来源：notes/sources/bilibili-2026-09-30-yu-kai-horizon-in-good-company/2026-09-30-yu-kai-horizon-in-good-company.bilingual.md

Horizon Robotics CEO 余凯 In Good Company 播客访谈（Nicolai Tangen 主持）

Speaker: 余凯（Dr. Kai Yu，Horizon Robotics 创始人 / CEO）、Nicolai Tangen（Norges Bank Investment Management CEO）
Ingest Date: 2026-09-30
Source: https://www.bilibili.com/video/BV18Aho6xEsE/
Duration: 字幕覆盖 00:00 - 00:47:04（视频总长约 94 分钟，本 srt 为已下载段落）
Subtitle Type: 英文 SRT（外部下载，773 条）
Format: 精校版（已加标点断句 + 轻度清洗口头语 + 修 ASR 错字；非逐字稿，原始逐字稿见同名 .original.srt / .original.md）

开场白

Yu Kai (opening teaser): In some sense, this is a terrible business — a heavy investment, very slow return. But in my view, sometimes a terrible business means a great business, because the entry barrier for all the other newcomers is extremely high.

余凯（开场预告）：从某种意义上看，这是一门糟糕的生意——投入重、回报极慢。但在我看来，有时候糟糕的生意恰恰意味着伟大的生意，因为对所有后来者的进入门槛都极高。

Nicolai: Hi everybody, I’m Nicolai Tangen, the CEO of the Norwegian Sovereign Wealth Fund. Ten years ago, we all assumed self-driving cars would come out of Silicon Valley. But today, the most advanced cars on the road are Chinese, and Western car makers are going to China to buy the technology.

尼古拉：大家好，我是尼古拉·唐恩，挪威主权财富基金的首席执行官。十年前，我们都以为自动驾驶汽车会出自硅谷。但今天，路上最先进的汽车是中国的，而西方车企正跑到中国去买技术。

Nobody sits closer to this shift than my guest, Yu Kai, who’s the founder and CEO of Horizon Robotics — which supplies the chips and software that let Chinese cars drive themselves. One of China’s leading AI scientists before he became an entrepreneur, few people can explain better how China got there and where it goes next. So, Kai, welcome.

没有人比我的嘉宾余凯更贴近这场变局。他是地平线机器人（Horizon Robotics）的创始人兼 CEO，这家公司提供的芯片和软件让中国汽车得以自动驾驶。在成为创业者之前，他是中国顶尖的人工智能科学家之一，很少有人能比他更清楚地解释中国是如何走到这一步的、接下来又会走向哪里。那么，余凯，欢迎你。

Yu Kai: Hi, Nicolai.

余凯：你好，尼古拉。

一、地平线做什么

Nicolai: Let’s just start with a simple question. What does Horizon Robotics do?

尼古拉：我们先从一个简单的问题开始吧。地平线是做什么的？

Yu Kai: Well, essentially what we are doing is the computing platform for all generic robots — so including self-driving cars. But of course, our current business is nearly 100% focused on passenger vehicles.

余凯：本质上，我们做的是面向所有通用机器人的计算平台——当然也包括自动驾驶汽车。不过目前，我们的业务几乎 100% 聚焦在乘用车上。

Nicolai: So if I sit in a car that you supply into, what does your chip do?

尼古拉：那如果我坐在一辆用了你们产品的车里，你们的芯片在做什么？

Yu Kai: Essentially it’s running the software all the way from camera perception to trajectory planning and control. It’s kind of like the brain for all the cars.

余凯：本质上，它运行着从摄像头感知一直到轨迹规划与控制的整套软件。它有点像所有汽车的大脑。

Nicolai: Now your chips run roughly one out of four cars in China.

尼古拉：现在中国每四辆车里大约就有一辆用的是你们的芯片。

Yu Kai: Just tell me — one out of three. I think that’s the latest track record. Thank you.

余凯：你尽管说——是每三辆车里就有一辆。我想这是最新的成绩。谢谢。

Nicolai: Who are your customers?

尼古拉：你们的客户是谁？

Yu Kai: The major Chinese OEMs — like BYD, Geely, Chery, Changan, and Great Wall. And also some of the global OEMs — like Volkswagen, Toyota, Suzuki. Quite a few local domestic and global players.

余凯：中国的主要整车厂——比如比亚迪、吉利、奇瑞、长安和长城。也有一些全球车企——比如大众、丰田、铃木。本土和全球的玩家都相当不少。

Nicolai: Who do you compete with?

尼古拉：你们和谁竞争？

Yu Kai: Oh, very good question. Mobileye, Huawei, and quite a few domestic players like Black Sesame — I’m not sure if you’ve ever heard of that. And Qualcomm, MediaTek, Texas Instruments, Renesas, NXP. Quite a lot of global and China technology companies.

余凯：哦，这个问题问得好。Mobileye、华为，还有不少像黑芝麻这样的本土玩家——不知道你有没有听说过。另外还有高通、联发科、德州仪器、瑞萨、恩智浦。全球和中国的科技公司都很多。

二、为什么是地平线

Nicolai: Why are you the best?

尼古拉：你们为什么是最好的？

Yu Kai: A very good question. Really, because we are very focused. We focus on this single one thing — to enable automatic driving, all the way to robotics. We are not doing anything else. So the intensity of focus is one thing. The other thing, fundamentally, is our vision and mission.

余凯：这个问题问得很好。真的，是因为我们非常专注。我们只专注这一件事——赋能自动驾驶，一直延伸到机器人。别的什么都不做。所以专注度是一方面。另一方面，更根本的，是我们的愿景和使命。

If you look at the name of our company — eleven years ago when we founded this company, Horizon Robotics — back then in China there was even no such a business at all; even the EV business was just starting. We had this vision, which was not an existing opportunity. We envisioned that in the future the world will be full of all kinds of different robots — working in a farming field, on a manufacturing assembly line, in a logistics warehouse, household service, elderly care, and of course self-driving cars. So we envisioned there should be a unifying platform, just like Windows and Intel did for personal computing, just like ARM and Android did for mobile computing. We envisioned a future computing platform that is generic for all kinds of robotics across different verticals.

看看我们公司的名字——十一年前我们创办这家公司，取名地平线机器人——那时候在中国，这根本还谈不上是一门生意，连电动车行业都才刚起步。我们有这样一个愿景，而它当时并不是一个现成的机会。我们预见到，未来世界将充满各种各样不同的机器人——在农田里、在制造业流水线上、在物流仓库里、在家庭服务中、在养老照护里，当然还有自动驾驶汽车。所以我们预见到，应该会有一个统一的平台，就像 Windows 和英特尔之于个人计算，就像 ARM 和安卓之于移动计算。我们预见到一个面向不同垂直领域、通用于各类机器人的未来计算平台。

Nicolai: When exactly did the vision hit you? In the shower, or working on a mountain?

尼古拉：这个愿景具体是什么时候击中你的？洗澡的时候，还是在山上工作的时候？

Yu Kai: That’s a very good question. The vision actually hit me — I didn’t find this opportunity, I found this vision. I’ve been working in machine learning for over thirty years. Since my early days as a junior student in college, I by chance read a book about machine learning, and I was so amazed that it’s so fantastic: to develop a computer algorithm that one day can be like a human brain, getting smarter and smarter through experience, by learning more and more from data. So I decided to devote my whole career into this single subject — machine learning.

余凯：这个问题问得很好。其实这个愿景是击中了我——我没有找到这个机会，我找到的是这个愿景。我在机器学习领域工作了三十多年。大学三年级时，我偶然读到一本讲机器学习的书，当时我被震撼到了：这太不可思议了——去开发一种计算机算法，有朝一日能像人脑一样，通过经验、通过从数据中不断学习，变得越来越聪明。于是我决定把整个职业生涯都投入到这一件事上——机器学习。

That was for my undergraduate thesis, my master’s thesis, my PhD, and my first job back at Siemens — and later in Silicon Valley, joining NEC Labs, and then joining Baidu, and then founding Horizon Robotics. I changed locations, I changed roles — from individual researcher, to team leader, to head of a research lab, to founder of a technology company. But there’s one thing that never changed: my mission and my vision to develop machine learning algorithms to make machines getting smarter and smarter.

这贯穿了我的本科毕业论文、硕士论文、博士研究，以及我的第一份工作——回到西门子；后来去硅谷加入 NEC 实验室，之后加入百度，再之后创办地平线。我换过地方，换过角色——从独立研究员，到团队负责人，到研究院院长，再到一家科技公司的创始人。但有一件事从未改变：我的使命和愿景——开发机器学习算法，让机器变得越来越聪明。

三、为什么不自己造车

Nicolai: Two companies we’ve had on the podcast — XPeng and Xiaomi — they also make phones, cars, in addition to chips. Why did you decide never to build a car?

尼古拉：有两家上过我们播客的公司——小鹏和小米——它们除了芯片，也做手机、做汽车。你为什么决定绝不造车？

Yu Kai: I always position ourselves as going to be like Microsoft and Intel — not for PC, but for robots. I think a platform, an enabling platform, is a much more interesting position, not just in terms of the business model, but also in terms of the impact you can make to the whole world.

余凯：我一直把我们定位成要做微软和英特尔那样的公司——不是为 PC，而是为机器人。我认为平台、一个赋能平台，是更有意思的位置，不仅从商业模式上说，也从你能对全世界产生的影响上说。

For example, back in the days of personal computing — Windows and Intel, they are not just enabling Lenovo or Dell; they actually helped the entire mankind, a new kind of technology civilization, a new era of technology. The personal computer enabled all the white-collar people to dramatically increase their productivity — that’s an amazing impact. XPeng, they actually maybe help one brand to become very appealing, very attractive to consumers. But I want to enable and help other brands, other cars.

举个例子，在个人计算时代——Windows 和英特尔不只是成就了联想或戴尔，它们实际上帮助了全人类，开启了一种新的技术文明、一个新的技术时代。个人电脑让所有白领的生产力大幅提升——这是了不起的影响。小鹏也许能帮助一个品牌变得非常有吸引力、非常受消费者欢迎。但我想赋能并帮助更多的品牌、更多的车。

Nicolai: Is it difficult to build a car?

尼古拉：造车难吗？

Yu Kai: Yeah, of course.

余凯：当然难。

Nicolai: But you have the technology, you have the chips. If I asked you to build a car, how long would it take?

尼古拉：但你有技术，有芯片。如果我要你造一辆车，需要多长时间？

Yu Kai: I wanted to achieve “Horizon Inside.” I think that’s a much bigger dream, much more interesting for me. I don’t want to be a car maker; I want to help more car brands to become more attractive.

余凯：我想实现的是“Horizon Inside”。我觉得那是一个大得多的梦想，对我来说有意思得多。我不想做汽车制造商；我想帮助更多的汽车品牌变得更有吸引力。

四、中国市场的残酷竞争

Nicolai: Now the Chinese car market is brutally competitive. What is driving the Chinese car market just now? Is it price, intelligence?

尼古拉：现在中国汽车市场的竞争极其残酷。眼下驱动中国汽车市场的是什么？是价格，还是智能化？

Yu Kai: The current Chinese car market is really crazy. You don’t see any market like this anywhere. There are probably sixty to seventy different car brands, and for the first half of this year, more than five hundred eighty new car models launched. It’s an extremely tough market.

余凯：现在的中国汽车市场真的很疯狂。你在别的地方看不到这样的市场。大概有六七十个不同的汽车品牌，而今年上半年，上市的新车型超过五百八十款。这是一个极其艰难的市场。

But the market is really driven by competition. On the other hand, it’s also driven by consumer demand. Consumers are expecting more and more smart features — automatic parking, autonomous driving, more fancy entertainment assistance. All of those factors are driving the market, not just competitive but also very innovative.

但这个市场确实是被竞争驱动的。另一方面，它也是被消费者需求驱动的。消费者期待越来越多的智能功能——自动泊车、自动驾驶、更花哨的娱乐和辅助功能。所有这些因素都在推动市场，不仅是竞争激烈，而且非常有创新力。

Nicolai: What does it do to Chinese manufacturers, when they have such strong competition at home — when they get out to, let’s say, Europe, is it easy-peasy to compete?

尼古拉：当中国厂商在国内面临如此激烈的竞争，这对它们意味着什么——当它们走出去，比如去欧洲，竞争是不是轻而易举？

Yu Kai: We position ourselves not first — we are not competing with any car makers; we are actually enabling them, helping them. We are building strong partnerships with all the car makers and ecosystem partners, including automotive suppliers — for example, Bosch, Denso, and many other global and local suppliers.

余凯：先说我们的定位——我们不与任何车企竞争；我们实际上是在赋能它们、帮助它们。我们与所有车企和生态伙伴建立牢固的合作关系，包括汽车供应商——比如博世、电装，以及许多其他全球和本土供应商。

Our philosophy is really to make friends and build long-term trust with our customers and partners. Essentially, this is fundamentally in the gene of our company. We don’t have too much culture from the competition perspective; we’re more about what kind of value we can create.

我们的理念就是广交朋友，与客户和伙伴建立长期信任。从根本上说，这已经刻在我们公司的基因里。我们没有太多从竞争出发的文化；我们更关心的是能创造什么样的价值。

But luckily, if you look at our business, fundamentally compared to those car makers, we are not so crowded. In China, pretty much the top three players for automotive computing have consolidated to NVIDIA, Huawei, and Horizon Robotics. I’m very proud of that — partially because we chose a position where the business has an extremely high entry barrier. It’s not like a bunch of guys with VC funding can easily get into this business, because this is all mission-critical automotive electronics; it has to have the highest standard for quality and safety.

但幸运的是，如果你看我们的业务，与那些车企相比，我们这里基本没那么拥挤。在中国，汽车计算的前三名基本上已经收敛为英伟达、华为和地平线。我为此非常自豪——部分原因是我们选择的位置有着极高的进入门槛。这不是一群拿着风投的钱的人就能轻易进入的生意，因为这些都是关键任务级的汽车电子，必须达到质量和安全的最高标准。

五、”糟糕的生意”反而是好生意

Yu Kai: Let’s say we are doing this chip design for auto models. It’s not like a consumer type of silicon; it has to go through all the regulations and tough qualification and testing processes. Normally if we initiate a new project, a new generation of project, it will take three to four years to get the product — the chip and the solution — ready for car makers. Then, if luckily enough we win a new nomination from one car maker, we need to also go through the cycle of production for the car platforms — that’s another three to five years in China, two years in Europe, a layer slower, typically five years globally. So think about it: it’s at least five to eight years for the whole cycle.

余凯：就说我们做车规芯片设计。它不像消费级芯片，必须走完所有法规流程和严苛的认证与测试。通常我们启动一个新项目、新一代项目，要花三到四年才能把产品——芯片和方案——做到可以交付车企。然后，如果运气够好，拿到某家车企的新定点，我们还要陪着走完车型平台的生产周期——在中国又是三到五年，欧洲是两年，节奏慢一层，全球通常要五年。所以你想一想：整个周期至少五到八年。

In some sense, this is a terrible business — a heavy investment, very slow return. But in my view, sometimes a terrible business means a great business, because the entry barrier for all the other newcomers is extremely high. And a good business is something where you can sleep well overnight. That’s my opinion.

从某种意义上说，这是一门糟糕的生意——投入重、回报极慢。但在我看来，有时候糟糕的生意意味着伟大的生意，因为对所有新进入者的门槛都极高。而好生意的标准，是你晚上能睡个好觉。这是我的看法。

六、自动驾驶的三步路线图

Nicolai: Talking about sleeping well, the autonomous journey — you speak between hands-off, eyes-off, mind-off. These are three steps a car will go through. In January last year you put dates on them. So we’ll have eyes-off in three years?

尼古拉：说到睡个好觉，在自动驾驶这条路上——你提到过 hands-off、eyes-off、mind-off（脱手、脱眼、脱脑）。这是汽车要经历的三步。去年一月你给它们都定了时间点。那么三年后我们就会有 eyes-off 了？

Yu Kai: Absolutely. I said that three years ago — we’ll have eyes-off.

余凯：绝对会。我三年前就说过——我们会实现 eyes-off。

Nicolai: Where are we today?

尼古拉：我们现在走到哪一步了？

Yu Kai: We are in very good progress. In the industry, they have the standard Level 3, Level 4, Level 5 — but from the engineering standpoint, for consumers, they have no clue.

余凯：我们进展得非常好。行业里有标准的 L3、L4、L5——但从工程角度看，对消费者来说，他们根本搞不清。

Nicolai: What is Level 3?

尼古拉：什么是 L3？

Yu Kai: I always try to connect our technology and product to something my mom can understand. That’s my way to interpret the technology stages: hands-free, hands-off, eyes-off, and mind-off.

余凯：我总是尽量把我们的技术和产品，转化成像我妈也能听懂的说法。这就是我解读技术阶段的方式：免手（hands-free）、脱手（hands-off）、脱眼（eyes-off）、脱脑（mind-off）。

Nicolai: So what would you tell your mom right now?

尼古拉：那你现在会怎么跟你妈说？

Yu Kai: Right now, we are in very good progress to achieve fully hands-off. I made a statement last year: by the end of 2028, we’re going to achieve fully hands-off; by the end of 2030, achieve fully eyes-off; and by the end of 2035, 100% mind-off — basically you can sleep in your running car. We are in very good progress.

余凯：目前，我们在实现完全脱手方面进展非常好。我去年做过一个表态：到 2028 年底，我们要实现完全脱手；到 2030 年底，实现完全脱眼；到 2035 年底，实现 100% 脱脑——基本上你可以在行驶中的车里睡觉。我们进展得非常好。

Our latest product — the software stack called Horizon SuperDrive, running on our Journey 6T SoC — is essentially fully hands-free. It’s like a smarter version of automatic transmission. Over the last twenty years, the whole car industry evolved from manual transmission, where you need to control all the gears, into automatic transmission. Now I call this getting into the smarter version of automatic transmission — you just need to stare ahead and nothing else; you need to take care.

我们最新的产品——运行在我们征程 6T SoC 上的软件栈 Horizon SuperDrive——本质上已经实现了完全免手。它就像自动变速箱的智能版本。过去二十年，整个汽车行业从需要你控制所有挡位的手动变速箱，进化到了自动变速箱。现在我把这称为进入了自动变速箱的智能版本——你只需要盯着前方，别的什么都不用做；但你仍然得留心。

Nicolai: What is lacking? What part of the technology are you seeing as the critical element you’ve left?

尼古拉：还缺什么？你认为哪部分技术是你还欠缺的关键要素？

Yu Kai: We need much bigger road tests, and we need much better numbers for statistics to prove that the technology is safe enough to claim eyes-off. We are in very good progress approaching hands-off, but we need much bigger road tests to prove eyes-off.

余凯：我们需要规模大得多的路测，也需要好得多的统计数据，来证明这项技术足够安全，可以宣称实现脱眼。我们在接近脱手方面进展很好，但要证明脱眼，还需要大得多的路测。

七、自研 AI 模型

Nicolai: Do you build your own AI models?

尼古拉：你们自己构建 AI 模型吗？

Yu Kai: Sure, the self-driving model. Remember we position ourselves as Intel — not just Intel. We position as Wintel. And I think Wintel is an amazing business model.

余凯：当然，自动驾驶模型。记住，我们把自己定位为英特尔——不只是英特尔，我们定位为 Wintel。我认为 Wintel 是一个了不起的商业模式。

Nicolai: Where is the future breakthrough going to come from? Is it the model, the data?

尼古拉：未来的突破会来自哪里？是模型，还是数据？

Yu Kai: It’s a combination of three factors together. This is the typical formula of today’s business: you need much bigger compute running locally on the vehicle; you need continuously collecting bigger and bigger data — but meaningful data, not just quantity but also quality, meaning many meaningful corner cases; and you need bigger and bigger models in terms of number of parameters.

余凯：是三方面因素结合在一起。这是当今这门生意典型的公式：你需要在车上本地运行的大得多的算力；你需要持续收集越来越大、但要有意义的数据——不只是量，还要有质，也就是大量有意义的极端场景（corner cases）；你还需要参数量越来越大的模型。

八、公司文化：不考核工时

Nicolai: How many people are in the company? What do they do?

尼古拉：公司有多少人？他们都做什么？

Yu Kai: We are a very technology-savvy company. Most of our employees are engineers — either software engineers or hardware engineers. And they work extremely hard.

余凯：我们是一家非常懂技术的公司。大部分员工都是工程师——要么是软件工程师，要么是硬件工程师。他们工作极其努力。

Nicolai: You don’t produce the chips; you outsource the production, right?

尼古拉：你们不生产芯片，生产是外包的，对吧？

Yu Kai: Actually, on the hardware side we are a typical fabless company. We design the chip but we don’t manufacture the chip; we outsource to our partners. We also developed the software, the end-to-end neural network for self-driving.

余凯：其实在硬件方面，我们是一家典型的无晶圆厂（fabless）公司。我们设计芯片，但不制造芯片，制造外包给合作伙伴。我们同时也在开发软件，也就是用于自动驾驶的端到端神经网络。

Nicolai: What does “working hard” mean?

尼古拉：“工作努力”具体指什么？

Yu Kai: We have a typical technology culture. We don’t have rules or policies like “you need to come to work at 8 AM.” Actually, typically most of our engineers come to work around 10 AM, and they freely organize their working time. There’s no rule like “before which time you are not supposed to leave the office.” The sense of ownership of our engineers is very high.

余凯：我们有典型的技术公司文化。我们没有那种“你必须早上 8 点来上班”的规定或制度。实际上，我们大多数工程师通常上午 10 点左右来上班，他们自由安排自己的工作时间。也没有“几点之前不许离开办公室”这种规定。我们工程师的主人翁意识非常强。

Nicolai: How long days do you think on average they work?

尼古拉：你觉得他们平均每天工作多长时间？

Yu Kai: Actually I have no clue. Quite some of our engineers even work from home. We don’t measure that by working hours. I know some Chinese companies measure by working hours; we never do. We care about how far we can achieve.

余凯：其实我也不清楚。我们有不少工程师甚至在家办公。我们不按工时来衡量。我知道有些中国公司按工时考核；我们从不这样。我们关心的是能走多远、能做成什么。

九、中国速度：两年 vs 五年

Nicolai: You said the model cycle time in China is two years, whereas it’s five years in Europe. What explains the difference?

尼古拉：你说车型的开发周期在中国是两年，而在欧洲是五年。这个差异怎么解释？

Yu Kai: As far as I know, because I work with so many Chinese OEMs — yes, they are working extremely hard. They have a very strong sense of competition that urges them to try all the ways to improve the process. They don’t have summer vacations like thirty or forty working days during the summer. They work really hard.

余凯：据我所知——因为我和这么多中国车企合作——是的，他们工作极其努力。他们有非常强烈的竞争意识，驱使他们想尽办法改进流程。他们没有那种夏天三四十个工作日的暑假。他们真的非常拼。

Nicolai: What’s the key to innovation? How do you create a really innovative culture?

尼古拉：创新的关键是什么？你如何打造一种真正有创新力的文化？

Yu Kai: First of all, we are a company driven by vision; we are not opportunity-driven, we are not competition-driven. We work on this because we really want to push the boundary of the entire society and industry. Self-motivation is very important. We don’t want to be the same as any other company; we don’t want to do the sort of things other companies are doing. That’s also why our business is not so crowded in China.

余凯：首先，我们是一家由愿景驱动的公司；我们不是机会驱动，也不是竞争驱动。我们做这件事，是因为我们真的想推动整个社会和行业的边界。自我驱动力非常重要。我们不想和任何其他公司一样；我们不想做别的公司在做的那类事情。这也是为什么我们在中国的生意没那么拥挤。

This is a real phenomenon: in China, if any business looks even somewhat interesting — not even profitable, just somewhat interesting — typically hundreds of companies jump in to compete. It’s crazy. Look at AI, embodied AI — so many companies, and so many car companies. We are working on something we believe in, but not necessarily other people love to jump into.

这是一个真实的现象：在中国，只要哪门生意看起来稍微有点意思——甚至不用赚钱，只要有点意思——通常就会有几百家公司一拥而上。太疯狂了。看看 AI、具身智能——那么多公司，还有那么多车企。我们做的是我们相信的事，但不一定是别人爱扎堆去做的事。

Nicolai: With all these companies jumping into these sectors, what does it mean in terms of the supply chain and the choices you have as a manufacturer? How does it all go together?

尼古拉：这么多公司涌入这些领域，对供应链、对你作为制造方的选择意味着什么？这一切是怎么拼到一起的？

Yu Kai: We just choose the best to work with, because we don’t have any luxury — something we need to be very innovative on, for example the chip architecture, the compiler, the algorithm. Some things we respect the tradition very much, because this is really a life-and-death, mission-critical matter: if any car accident happens, it’s not just impacting the driver or passenger; it’s sometimes a public safety issue. So we need to be religiously respecting tradition. We just choose the best partner.

余凯：我们就选最好的来合作，因为我们没有可以奢侈的余地——有些地方我们必须非常创新，比如芯片架构、编译器、算法。而有些东西我们非常尊重传统，因为这真的是生死攸关、关键任务级的：一旦发生车祸，影响的不仅是驾驶员或乘客，有时还是公共安全问题。所以我们必须近乎虔诚地尊重传统。我们就选最好的合作伙伴。

十、中国供应链：德国朋友的”仙境”

Nicolai: What advantage do you think Chinese auto manufacturers now have from that breadth and depth of supply chain, which we don’t have in Europe?

尼古拉：你认为中国车企现在从这种供应链的广度和深度中获得了什么优势，而这是我们在欧洲所没有的？

Yu Kai: I worked very closely with many global automotive companies like Bosch, Denso, Volkswagen. I talked to them about that. It turned out that over the last maybe two decades, China has become a country where the supply chain of components and technologies is so rich.

余凯：我和很多全球汽车公司合作非常紧密，比如博世、电装、大众。我和他们聊过这一点。事实证明，在过去大概二十年里，中国已经成为一个零部件和技术供应链极其丰富的国家。

One of my friends, a German, told me he thinks China nowadays is like a wonderland for supply chain: no matter what kind of new ideas you have, you can always find partners and suppliers in China — either to supply the technology component to you, or to work with you to fulfill your requirement. And the speed is incredible. That gave China an overall fantastic environment to innovate. The kind of overall ecosystem is there.

我有一个德国朋友，他跟我说，他觉得如今的中国就像供应链的仙境：不管你有什么样的新想法，总能在中国找到伙伴和供应商——要么给你供应技术部件，要么和你一起把需求做出来。而且速度快得惊人。这给了中国一个整体上极佳的创新环境。那种整体生态就在那里。

You don’t need to write an email to somewhere else; you just make a phone call or send a WeChat message, or take a flight from Beijing to Shenzhen — just three hours. Including myself and all of my friends, we travel so frequently within China; we take the whole of China as one village.

你不需要给别的地方写邮件；打个电话、发条微信，或者从北京飞深圳——只要三个小时。包括我自己和我所有的朋友在内，我们在中国国内出行非常频繁；我们把整个中国当作一个村子。

十一、大众裁员：中欧需要新心态

Nicolai: When you see news such as Volkswagen laying off one hundred thousand people, what reflections do you have?

尼古拉：当你看到大众裁员十万人的新闻时，你有什么感想？

Yu Kai: We position ourselves as a strong partner to Volkswagen and many other global companies, because we don’t face any competition. In Europe, if you look at semiconductor companies like Infineon, NXP, and STMicroelectronics — we are not competing with them, because their focus is on analog chips, MCUs, those kind of strengths. But we are working on computing for cars, which is exactly what they’re not doing. So in that sense, we find very strong synergy to work together.

余凯：我们把自己定位为大众以及许多其他全球公司的强大伙伴，因为我们不构成竞争关系。在欧洲，你看英飞凌、恩智浦、意法半导体这些半导体公司——我们不和它们竞争，因为它们专注的是模拟芯片、MCU 这类强项。而我们做的是汽车计算，恰恰是它们不做的。所以从这个意义上说，我们之间有非常强的协同空间，可以合作。

In my view, both European friends and the Chinese need to change the mindset. For example, several decades ago, the general pattern was global companies trying to make money from China. Nowadays that motive is getting very difficult. And now there could be some other mindset: Chinese companies now have very competitive products, we get into Europe to take their market shares and make money from them — that’s also not healthy.

在我看来，欧洲朋友和中国人双方都需要转变心态。比如，几十年前，普遍的格局是全球公司试图从中国赚钱。如今这种想法已经越来越难实现。而现在可能又出现另一种心态：中国企业现在有非常有竞争力的产品，我们进入欧洲去抢它们的市场份额、从它们身上赚钱——这也不健康。

We all need to embrace a new mindset. The new opportunity is for us to work together, to explore new markets, to launch more competitive products, to make profit, to create value globally. This is the kind of mindset I share deeply with many of my friends in Germany, in France, and in other countries.

我们都需要拥抱一种新的心态。新的机会在于我们一起合作，去开拓新市场，推出更有竞争力的产品，赚取利润，在全球创造价值。这种心态，我和我在德国、法国以及其他国家的许多朋友都深有共鸣。

十二、车就是机器人

Nicolai: Your company is called Horizon Robotics. How far is a car from a robot?

尼古拉：你的公司叫地平线机器人。汽车离机器人还有多远？

Yu Kai: Still a little long way to go, but not that far. As I said, by the end of 2028, all passenger vehicles in China will be hands-free. So it’s already kind of semi-robots.

余凯：还是有一段路要走，但也不算太远。就像我说的，到 2028 年底，中国所有乘用车都会实现免手。所以它已经算是某种半机器人了。

Nicolai: What is your business going to be in robotics?

尼古拉：你们在机器人领域的业务会是什么？

Yu Kai: Actually we are already a leading player in China right now — one of the top two players for generic robotics computing. We are enabling a lot of robotics companies, supplying computing solutions to them.

余凯：其实我们现在已经是中国领先的玩家——通用机器人计算的前两名之一。我们在赋能很多机器人公司，为它们提供计算方案。

Nicolai: What’s the overlap in technology and systems between self-driving cars and humanoid robots?

尼古拉：自动驾驶汽车和人形机器人在技术和系统上的重合部分是什么？

Yu Kai: That’s a very good question. I believe eventually the two businesses’ technology will really converge, and they eventually become a unified computing platform. Cars will become robots, no big difference. Cars will sense the environment, understand the intention of the pedestrians, surrounding traffic participants, understand the gesture of the policeman, and make decisions in very sophisticated traffic environments. Fundamentally, that’s no big difference from any other computing pattern of robots.

余凯：这个问题问得很好。我相信这两块业务的技术最终会真正融合，最终成为一个统一的计算平台。汽车会变成机器人，没有多大区别。汽车会感知环境，理解行人和周围交通参与者的意图，理解交警的手势，并在非常复杂的交通环境中做决策。从根本上说，这和其他机器人的任何计算范式都没有多大区别。

十三、人形机器人：五年内能否上量

Nicolai: Tell me about the speed of development of humanoid robots in China. What is crazy?

尼古拉：跟我说说人形机器人在中国的发展速度。有什么疯狂的地方？

Yu Kai: Again, there are like hundreds of such companies, so intense. But I think all of them are still really a long way to go. If you look at autonomous driving — Horizon is probably the first autonomous driving startup in China — we’ve already been working here for eleven years and now we are top two. And now we’re top two in robotics. For humanoid robots, I think it will take many years to go.

余凯：还是那样，这类公司有几百家，竞争非常激烈。但我觉得它们全都还有很长的路要走。看看自动驾驶——地平线大概是中国第一家自动驾驶创业公司——我们已经做了十一年，现在做到前二。现在我们在机器人领域也是前二。至于人形机器人，我认为还需要很多年。

Nicolai: What do you do inside the robot? What is your software?

尼古拉：你们在机器人里做什么？你们的软件是什么？

Yu Kai: We supply chip solutions. For robots, we are not so much into the software; we are supplying chips to the robot companies.

余凯：我们提供芯片方案。在机器人领域，我们不太涉及软件；我们向机器人公司供应芯片。

Nicolai: When do you think we’ll have meaningful volumes of humanoid robots coming? Five years?

尼古拉：你觉得什么时候会有真正上量的人形机器人？五年？

Yu Kai: I know I am a little bit conservative compared to many other guys. I try to be objective, and also partially — as a CEO, as a founder, I’m basically a technology guy — a little bit more conservative and objective.

余凯：我知道和很多人相比，我有点保守。我尽量保持客观，另外也有部分原因——作为 CEO、作为创始人，我本质上是个技术人——会稍微更保守、更客观一些。

Nicolai: What will you personally use them for?

尼古拉：你个人会用它们来做什么？

Yu Kai: I think the number one scenario is home service. We enjoy all the fun things, but we don’t enjoy many of the tedious work — like cleaning tables after dinner. Sometimes we like cooking, but we don’t like cleaning.

余凯：我觉得第一大场景是家庭服务。我们享受所有有趣的事，但不喜欢很多枯燥的活儿——比如饭后擦桌子。有时候我们喜欢做饭，但不喜欢收拾。

Nicolai: I have a theory that divorce rates will go down when you don’t have to do boring things anymore.

尼古拉：我有个理论：当人们不再需要做无聊的事情时，离婚率会下降。

Yu Kai: Exactly. There is another thing: China is facing a very terrible aging society, and it’s coming so quickly, quite unexpectedly. The issue is getting worse actually, compared to Europe and even Japan. What about elderly care? What about the labor shortage? This is going to be a huge social and economic challenge for China.

余凯：正是如此。还有一点：中国正面临非常严峻的老龄化社会，而且来得非常快，相当出人意料。实际上，与欧洲甚至日本相比，这个问题还在恶化。养老照护怎么办？劳动力短缺怎么办？这对中国将是巨大的社会和经济挑战。

十四、赢下 EV 战，就赢下机器人战

Nicolai: What’s the overlapping supply chain between self-driving cars and robotics?

尼古拉：自动驾驶汽车和机器人之间重合的供应链有哪些？

Yu Kai: I think almost the same. If you win the EV race, you’ll win the robotics race. That’s fundamentally what I believe. My argument is: if you’re a winner for the automotive supply chain, you’re going to be a winner for robotics.

余凯：我觉得几乎是一样的。如果你赢下电动车这场仗，你就会赢下机器人这场仗。这是我根本的信念。我的论点是：如果你是汽车供应链里的赢家，你就会成为机器人领域的赢家。

If you look at the actuators, mechanical things, all the sensors — lasers, radars — currently applied to cars. Because the car business is big volume, suppliers will continuously improve the cost structure, the performance, make it very compact. So those guys will be in a very strong position for robotics — just like many companies that were winners for the mobile phone supply chain later became extremely competitive for the generic IoT business.

你看现在用在汽车上的执行器、机械件、各种传感器——激光雷达、毫米波雷达。因为汽车业务量大，供应商会不断改进成本结构、性能，把它们做得非常紧凑。所以这些玩家在机器人领域会处于非常有利的位置——就像当年在手机供应链里胜出的许多公司，后来在通用 IoT 业务上也变得极有竞争力。

Nicolai: Stupid question, but where does that leave the race between China and the US in robotics?

尼古拉：问个傻问题，这样一来，中美在机器人领域的竞赛会是什么局面？

Yu Kai: I don’t like this view of looking at this game. I don’t know this kind of zero-sum game. I’m not sure this is the truth. China is very strong in manufacturing. If the robotics industry in the US — if they work with the China ecosystem, they could become more innovative and innovate at a much faster pace. If they only rely on their own supply chain or manufacturing, they will end up with slower technology innovation. So it really makes sense to always find the synergy to work together and create value for the global together.

余凯：我不喜欢用这种视角来看这件事。我不认同这种零和博弈。我不确定这是不是事实。中国在制造上非常强。如果美国的机器人产业——如果它们能和中国生态合作，它们可以变得更有创新力，创新速度也会快得多。如果它们只依赖自己的供应链或制造，它们最终的技术创新会更慢。所以，始终去寻找协同、一起合作、共同为全球创造价值，才是真正有意义的。

十五、机器人把自由还给人

Nicolai: What is a city going to be like in ten to twenty years?

尼古拉：十年到二十年后，城市会是什么样子？

Yu Kai: I’m a huge fan of ancient Greek culture and history. In July I’ve been to Athens again; I’ve been there many times. It was a very golden age, where people really enjoyed free thinking, free exchange of ideas, so many inspiring philosophers making a huge impact on human civilization. They were free of labor work. I think civilization is so much related to the freedom of humans.

余凯：我是古希腊文化和历史的超级粉丝。七月我又去了雅典；我去过很多次。那是一个非常黄金的时代，人们真正享受自由思考、自由交流思想，涌现出那么多启人心智的哲学家，对人类文明产生了巨大影响。他们从体力劳动中解放了出来。我认为文明与人的自由高度相关。

If we have all the robots to do all the tedious and sometimes dangerous work, we can have more time — like you and me right now, chatting, exchanging ideas, hanging around for other things, more time for entertainment, more time for research, more time for arts, even more time for outings. That would be great.

如果我们有机器人来做所有枯燥、有时甚至危险的工作，我们就能有更多时间——就像你我现在这样，聊天、交流想法、为别的事情闲逛，更多时间用于娱乐、用于研究、用于艺术，甚至更多时间出去走走。那就太好了。

If you look at today, the traffic jam in Beijing and Shanghai — if you ever come to China, it’s terrible. Very often it’s a one-way commute from home to office; you’re stuck in traffic for almost two hours, and you waste two hours of your life doing nothing because you need 100% focus staring at the wheel. Are you the master of the car? No, in that situation you are a slave of the car, because you don’t have choice.

看看今天北京和上海的堵车——如果你来中国，那真是太糟糕了。很多时候，从家到公司的单程通勤，你会在路上堵将近两个小时，白白浪费生命里的两个小时，什么也做不了，因为你必须 100% 专注地盯着方向盘。你是车的主人吗？不，在那种情况下，你是车的奴隶，因为你没有选择。

The essence of robotics is to keep freedom, give the choice back to humans — the human can make a choice to do something or not doing something.

机器人的本质是守住自由，把选择权还给人类——人可以自己选择做点什么，或者什么都不做。

Nicolai: And what will the city look like physically?

尼古拉：那从物理形态上看，城市会是什么样？

Yu Kai: Even today, Beijing housing prices differ from school district to school district, from region to region, quite differently. Maybe in ten years, housing prices won’t be so different, because essentially every car is like a moving house; you have much more freedom, and you don’t have such kind of anxiety about distance. So the city will be very different, drastically different in traffic.

余凯：即使在今天，北京的房价也是学区与学区之间、区域与区域之间差别很大。也许十年后，房价就不会差那么多了，因为本质上每辆车都像一个移动的房子；你自由得多，也不会再有那种对距离的焦虑。所以城市会非常不一样，在交通上会有剧烈的变化。

Today, people are fighting for the lanes, the parking lots. If everything is self-driving, the efficiency of the roads will be much higher, because the cars are coordinated rather than competing against each other. The efficiency of traffic will be ten times improved. The environment will be much nicer — much less noise, much less stress, much less anxiety. The city will be much better.

今天，人们在争抢车道、争抢停车位。如果一切都自动驾驶，道路效率会高得多，因为车辆之间是协同的，而不是互相争抢。交通效率会提升十倍。环境会好得多——噪音少得多，压力小得多，焦虑少得多。城市会好得多。

十六、雅典：文化意味着一切

Nicolai: Take me back one second to your Athens visit. When you go up to the Acropolis and stand there in front of the Parthenon, what do you think?

尼古拉：带我回到你那次雅典之行。当你登上卫城，站在帕特农神庙前，你在想什么？

Yu Kai: It’s stunning. I’m so amazed. I think culture means everything. Human thinking is unbounded. We can do something much greater than what we can imagine today. It’s such an amazing thing. And in some sense, I think it’s a shame — it’s a shame that today, in this world, we are still doing so many terrible things, making the world a huge mess. It’s a shame we are not doing anything greater than the ancient Greek people.

余凯：太震撼了。我非常惊叹。我觉得文化意味着一切。人的思维是没有边界的。我们能做出比今天所想象的伟大得多的事情。这太了不起了。而从某种意义上看，我觉得很遗憾——很遗憾在今天这个世界，我们仍在做那么多糟糕的事，把世界搞得一团糟。很遗憾我们没能做出比古希腊人更伟大的事情。

十七、个人失败、禅与当下

Nicolai: Personal question: what’s your biggest personal failure in life?

尼古拉：问个私人问题：你人生中最大的个人失败是什么？

Yu Kai: To be honest, I’m not a very pessimistic guy. If I ever encounter any frustration or failure, I always turn into a new page and almost immediately forget. I have a very good ability to forget bad things. I’m an insanely optimistic guy. And I care about the power of now; I’m following the philosophy of Zen — the Chinese ancient Buddhism philosophy — really carrying the power of now.

余凯：说实话，我不是一个很悲观的人。如果我遇到什么挫折或失败，我总是翻篇，而且几乎立刻就忘了。我很擅长忘记不好的事情。我是个极度乐观的人。我在意“当下的力量”；我遵循禅的哲学——中国古代的佛教哲学——真正践行当下的力量。

Nicolai: What’s the key to the Zen philosophy for you?

尼古拉：对你来说，禅的哲学的核心是什么？

Yu Kai: Happiness is so simple. According to one of the greatest monks in history, he said many people are having a miserable life just because: when they are supposed to sleep, they think about work; when they are working, they are thinking about sleeping. The essence of happiness is very simple: when you are working, just focus on working; when you are sleeping, just sleep.

余凯：幸福其实很简单。历史上最伟大的一位僧人说过，很多人之所以过得痛苦，就是因为：该睡觉的时候，他们在想工作；工作的时候，他们又在想睡觉。幸福的本质非常简单：工作的时候就专注于工作，睡觉的时候就睡觉。

I focus the power of now because I’m running this company, Horizon Robotics. I just focus on this and nothing else — nothing more, nothing less, just doing this whole thing.

我专注于当下的力量，因为我在经营这家公司——地平线。我就专注于此，别的什么都不管——不多不少，就是把这整件事做好。

Nicolai: What’s something about you that most people don’t know?

尼古拉：有什么关于你的事是大多数人不知道的？

Yu Kai: There’s one thing I’m not very satisfied with myself: my weight. I just have too good an appetite. I love food so much. This is really something not so good.

余凯：有一件事我对自己不太满意：我的体重。我胃口实在太好了。我太爱吃东西了。这确实不太好。

Nicolai: What do you read?

尼古拉：你读什么书？

Yu Kai: I’m reading one book — I forgot the name. It tells the story about three companies: Microsoft, Intel, and Apple. It’s about the kind of decisions and strategies they’ve been making to create the era of personal computing. Strategic thinking.

余凯：我在读一本书——名字我忘了。它讲的是三家公司：微软、英特尔和苹果。讲的是它们为开创个人计算时代所做的那些决策和战略。战略思维。

十八、给年轻人的建议

Nicolai: What’s your advice to young people?

尼古拉：你对年轻人有什么建议？

Yu Kai: Doing what you love, and loving what you’re doing — probably equally important. Many people always try to dream to do what they love, but in reality you are not always so lucky. Sometimes you need to accept, sometimes accommodate. But it’s nothing personal; stick with it. Enjoy whatever you have right now is also a very good attitude. Sometimes you love what you are supposed to do, or what you are requested to do, and then maybe this whole thing can become something you really love to do. We need to have such an attitude: love what we do, and do what we love.

余凯：做你热爱的事，热爱你正在做的事——这两者大概同样重要。很多人总是梦想着去做自己热爱的事，但现实中你并不总是那么幸运。有时候你需要接受，有时候需要适应。但这不是针对谁；坚持下去。享受你现在所拥有的，也是一种很好的态度。有时候你会爱上你应该做的事、被要求做的事，然后这整件事也许就会变成你真正热爱的事业。我们需要这样的态度：爱我们所做的事，做我们所爱的事。

Nicolai: Well, I think that’s a wonderful place to end. Thank you. What an incredible story, what an incredible company. I love your life philosophy. There is so much to take away from this.

尼古拉：好，我觉得这是一个绝佳的收尾。谢谢你。多么不可思议的故事，多么不可思议的公司。我很喜欢你的生活哲学。从这段对话里能带走的东西太多了。

Yu Kai: Thank you.

余凯：谢谢你。

原始素材