Li Xiang: What top founders share is not picking right, but inhumanely-high iteration speed
He says that picking the right category and persisting long-term are just entry-level skills for top founders—what actually separates them is the inhumanely fast iteration speed practiced by Nvidia, Bytedance, and Meituan.
The video won't play here. Listen to the audio instead:
The argument · tap a timestamp to hear it
Reframe college as a choice problem, not a right-or-wrong problem
Li Xiang's monthly income at 18 was ten times his parents'. Rather than argue over whether he should attend university, he reframed the question as a choice between college and entrepreneurship, forcing his parents to weigh trade-offs instead of issuing commands. He showed them bank records proving he could support himself. The reframe worked—they let him go.
— Li XiangOnly two organizational models win globally; he is building a third
The world's winning organizations fall into two categories: those built on strict process (Huawei, IBM) for complex products where any broken link fails the whole system, and those built on high talent density plus simple tools (Bytedance, Google) for platforms and models. Li Auto tries something different—treating talent density as the body and validated best practices as tools, not structure, to handle both supply-chain complexity and AI software simultaneously.
— Li XiangRange anxiety forced the counterintuitive choice of extended-range hybrid
Industry consensus held that extended-range was backward. Li Xiang's decision didn't come from market research—it came from real experience. Driving a Model S to a lakeside resort, he encountered a broken charging station; he barely limped back on downhill momentum. His team resisted, wanting a pure plug-in hybrid instead. He showed them Nissan's e-power, which uses less than one kilowatt-hour and drives extended-range the entire way, yet feels as smooth as a pure electric. That proof convinced them.
— Li XiangThe investor he'd rejected a year earlier rescued him from fundraising ruin
During Ideal One's production phase, Li Xiang met 150 investors. None committed. Tesla's valuation had crashed to $30 billion, NIO's stock had fallen from 6-7 yuan to around 1 yuan, and the sector was in free fall. Li Xiang was so ill he needed an hour on an airport lounge chair before each meeting. Wang Xing (Meituan founder) had once asked to invest; Li Xiang refused, wanting to avoid Didi's control. This time, Li Xiang simply chatted with him—never asking for money, only discussing value. Wang Xing spontaneously offered 300 million yuan.
— Li XiangHis five-million-yuan luxury-car claim came from test-driving, not positioning
Li Xiang's claim that the Ideal One was "the best family SUV under 5 million yuan" drew fire from reviewers. The genesis wasn't marketing strategy—it was empiricism. He test-drove cars worth 3 to 10 million yuan owned by industrialists he knew, and found them disappointing. He concluded that at that price point, a car solves social identity problems, not transportation. Wealth simply signals status (e.g., stranger approach at nightclubs). This insight later became L9 marketing copy.
— Li XiangCompetitors orchestrated the Mega 'coffin car' scandal, not grassroots criticism
After three years in development, the Ideal Mega launched into coordinated "coffin car" attacks. The design team was devastated; some urged an immediate redesign. Li Xiang assessed it as orchestrated black PR—he knows people arrested for similar campaigns. Modern black PR no longer relies on extortion; it monetizes traffic itself, and many operators aren't even from the car industry. The Mega did have genuine issues: the interior copied the L series with no new experience and higher pricing. He fixed it later with a rotating seat and price cuts, but never directly rebutted the coffin car narrative.
— Li XiangTop founders don't succeed by picking right; they win through inhumanely frequent iteration
Li Xiang identifies three traits among top founders he's observed. First two are obvious: picking the right category, and persisting long-term. The third is completely inhuman—facing real market feedback at extreme frequency. Huang Renxun releases one GPU generation per year while competitors do one every three. Bytedance runs dozens of parallel versions. Meituan adjusts operations so frequently outsiders never see it. This is reinforcement learning: pretraining matters, but growth comes from constant real-world iteration.
— Li XiangA shareholder coup taught him that asking for help matters more than dying alone
In 2008, three minor shareholders—one had just borrowed from Li to buy a house—cornered him in a teahouse, demanding he and co-founder Fanzhen resign, keeping only 15% equity. They blamed cash-flow decline after hiring Qin Zhi and blamed failed fundraising. An angel investor backed Li and stopped them. Weeks later, the ringleader revealed the real reason: Li had been grinding alone, never consulting the team, making everyone feel disposable. This taught Li two lessons: tell core management immediately when trouble hits, and stop punishing himself to isolation.
— Li XiangIn their own words · checked verbatim
I didn't argue with my parents about whether to go to college. I asked them: college or starting a business? I turned it into a choice problem, not a right-or-wrong problem.
我并没有跟父母在讨论是否上大学这件事 我跟父母讨论的是上大学还是创业这个事 就是跟父母一起来做个选择题 不是做个对错题
Li Xiang27:15
There are only two successful organizational models in the entire world. One has extremely strict process—because the category is complex, it's bottleneck theory.
全世界的组织方式最成功的就两大类 嗯 这个一大类是有特别严格的流程 对吧 因为可能他这个品类很复杂是个木桶理论
Li Xiang1:07:37
I showed them something completely different: Japan's best-selling car is Nissan's e-power, which uses an extended-range structure. It uses less than one kilowatt-hour, can't drive purely electric—pure extended-range the whole way.
我向他们 让他们看了一件事情 其实完全不一样 就是在日本销量最好的车 是日产的e-power 用的是增程结构 但是它用的是一个 不到一度电的 然后不能纯电驱动的 纯粹的一个增程结构 全程都要增程
Li Xiang1:34:50
After I was done, I didn't ask him for anything—just chatted casually. We'd meet at the hotel lobby during lockdown when many places were closed. I didn't bring up funding. Then at the end he just said, 'Can I invest?' I was shocked.
然后讲完以后 我都没找他要通知 我只是找他聊一聊 而且我们聊的地方 就是经常在那遇到你 就是那个 什么饭馆 凯越 在房间凯越 酒店大堂 然后跟他聊的 对吧 因为那时候疫情 你知道聊的就是 然后很多地方去不了 我都没提 他最后来就说 我能投你们吗 我当时就惊了
Li Xiang1:43:53
Later I looked it up myself online, did the research. When a car costs several million yuan, it's not solving a transportation problem at all—it's solving a completely different kind of problem.
然后后来我自己甚至上网 动手找答案 看了一些东西 然后大概意思是说 就一个车很不正常的 卖到几百万以后呢 它解决的根本不是交通工具问题 它是解决其他的功能问题
Li Xiang2:03:02
Now they're just using you for clicks. They monetize the traffic itself, so they're not even necessarily from the car industry.
现在是拿你做流量 靠别的 靠流量变现去了 因为流量是本身就可以变现了 所以他也不一定是 很多都不是汽车行业的
Li Xiang2:19:07
Look at Huang Renxun—most people can't do this, absolutely can't. When I was in graphics cards, it was obvious: he releases a new generation every year; competitors do one every three years.
然后你看包括黄仁勋 然后因为 这个应该是绝大多数人做不到的 完全做不到 我做显示卡的时候就很明显 就是他一年一代产品 别人三年一代产品
Li Xiang2:45:23
He said, 'You were always grinding alone. While you were grinding, we felt like you thought we didn't matter. We fought through this with you. You weren't trying to build together with us. You stopped consulting us on any problems.'
他说你老一个人死扛 然后在你死扛的时候 我们最大感觉就是你觉得我们不重要了 我们跟你一起拼过来的 你没有试图跟他们一起共同成的 你遇到任何困难都不跟我们商量了
Li Xiang3:16:37
Figures
| Monthly income at 18 compared to parents | Over 20,000 yuan, roughly 10× their income | 20:16 |
| First computer price | 8,000 yuan | 6:10 |
| Telecom Australia's stake in Autohome | 55% | 49:25 |
| Ideal One lifetime sales and revenue | 200,000 units, roughly $10 billion | 2:10:02 |
| Tesla valuation during funding crisis | Dropped to $30 billion | 1:36:51 |
| NIO stock price during funding crisis | Fell from 6-7 yuan to around 1 yuan | 1:36:51 |
| Investors Li Xiang met during Ideal One fundraising | 150 | 1:36:51 |
| Wang Xing's investment in Ideal | 3 billion yuan | 1:44:53 |
| Autohome's 2013 IPO launch and first-day gain | $13 IPO price, rose to over $30 on day one | 3:23:39 |
Glossary
- GTD
- Task-priority time management: A system centered on listing tasks and ranking by priority.
- PHEV
- Plug-in hybrid: A hybrid where the engine drives the wheels directly, not just powering a generator.
- e-power
- Nissan's extended-range system: The engine only powers a generator; it cannot drive the wheels directly.
How to listen
Founders and investors interested in how repeat entrepreneurs design organizations, make technical choices, and navigate brand and PR decisions.
1:50-2:02, detail on luxury car reviews and different model pricing—doesn't affect the main arguments.