laitimes

Talking to He Xiaopeng about the transformation of large models for autonomous driving: building a car is like swimming in a sea of blood, but I am more confident

Talking to He Xiaopeng about the transformation of large models for autonomous driving: building a car is like swimming in a sea of blood, but I am more confident

Zhang Xiaojun's business interview

2024-07-16 08:00Published in Beijing, Tencent News, original creator, creator in the financial field

The full text is 1372 words, and it takes about 4 minutes to read, so help me mark the key points

Highlight the point

01 He Xiaopeng, chairman and CEO of Xpeng Motors, said that China needs a general model company, but it needs a company that can read the model and put the model into the application of the project.

02 He believes that building a car is like "swimming in a sea of blood", which requires continuous innovation and change.

03He Xiaopeng shared the fresh experience of test driving Tesla FSD and Waymo in United States, and believes that large model companies need to make money in another dimension, that is, autonomous driving.

04In addition, he also discussed the future development of Xpeng Motors, including naming, investors, and joining Xpeng Motors.

05 Finally, He Xiaopeng said that Xiaopeng Motors will become one of the companies that will put large models into applications for engineering landing.

The above content is generated by Tencent's hybrid model and is for reference only

Many people say that large models will transform autonomous driving and accelerate the arrival of an inflection point, but how exactly does it happen?

In this episode, I invited He Xiaopeng, chairman and CEO of Xpeng Motors.

In the past one and a half years, as an automotive CEO, He Xiaopeng has done a lot for the big model — he asked the team to read papers, visited the United States three times, test-drove Tesla FSD and Waymo, and visited Huang Jenxun. In his view, China needs a general model company, but China also needs a company that can read the model and put the model into the application of the project.

Standing at the node of the 10th anniversary of the establishment of Xpeng Motors, I also discussed with him friendship, business wars, peaks and Waterloo.

He described building a car as "swimming in a sea of blood".

Talking to He Xiaopeng about the transformation of large models for autonomous driving: building a car is like swimming in a sea of blood, but I am more confident
Talking to He Xiaopeng about the transformation of large models for autonomous driving: building a car is like swimming in a sea of blood, but I am more confident
Talking to He Xiaopeng about the transformation of large models for autonomous driving: building a car is like swimming in a sea of blood, but I am more confident

Our podcast program was first published on Tencent News, so you can go and follow it, so that you can get the program information and more news information as soon as possible :)

Talking to He Xiaopeng about the transformation of large models for autonomous driving: building a car is like swimming in a sea of blood, but I am more confident

Friendship and stock prices

02:11 Why did Lei Jun invite you, Li Xiang and Li Bin to attend the Xiaomi SU7 launch event at the same time?

03:12 They said, "I guess you're dizzy, crying in the toilet," and they made me laugh

03:25 At that press conference, Li Xiang, He Xiaopeng, Li Bin, and Wei Jianjun were seated

04:17 What do we sit down and discuss? Why am I laughing?

05:50 When did you find out that the stock price of Xpeng Motors has fallen? What's the mood?

06:05 My opinion of Lei Jun and his first car

Genes and limitations

06:27 The source and gene of Xpeng Motors as intelligent driving

10:30 I have accumulated knowledge from starting a business and doing UC Browser

14:14 Wei Xiaoli is an entrepreneur who entered the Internet to build cars, but his genes and thinking are completely different

15:56 My reflection: If I had a car brand logic, I would cut a little higher; If I understand the automotive system better, I will start with the P7 first, and then move on to the system of maybe G7

17:25 On the naming of "Xpeng Motors" and from investors to joining Xpeng Motors

20:18 The psychological highs, lows, and apprehensions of the 10 years of UC entrepreneurship

Large-scale models transform autonomous driving

21:32 A Realistic Question: Can Autonomous Driving Help You Sell Cars?

23:33 No one in the big model company makes money, what it can make money in another latitude is autonomous driving

24:33 Autonomous driving has made a large model, and it also needs to have "brakes" and "goalkeepers"

25:28 What did I do after ChatGPT came out? Read papers, travel to United States, visit Jensen Huang

26:00 The big model is not the thinking logic of people, it is the thinking logic or reasoning logic of the machine

27:25 Software is used quickly from scientific research to scale, and hardware is extremely difficult

29:58 How does the large model land on the car? According to its logic, we rewrite our data flow logic, increase the collection of data, do pre-training, and what is written at the beginning is completely unusable, and requires a huge amount of training

32:58 Will you talk to the CEO of the car company, Li Xiang, Li Bin, and Lei Jun about the big model?

33:20 China needs a general model company, but China also needs a company that can read the model itself and put the model into application for engineering implementation - Xpeng is one of them

35:10 What do you think Musk is doing both Tesla and xAI

36:54 Inspiration for Wu's departure and visit to Jensen Huang

enemies and moats

40:32 Today, with Huawei's entry into the game, its intelligent driving ability is also considered to be the first echelon, and its style is so fierce, do you feel the impact?

41:32 Some people say that Xiaopeng should not sell cars at all, and learn from Huawei to be an intelligent provider

42:45 Big models are bad for small companies: the stronger the big AI, the more data, the more data, the more fees, the more costs, the more cars to sell to support the system, this is a network effect

46:32 How much does it cost, how much time, and what are the basic elements to replicate Tesla's end-to-end FSD (Full Self-Driving)?

47:40 Will the future of multimodal large models, autonomous driving models, and robot brains be unified?

Troubled times are coming

48:59 "Whoever the brothers want to harm, let him build a car"

51:40 "I just feel like I'm swimming in a sea of blood"

53:31 Don't "kick the accelerator and brake" on the human side

55:49 G9 accident reflection and the difficulty of change

57:00 We have no reporting relationship with President Wang Fengying

01:00:28 It's not a strong man who breaks his wrist, it's "a strong man who breaks his head"

01:01:35 Knockouts, All-Star breaks, my goal is to be number one

01:03:01 The combination of vertical and horizontal links is frequent, and troubled times are about to appear

01:05:44 Car companies are like "playing cards", why did Xpeng release the MONA M03 card this year?

01:10:33 Me in Change: A little tantrum recently, from democracy to centralization, focusing on 2+3 issues

01:14:00 He Xiaopeng recently test-drove Tesla FSD and Waymo in United States, fresh experience and judgment

Straight to the past

Oral description of the global model in the past six months: Perplexity suddenly became popular and the AI application ecosystem that has not yet exploded

The first person to vote for MiniMax: MiniMax, GenAI Conference, and Big Model Playing Cards

Podcast Update|Talk to Li Zhifei about IPO, GPT-4o and the pain of cutting-edge technology entrepreneurship that you don't know

View original image 141K

  • Talking to He Xiaopeng about the transformation of large models for autonomous driving: building a car is like swimming in a sea of blood, but I am more confident
  • Talking to He Xiaopeng about the transformation of large models for autonomous driving: building a car is like swimming in a sea of blood, but I am more confident
  • Talking to He Xiaopeng about the transformation of large models for autonomous driving: building a car is like swimming in a sea of blood, but I am more confident
  • Talking to He Xiaopeng about the transformation of large models for autonomous driving: building a car is like swimming in a sea of blood, but I am more confident

Read on