get done” — unless the US insists on “derogatory language.” His broader moral objection: “I just don’t know why vilifying a billion people is a good idea.”
• Brad’s market note: Trump tweeted a threat of 200% tariffs if rare-earth magnets don’t flow, amid talk of a September China visit — he sees the deal as one of the biggest variables for back-half growth and sentiment, and reads Trump as “a self-described deal junkie… a pragmatist, not an ideologue.”

6. Subsidies don’t explain China’s lead

• Stella Li’s retort when Gurley raised subsidies at BYD: “If I’m getting all this government subsidy money, can you please find it and show it to me? We’re a public company.” Gurley’s mirror: the US hands out factory subsidies, a decade of EV credits, and government money to Intel — “it’s very unclear to me what we’re pointing at and accusing.”
• The thought experiment that lands: Elon published all Tesla patents — free IP for Ford and GM. Add subsidies: would they be competitive with China? Brad: “No, sir.” Gurley: “the thing we’re accusing them of being the reason they’re succeeding, if you flipped and gave that to US companies, none of us have confidence that would work.” Brad’s synthesis — the cheating narrative “allows us to delude ourselves into believing we don’t need to get better ourselves,” like a kid blaming a lost race on cheating.
• On the 20-year “who won” question, Gurley reframes via finite-vs-infinite games: the US benefited post-WWII because Asia and Europe “had been blown up”; Deng’s capitalism lifted ~500M people out of poverty — “I don’t think they really want to say we shouldn’t have let 500 million people out of poverty.” The bigger complaint: naive trade policy while the US simultaneously moved toward a regulatory state.

7. China’s AI stack stays open

• The 14th five-year plan mentioned open source — a 20-year lineage, not an overnight reaction — and Gurley urges everyone to read the next plan when it drops, since it historically directed heavy investment that could lead to global competitiveness. Consequence: entrepreneurs there aren’t particularly concerned about a model monopolist, so they feel “more relaxed” building products across models.
• The scorecard as he heard it: DeepSeek has “the most intellectual brand” and national-pride halo; likely Qwen matters because Gurley thinks Alibaba is roughly a 70% player in China cloud; ByteDance is “the company to watch on anything consumer” with the app closest to OpenAI’s; Tencent has the WeChat asset but hasn’t been aggressive; Xiaomi is a wildcard because of Lei Jun and phone share.
• On staying at the frontier: Gurley had that sense before he went, because of the number of competitive open-source models and the way they can train each other. Open models create the EV-style hyper-competition “maybe more, because of the way the models can help one another. You can’t take someone else’s EV and make yours better, but here you can.”

8. Private markets pressure Google

• Brad opened by calling the OpenAI/Anthropic rounds “bigger private IPOs than any public IPO in the last 5 years”; Gurley argued public incumbents “haven’t internalized” that private markets will fund unprecedented burn, and that cash burn is a proxy for risk. So “if everything’s at stake for Google, should they be willing to lose five or 10 billion dollars? Because the startup attacking their space is.”
• The precedent he keeps: Rich Barton at Zillow “took a lot of heat” chasing Opendoor but “engaged and played that game on the field” — and it was “probably the right thing to do.” Google has done this before — Kubernetes against AWS, Android against Apple — so “maybe they should be more aggressive even still because of what’s at stake.”

9. China’s ecosystem sends mixed signals

• China’s venture market is “in a bit of a lull”: the Jack Ma episode, DiDi being pulled back from US markets, the for-profit-education companies being taken out and Tencent going flat for a couple of years because of gaming reforms; Sequoia and GGV split in half, leaving only HongShan (Neil Shen), ZhenFund and IDG meaningfully active. RMB capital is scarce, and provinces now invest on terms “you and I would consider non-starters.” The operating rule everyone knows: “don’t be the tallest tree.” Yet entrepreneurial energy is intact — locals “heavily dispute” the FT’s collapsing-startups chart, and Shenzhen (DJI, BYD, Huawei) is “a new young 20 million people highly energetic town.”
• The counter-flow to watch: China’s new K visa may allow technology students to come without a job requirement, as Gurley heard stories of groups of 50-100 admitted PhD students being told they could not attend in the US. Gurley’s hedge intact: “I don’t know how successful they’ll be” — but with ~50% of AI researchers in America being Chinese, decisions matter.
• Ground-truth details: credit cards are rarely accepted — payment is near-100% WeChat Pay/Alipay, with table-level QR codes to order, pay and walk out, “universal” from high-end hotel to street vendor. (On crypto policy Gurley admits flatly: “I don’t know the answer to that question.”)
• Brad’s closing thesis, and the episode’s through-line: the US is in “an incredible position” but “if we focus too much on how do we slow down China, we take the eye off the ball on how to accelerate our own race.” Gurley’s coda: read the global car CEOs, read Breakneck (Tyler Cowen: possibly best book of the year) — “just make sure you have the exact right information as you then go to make decisions.”

Verification Notes

• Raw captions render several resolved names ambiguously (“Xiai,” “BYU,” “Quinn,” and “Jable Circuit”); the digest uses qualified references where those identifications matter.