To reduce costs for car companies, chip companies are trying their best

Publisher:SereneVoyageLatest update time:2024-05-10 Source: 汽车公社 Reading articles on mobile phones Scan QR code
Read articles on your mobile phone anytime, anywhere

Reducing costs and increasing efficiency will become the main theme in the next few years.


The price war is getting more and more intense. In the past year, the implementation of cost reduction methodology and strategy in the field of intelligence has become a "hot topic" that automobile manufacturers have focused on studying.


Everyone is thinking about how to use technological innovation and management model innovation to achieve large-scale cost reduction and efficiency improvement.


Xiaopeng Motors once proposed that in the next one or two years, the cost of car manufacturing will directly affect market competitiveness, and the cost of autonomous driving hardware must be reduced by half by 2024.


Musk has released the same message in all his speeches and press conferences: cost reduction and efficiency improvement will become the main theme of Tesla in the next few years. The ultimate goal is to achieve lower prices for new cars by reducing costs, thereby harvesting a larger global market.


In the past 2023, NIO has also made many efforts in management efficiency and market services, such as layoffs and opening up battery swaps. Li Bin himself also admitted that reducing costs and increasing efficiency is currently a very high priority within NIO.


In this environment, upstream suppliers are also looking for new solutions, not only to consider the cost reduction of the entire system for car company customers, but also to meet the higher-level market requirements in product and technology iteration. Judging from the Beijing Auto Show, this trend is more obvious in the intelligent driving track and car chip track.


Especially in the field of automotive chips, people no longer emphasize computing power in a single dimension when introducing products and deploying strategies . We found that chip players pay more attention to advantages such as cost, especially leading companies, which are all looking for a "cost-effective" route to reduce costs for car companies, so as to occupy more market high ground.


01. Cost reduction methodology


In the view of many local chip manufacturers, saving development cycles for customers is essentially reducing costs for car companies, and this approach is one of the advantages of domestic chips .


Zhang Xitong, product and marketing director of CoreDrive Technology, told Auto Commune that as a local manufacturer, one of its advantages is that the speed of development and adaptation is much faster than that of foreign manufacturers. At this stage, CoreDrive can help domestic companies save 5 to 6 months of development cycle.


"Most of the adaptation work can be done in advance for the OEMs. For example, we can start working on advanced information security methods before the OEMs start to implement them. If the OEMs choose our solutions, they can save this cycle directly. This is something that foreign chip manufacturers may not be able to do."


The other is to support the customized needs of domestic customers.


"Domestic car companies are actually quite different from foreign ones in production. Domestic competition is very fierce, and they need to meet many differentiated demands. Chips need to support customization from the bottom up." In this regard, Zhang Xitong believes that domestic chip companies have a great advantage, and it is also one of the moats of "dark horses" in automotive chips such as Xinchi.


Wang Kai, founder, director and CEO of CoreEngine Technology, also said that automobile manufacturing is currently developing in a new direction, and everyone is thinking about how to reduce costs and increase efficiency for automobile companies and integrate rigid demands. Bringing cabins, berthing and travel together is a good solution.


In essence, it provides a chip solution with higher integration, which is a horizontal functional expansion. It integrates multiple functions into fewer chips, thereby reducing the cost of intelligence for automakers.


In the field of intelligent driving, the Black Sesame Huashan Series A1000 is China's first automotive-grade single SoC chip platform that supports integrated driving and parking domain controllers . It is also the most mature and mass-produced autonomous driving chip for car companies in China. At present, A1000 is in full mass production and delivery.


Luo Qing, former head of Continental's Autonomous Driving and Mobility Business Group in China, once said that the development of the integrated driving and parking system is product-driven, taking into account both functional experience and cost. Integrating the two systems into one can reduce costs by about 50%.


This is extremely attractive to car companies.


Cross-integration is also a challenge for chip companies. Zhang Xitong of CoreChips believes that this difficulty is mainly reflected in three aspects.


First, the integration of more functions requires higher processing power of MCU . Second, security needs to be considered. Cross-domain integration has crossed into another domain, which may be power and chassis. In this case, higher security is definitely required. Third, some isolation mechanisms may be needed between various businesses, and products also need to support virtualization through various software and hardware methods.


At present, the majority of mass-produced solutions include cabin-and-berth integration and driving-and-dock integration. Among all the integrations, cabin-and-drive integration is the most difficult.


What is the difficulty for chip suppliers in integrating cabin and cockpit?


Sun Mingle, CTO of CoreDrive Technology, believes that, first of all, the functional safety levels are different, and the ADAS functional safety level is higher. In terms of the entire design, if a single chip is used, the entire chip must be able to reach the ADAS level to be realized.


Secondly, the heat dissipation challenge is even greater. Originally, the two controllers had their own boxes, and when they were stuffed together, the power consumption was put together, making thermal design more difficult.


Finally, in terms of complexity, hardware decoupling has been done and the interaction is actually very limited.


After putting them together, the interactions are all inside the chip, which makes it more efficient, but running so many things on a single processor at the same time is more difficult than before in terms of software , underlying software, or the entire software architecture.


GigaDevice is a technology company in the China Chip Zone. In recent years, it has been committed to developing memory technology, MCU and sensor solutions.


At the auto show, the company's head told Auto Commune that this was their first time participating in the Beijing Auto Show, and that the China Chip Alliance organized everyone to get together, which was a good opportunity for a collective appearance and to showcase the latest technologies.


"Chinese companies have their own methodology. In fact, reducing costs for car companies is one of our moats. Of course, for this moat to be sustainable and effective, it depends on technological innovation on the one hand and cooperation with car companies on the other."


02. “Computing power is king” is gone forever


More chip companies have realized that as car companies pursue cost reduction and efficiency improvement, customers are more concerned about the comprehensive indicators of chip products, such as performance, price and power consumption. The era of single-dimensional computing power and "computing power is king" is gone forever, and cost performance has become an important consideration for car companies in purchasing.


In essence, just piling up computing power is definitely not enough, and it requires the coordination of sensors and other software and hardware.


Take Tesla as an example.


At present, the use of large models in cars has become a hot topic, but the matching degree between large models and car chips directly affects the utilization rate of computing power. Tesla has specially designed a chip with a computing power of 144Tops for its own algorithm. The reason why it can outperform chips in the range of 500Tops in performance is precisely because of its high degree of synergy.


Moving back in time, Tesla's self-developed HW3.0 chip in 2018 improved intelligent driving performance by up to 20 times, but the computing power of a single chip was only 72tops.


Horizon Robotics CEO Chen Liming also said that the competition in the intelligent driving field is becoming more and more intense, and everyone is under great pressure, from OEMs to suppliers. This has forced Horizon Robotics to focus on developing cost-effective and high-performance products, rather than just pursuing high computing power.


“From a system perspective, we optimize costs.”


In his opinion, local optimality does not mean overall optimality or system optimality, and we must consider how to optimize from a system perspective.


For players like Horizon, the most important task is to reduce system costs while improving the performance and user experience of the entire system. Development costs and system costs are both crucial to Horizon.


Chen Shujie, vice president of Xinchi, also told Auto Commune that about three or four years ago, competitors were all competing for " AI computing power ", but Xinchi did not follow suit at that time.


“Now, everyone is actually doing a withdrawal of computing power.


CoreChi focuses on L2+ computing power, which is still the most practical computing power. "We didn't go out first, wasting a lot of money, and the R&D cost invested is more cost-effective. In fact, everyone is now thinking, do all cars really need to achieve 8295 GPU computing power?"

[1] [2]
Reference address:To reduce costs for car companies, chip companies are trying their best

Previous article:How will driverless cars develop in the future?
Next article:What will the future of cars look like? Take a look at Geely's latest technology

Latest Automotive Electronics Articles
Change More Related Popular Components

EEWorld
subscription
account

EEWorld
service
account

Automotive
development
circle

About Us Customer Service Contact Information Datasheet Sitemap LatestNews


Room 1530, 15th Floor, Building B, No.18 Zhongguancun Street, Haidian District, Beijing, Postal Code: 100190 China Telephone: 008610 8235 0740

Copyright © 2005-2024 EEWORLD.com.cn, Inc. All rights reserved 京ICP证060456号 京ICP备10001474号-1 电信业务审批[2006]字第258号函 京公网安备 11010802033920号