Intelligent connected car chips interpret "three companies are promoted"

The smart connected car chip arena has been in turmoil again recently.

NVIDIA, which has "rolled" the computing power of autonomous driving SoC to 2000TOPS, has reached a cooperation with MediaTek to provide artificial intelligence cockpit solutions for software-defined cars; Qualcomm, which uses "connection" to empower smart cockpit platforms, and China Science and Technology Thunder have established an intelligent connected car alliance The innovation center provides underlying software and hardware support to domestic partners; Intel, which has strengthened its autonomous driving capabilities through the acquisition of Mobileye, showcased the Chongqing Internet of Vehicles project in cooperation with Lenovo, demonstrating Intel's capabilities in hardware products and software development kits. Historically, the "three families of Han, Zhao, and Wei divided into Jin" was regarded as the watershed between the end of the Spring and Autumn Period and the beginning of the Warring States Period.

Nowadays, under the new trends in the automotive industry, NVIDIA, Qualcomm and Intel seem to be performing a "three-pronged" drama in the field of intelligent connected cars, accelerating the transition of cars from manufacturing attributes to "consumption + technology" attributes.

The "sharp sword" to seize the high ground

Every chip manufacturer that dominates the field of intelligent connected cars carries a "sharp sword" to seize the high ground.

Source: Compilation of public information

NVIDIA, which has benefited from AI computing power, reached the top of the trillion-dollar market capitalization a few days ago. Relying on its computing power advantage, NVIDIA has achieved the industry's "top match" in the computing power competition of smart connected car chips: the smart car chip released by NVIDIA - NVIDIA DRIVE Thor (Thor) has a computing power of 2000TOPS per chip, with a performance of about It is 8 times that of the current mainstream NVIDIA Orin chip.

But NVIDIA, the king of computing power, has also been frustrated in the smartphone era. In 2008, Nvidia, which has been deeply involved in the field of GPU, launched Tegra, a mobile processor based on Arm and Geforce architecture, which quickly occupied the game console and tablet markets that require high image processing. However, in the field of smartphones, due to failure to seize the opportunity of integrating processors and basebands in the 3G and 4G eras, NVIDIA's third- and fourth-generation Tegra processors have been somewhat lonely in the field of mobile phones due to factors such as higher power consumption.

The failure of the smartphone prompted Nvidia founder and CEO Jensen Huang to make a desperate move. At the 2015 NVIDIA new product launch conference, Jen-Hsun Huang first released the Tegra X1 processor, and then dropped a blockbuster - DRIVE PX, a new smart car processing platform based on Tegra X1.

It is reported that the performance and development experience of the Tegra series on mobile platforms can support navigation, instrumentation, in-car entertainment and other related applications. Tesla has already used the Tegra 3 processor on the Model S to support the central large-size touch screen . After that, Nvidia successively cooperated with Audi and other car companies. At that time, the field of autonomous driving was barely emerging, but industry standards were still blank. Nvidia, which took the lead in moving towards smart cars, relied on the innate advantages of GPUs in large-scale parallel computing to take the lead in the fields of autonomous driving technology and machine learning, becoming the first chip manufacturer to provide super car processors.

Unlike Nvidia, which failed in mobile phone chips, Qualcomm has extended wireless connection capabilities from mobile phones to cars, using it as a "sword" to capture the high ground of smart cockpit chips. 20 years ago, Qualcomm built General Motors' OnStar CDMA  1x in-car networking solution in the United States. From that point on, Qualcomm realized that every car needed connectivity capabilities. Between 2013 and 2014, Qualcomm began to provide in-vehicle infotainment products and received an olive branch from Audi to bring the Snapdragon platform to Audi's cars.

Overall, Qualcomm inherits profound mobile communication technologies such as radio frequency, millimeter wave , and MIMO (Multiple Input Multiple Output), and has strong comprehensive capabilities in CPU, GPU, NPU, etc. Qualcomm also uses its huge smartphone shipments to spread its R&D and production costs. For example, the Snapdragon 8155, which is sought after by car companies such as Mercedes-Benz, Audi, Honda, Geely, Great Wall, and BYD, is based on the Snapdragon 855 chip installed in flagship mobile phones such as Xiaomi 9 Pro and Samsung Galaxy Note10.

"The similarity between mobile phone chips and smart cockpit chips lies in the highly consistent demand for communications, entertainment, and high-definition display." Chi Xiannian, chief researcher and senior engineer at CCID Consulting Integrated Circuit Center, told a reporter from China Electronics News that Qualcomm inherited Competitive advantage of smart phone chips, dimensionality reduction cuts into the smart cockpit chip market, forming a greater competitive advantage.

For Intel, which hopes to make a difference in the field of assisted and autonomous driving, joining hands with Mobileye, which accounts for more than 70% of the assisted driving system-level chip market share, is an excellent entry point to enter the game. In 2017, Intel acquired Mobileye at a premium and co-designed algorithms and chips, filling in a missing piece of its own strategic puzzle: algorithms and corresponding dedicated algorithm processor IP design experience. Through "1+1", Intel has improved the performance, reduced power consumption and cost of autonomous driving products, shortened customer introduction time, and gained commercial competitive advantages.

Frequent moves to build a technological moat

If competitors are strong, they must build a high technological moat. In the slightly crowded arena of smart connected car chips, Nvidia, Qualcomm and Intel have recently made frequent moves to further consolidate their advantages. Not content with improving the computing power and performance of a single chip, the key to NVIDIA's move is to further expand its computing power advantages to cover more areas such as connected vehicle infotainment, cockpit convenience functions and cockpit safety functions.

Recently, NVIDIA chose to cooperate with MediaTek to provide a complete AI smart cockpit solution for software-defined cars. It is reported that Nvidia currently not only provides GPUs for laptops, desktop computers, workstations and servers, but also provides SoCs for automotive and robotic applications. With its new GPU chiplet, NVIDIA expands its leadership in GPU and accelerated computing to a larger market.

NVIDIA DRIVE Concierge Platform NVIDIA told a reporter from China Electronics News that this cooperation has brought more in-vehicle infotainment platform options to automakers that support the NVIDIA DRIVE platform. NVIDIA GPU chips are integrated into the automotive-grade SoC design architecture, and the chips are connected through ultra-fast chip interconnect technology.

NVIDIA's DRIVE OS, DRIVE IX, CUDA and TensorRT software technologies will run on the new automotive-grade SoC to achieve connected in-car infotainment, cockpit convenience functions and cockpit safety functions. "AI and accelerated computing are driving the transformation of the entire automotive industry." Huang Renxun said. The combination of MediaTek's leading SoC with NVIDIA's GPU and AI software technology will bring higher security and innovative interconnected services to the automotive market segments from entry-level to luxury-level. Qualcomm further uses mobile connections to meet the high computing power demands brought about by intelligence. At the 2023 Qualcomm Automotive Technology and Cooperation Summit held at the end of May, Qualcomm demonstrated many of its blockbuster products related to cockpit chips.

It is reported that the Snapdragon 8295 uses a 5nm process that is synchronized with the flagship mobile phone chip platform Snapdragon 888 and 888+, and its AI computing power reaches 30TOPS. Compared with the previous generation product Snapdragon 8155 using the 7nm process, the overall GPU performance of Snapdragon 8295 has been improved by 2 times, and the 3D rendering performance has been improved by 3 times. It also adds integrated electronic rearview mirror, machine learning vision (rear, surround view), Passenger monitoring and information security functions.

Qualcomm Snapdragon Digital Chassis

The reporter learned from Qualcomm that Qualcomm currently ranks first in the fields of in-vehicle networking and automotive wireless connections. In China, more than 50 cars equipped with the Snapdragon Automotive 5G platform have been launched in just a few years. "Perception" is the key word for Intel's efforts in intelligent connected cars. Recently, Intel and Lenovo jointly demonstrated two Internet of Vehicles projects in Chongqing for the first time.

Self-driving buses equipped with Lenovo and Intel's roadside computing technology "Vehicle-road collaboration and intelligent transportation put forward higher requirements for traffic roadside perception and calculation." Intel technical experts told reporters from China Electronics News that vehicle-road collaboration and Intelligent transportation has very high requirements for real-time, accuracy and high reliability of information collection and processing. Since the load of information processing is mainly carried out on the chip, the accuracy, real-time and high reliability requirements for chip processing are obviously much higher than those on highways.

In this regard, Intel provides solutions with perception fusion technology as its core. Mobileye also has profound knowledge in the field of lidar , known as the "eye of the vehicle" . Therefore, close cooperation with Mobileye is also one of the keys to Intel's development in the field of autonomous driving. The relevant person in charge of Mobileye told a reporter from China Electronics News that Intel silicon photonics technology can help Mobileye develop lidar systems to meet the performance and cost requirements of consumer self-driving car deployment. Here, Intel has a strong advantage in integrating active and passive laser components on the chip.