On May 10, Geely's Zeekr was listed on the New York Stock Exchange under the ticker symbol "ZK". It landed on the market with an issue price of $21, but its share price soared, closing at $28.26 on the first day, with a market capitalization of $6.898 billion (about 49.85 billion yuan), almost equal to Xpeng's $7.3 billion. On the second day of listing, it continued to rise sharply, with an increase of more than 10% and a market value of more than 7.5 billion.
This listing also broke a record for new energy vehicles, with only 37 months from the brand launch to a successful IPO. In comparison, Tesla took seven years, while Xpeng, Ideal, and NIO took six, five, and four years, respectively.
The reason why ZEEKR cars can be launched so quickly may come to mind the positioning of high-end luxury, the differentiation strategy, and the vast architecture of SEA.
However, behind the success of ZEEKR, there is also an agile, flexible and efficient digital system that is unknown to the public.
01 From a manufacturing company to a user company
ZEEKR was born out of Geely Holdings, and CEO An Conghui is a veteran of Geely Automobile, who joined Geely in 1996 and has been involved in the growth of the entire company.
When An Conghui bowed down to enter this car-making boom, it meant that as the "old mage" of the car rim, he lost his past, re-entered the "novice village", and carried out a "second venture". He believes that ZEEKR must become a "user enterprise", not only relying on internal teams to develop products, but also co-creating with the majority of users and partners. For this reason, at the age of 51, he took off his suit and changed into casual clothes, and did not let everyone call Mr. An, but called him "Congcong".
In the process of co-creation with users, the role of ZEEKR APP is particularly important. It is the link between ZEEKR and users, and can do direct communication and interaction with users online, it can widely collect user feedback problems, and for the problems raised by users, ZEEKR will confirm and solve them in time.
An Conghui said that with the development of smart cars, it is inevitable that some problems will occur, whether it is on the product side or the user side. When users find problems and make suggestions, dissatisfactions, or even sharp criticisms, it does not mean that they are picky, but that they care about ZEEKR.
With the explosive growth of the number of registered car owners of the ZEEKR APP, it is a great challenge for ZEEKR to face the stability, fluency and rapid response to the needs of the APP. For example, it is important to ensure that core services are continuously online 24/7, especially in response to new model launches and sudden hot events, and the system needs to cope with sudden large traffic to ensure availability and user experience. Rapid business development requires rapid iteration, but in order to ensure stability, release can only be carried out during off-peak business hours. In addition, inconsistent gateway architectures lead to long request links, which also increases costs and affects request response time.
In order to solve these challenges, ZEEKR has cooperated with Alibaba Cloud to upgrade the APP architecture in a cloud-native manner. The ZEEKR automotive APP system adopts Alibaba Cloud's microservice engine MSE to improve the overall stability and development iteration efficiency of the microservice system, while reducing the resource cost of the ingress gateway and the network latency of requests.
One of the features of this solution is that Alibaba Cloud's MSE Sentinel real-time monitoring and link function can quickly locate slow calls and unstable services, and carry out flow control and concurrency control in a timely manner to ensure that the system can quickly return to normal levels from the brink of collapse. At the same time, when multiple microservices on the customer's core business link need to be released at the same time, Alibaba Cloud implements full-link grayscale for multiple services based on MSE cloud-native gateway and traffic grayscale marking.
At the same time, the cloud-native gateway combines the traffic gateway and the microservice gateway into one, and has built-in service governance capabilities, which not only reduces resource costs by 50%, but also shortens request time and reduces O&M complexity. Compared with the traditional VM-based deployment method, the ACK containerized deployment method is adopted, which helps ZEEKR improve resource utilization by 20% and O&M efficiency by 50%.
However, the sword must be sharpened on stone, and a man must be trained in things. The automotive industry's press conference is an important moment for car companies, especially when prices are announced. On February 27 this year, ZEEKR released the new ZEEKR 001 model. When the price was announced, the number of business requests of ZEEKR APP surged by 5-6 times, and the system was under tremendous pressure.
02 PaaS raises the security ceiling of the Internet of Vehicles
ZEEKR's success stems not only from their pursuit of the ultimate user experience, but also from its excellent security. In the field of new energy vehicles, safety is the biggest "luxury". In case of living safety accidents, ZEEKR's vehicles can always protect the safety of drivers and passengers, so they are nicknamed "Highway Tenterkr".
This safety is not only reflected in the vehicle's construction, three-electric system and manufacturing process, but also in the excellent ability of the Internet of Vehicles.
In order to reach the needs of users in the first time and respond in a timely manner, so that smart cars are closer to customers, ZEEKR is promoting a new generation of distributed Internet of Vehicles, so as to achieve the ultimate user experience, but to achieve this goal, it is necessary to solve the challenges of collecting, storing and processing a large amount of data.
First of all, in terms of data upload bandwidth, as the number of vehicles increases and the amount of data uploaded by a single vehicle increases, the concurrent bandwidth of the entire public network may reach the level of 100 Gbit/s. Secondly, in terms of storage capacity, the scale of data storage for one year will also reach the petabyte level. Finally, in terms of real-time data processing, the data upload, analysis, and analysis of the whole link need to be completed within seconds, which puts forward extremely high requirements for data processing efficiency.
For example, new forces such as Wei Xiaoli have been using cloud computing platforms to solve the problem of response speed since about 2016. They use the cloud's flexible resources and agile business iterations to quickly meet demand without having to set up their own data centers.
Image source: Visual China
ZEEKR has taken a similar approach, but it doesn't just focus on the scalability of cloud computing services. In ZEEKR's view, scalability is not enough to cope with the growing demand for data and services. It also requires the integration of a range of capabilities, including data storage, computing, analytics, and compliance.
To this end, ZEEKR chose to rely on Alibaba Cloud's computing resource basic services, and deeply utilized Alibaba Cloud's Platform-as-a-Service (PaaS) services. This enables ZEEKR to make better use of various services in the cloud to achieve efficiency and cost reduction at the governance level, including cost optimization, stability and compliance. Specifically:
- Massive storage and data processing capabilities: Intelligent vehicles need to process and store data in real time while driving. In the former, ZEEKR detects abnormalities in real time and warns users by monitoring the operation status of the vehicle to avoid safety accidents; The latter, ZEEKR, uses the hot and cold layering technology of Alibaba Cloud OSS to distinguish the Internet of Vehicles data according to the storage time, which greatly reduces the storage cost.
- Auto scaling: The amount of data in the Internet of Vehicles fluctuates greatly with time and user behavior, so ZEEKR chooses Alibaba Cloud serverless Kafka, Function Compute FC and other products to automatically adjust resources according to demand. In this way, it can automatically expand the capacity during peak data hours to ensure the normal operation of the system, and release resources and control costs during low peak hours.
From the first day of its birth, ZEEKR chose to rely on Alibaba Cloud to achieve "agile, efficient, and secure" iterations. By the end of this year, the cumulative sales of ZEEKR cars are expected to reach 400,000 units, and such a high-concurrency, stable and reliable Internet of Vehicles experience support is also based on the capabilities provided by Alibaba Cloud.
03 Cloud services help large models get on the car
At the same time, ZEEKR is committed to redefining the essence of automobiles under the trend of intelligence and digitalization. Traditionally, automobiles have been regarded as hardware-led industrial products, but now with the rapid development of data, algorithms, and software, automobiles are transforming from a means of travel to an intelligent mobile space with both hardware and software, and have a new soul.
Image source: ZEEKR Automotive
At present, the focus of the automotive industry is gradually shifting to large-scale model technology. In the past, smart cars mainly relied on rule writing to implement functions, but this approach has limitations. With the development of large AI models, end-to-end solutions have made great progress in accuracy and generalization capabilities, providing new opportunities and advantages for the development of intelligent seats and intelligent driving.
Further, with the in-depth development of artificial intelligence technology, multi-modal large models have become a new hot spot, and most in-vehicle scenarios in the industry can be realized through multi-modal large models. The multi-modal large model can process information of multiple modalities at the same time, and realize the cross-modal interaction and fusion of information. This allows machines to better understand human intentions and needs and provide more precise and personalized services.
The application of these new technologies not only makes cars smarter, but also brings users a more convenient and interesting life experience. The intelligent assistant equipped with Alibaba Cloud's Tongyi model can display rich images through the vehicle's central control screen according to the passenger's natural language commands, such as "a kitten climbs on the window sill", thereby increasing parent-child interaction and riding fun in the car. At the same time, through the distillation model, the large model can generate the required image in just a few seconds while ensuring the image quality, greatly shortening the image generation time and making the user experience smoother.
Although ZEEKR is a start-up, it has shown a strong ambition to respond to these trends. Their latest products, ZEEKR 007 and the all-new ZEEKR 001, are equipped with Kr AI large models, bringing users a personalized and fun experience. Through the large model, the ZEEKR intelligent assistant EVA can not only complete tasks such as finding news, arranging itineraries, translating, and literary pictures, but also become a productivity tool. After the press conference, the first press release generated by an electric vehicle was witnessed, marking the entry of the intelligent cockpit application into the era of large models.
Not only that, Zeekrypton also has greater "ambitions". Jiang Jun, vice president of ZEEKR Intelligent Technology, said that they believe that the development of AI large models is not limited to the role of voice assistants. Their goal is to make in-car interactions easier and more convenient by enabling AI to better understand the driver's intentions.
It is understood that the large model independently developed by ZEEKR is built on the Alibaba Cloud platform, relying on Alibaba Cloud AI infrastructure to provide strong training capabilities and inference capabilities, which not only has large training throughput, high efficiency, fast inference speed and short latency, but also supports various intelligent applications of ZEEKR automobiles. At the same time, it makes full use of Alibaba Cloud's ecological resources such as Alipay and AutoNavi Map to realize the upgrade of the intelligent cockpit, providing users with more convenient clothing, food, housing and transportation services and itinerary planning.
04 Symbiosis with the cloud has become an inevitable choice
In just three years, the sales volume of ZEEKR has increased from a few thousand units to more than 240,000 units, setting a record for the shortest time for new energy vehicles to be listed in the United States. Behind such rapid development, how to match the technical architecture of ZEEKR with its development and help it break through in the fierce market competition has become the key.
Image source: ZEEKR Automotive
In particular, with the acceleration of intelligence and globalization, advanced cloud-native technology architectures such as global multi-location multi-active, cross-AZ high availability, and off-line hybrid have become essential capabilities for car companies. At the same time, with the new paradigm brought by large models, the market penetration of large models will accelerate, and more killer applications will be born in the entire market to replace and penetrate the original enterprise processes, and these new technologies will be born first or even only on the cloud.
As mentioned above, in terms of the two key technology systems of cloud and AI, ZEEKR has cooperated with Alibaba Cloud since its inception, and uses Alibaba Cloud for the full stack of marketing, Internet of Vehicles, and intelligent cockpit. If you use three sentences to summarize the "long" "long" of ZEEKR in Alibaba Cloud, the new paradigm of cloud use is represented: one is the efficient and ultimate use of cloud, the second is to embrace the cloud-native advanced technology architecture, and the third is to reshape the business process natively with AI. From this, we can see that full-stack cloud migration has become an inevitable choice for smart electric vehicle companies, making them stand out in globalization and AI.
ZEEKR's IPO listing is just the beginning. Li Qiang, Vice President of Alibaba Cloud Intelligence Group and General Manager of Automotive Energy Industry, said, "The transformation of the global automotive industry has just begun, and the two major variables that will affect the global automotive industry in the future are global layout and intelligent innovation. The former determines the breadth of business coverage, while the latter relates to the depth of the product moat." Under these two trends, digital capabilities and intelligent capabilities are the real differentiated competitiveness of enterprises, and in this regard, ZEEKR and Alibaba Cloud choose to embrace the opportunity.