news

at the 2024 tencent global digital ecosystem conference held today, tencent released the core capability upgrades and practical case results of products such as smart car cloud and smart driving map, helping to build a new generation of "car-cloud integrated data closed loop" and provide infrastructure support "under the iceberg" for smart cars.

zhong xiangping, vice president of tencent group and president of tencent smart mobility, said: "on the one hand, tencent will give full play to its underlying technical capabilities of 'cloud-based, car-cloud integration' to help partners in the automotive industry build a closed-loop data capability for car-cloud integration, accelerate the functional iteration of autonomous driving and smart cockpits, and promote efficiency improvement in the industry chain; on the other hand, based on the advantage of 'ecological connection and serving people', we will help partners build a global service network, connect users' travel service scenarios, and continuously improve the smart travel experience."

so far, 90% of car companies have chosen tencent as their automotive cloud service partner, with cooperation covering more than 100 car companies and travel technology companies, and providing overseas cloud services to more than 30 car companies and travel technology companies. tencent's growth rate in the automotive public cloud market ranks first in the industry, which is twice the market average growth rate. tencent's smart cockpit solution has been installed in more than 15 million vehicles.

smart cars have gone through the era of "hardware construction" and "software definition" and are moving towards a new era of "ai drive". with the exponential growth of data scale and computing power requirements, building a closed loop of car-cloud data and more efficient collection, storage, processing and analysis of massive data will become the core "growth flywheel" of smart cars.

tencent has integrated leading technologies and products in computing power, network, data, and security, and launched the "car-cloud integrated data closed loop" solution, which has been put into practice with many car companies. tencent core provides five pillar capabilities, including car-side data solutions, cloud infrastructure, cloud data processing platform, cloud intelligent computing training platform, and full-link security and compliance services.

zhong xuedan, vice president of tencent smart mobility, said that tencent has always focused on building a good infrastructure "under the iceberg" to enable partners in the automotive industry to focus on the evolution of core capabilities, focus more on upper-level intelligent application innovation, focus on the iteration of models and algorithms, and focus on the continuous improvement of user experience.

tencent can provide a one-stop vehicle-side data solution, including data collection, data desensitization, data upload and data downlink channels, which can efficiently exchange massive data between the vehicle and the cloud. the solution has a wide range of system compatibility and can be well adapted to the weak network environment of the vehicle. in terms of encryption and decryption performance, compared with industry competitors, asymmetric encryption performance has increased by 50% and symmetric encryption performance has increased by 1.87 times.

at the cloud infrastructure level, tencent's global cloud infrastructure and 3,200+ edge acceleration nodes can provide high-quality nearby access services for smart cars. in terms of resource scale, tencent has deployed more than 1 million servers across the entire network, with a total computing power scheduling scale of more than 150 million cores, providing 16 eflops of intelligent computing power, and realizing the deployment of distributed hybrid computing power on and off the cloud.

the "tencent vector database" can support up to 100 billion vectors and a peak qps capacity of 5 million. its performance is more than 1.5 times higher than the industry average, and it can store 5 to 10 times more data with the same memory.

in terms of intelligent computing training, tencent can provide a high-performance, diversified computing platform and a complete engineering platform. the efficiency of large model training is accelerated by 30%, the reasoning performance is improved by up to 2.5 times, and the gpu utilization rate of reasoning scenarios is increased by 60%. ti, an out-of-the-box artificial intelligence development platform, can fully support one-click calling of pre-trained large models in various fields and achieve more than 2 times reasoning acceleration.

with the rapid development of autonomous driving technology, the human-car co-driving model will become the mainstream for a long time in the future. however, this model also brings many challenges, such as the frequent switching between human and car driving on complex roads, the fear caused by the inconsistency between the self-driving path and the cockpit navigation path during driving, and the linkage between outdoor driving and indoor parking.

on the day of the conference, tencent officially released the "smart driving map 8.0" cabin-driving integrated solution, which further classified, integrated and processed map data elements at all levels, completed the construction of a "one map" system, and met the needs of both human and vehicle driving. in the field of human driving, it provides drivers with a more realistic, better-used and clearer vehicle-side navigation map. in the field of vehicle driving, it can provide lighter and more open data support for autonomous driving, helping the rapid iteration and low-cost upgrade of autonomous driving systems.

in terms of one-stop service capabilities, tencent is committed to providing seamless navigation services from outdoor roads to indoor parking spaces. users can easily find parking spaces by seamlessly switching between indoor and outdoor maps.