At present, the automotive industry is undergoing a profound transformation from electrification to intelligence, and the evolution of semiconductor technology has become the core engine driving this change. With the advent of the “post-Moore era”, the law of scale, represented by large models and NPUs (neural network processors), has reshaped the computing paradigm, shifting from hardware-driven computing to algorithm-driven computing. In this context, intelligent assisted driving, as a key application scenario, is accelerating the realization of intelligent automobile scenarios.
At the same time, local Chinese brands have shown strong momentum in technological innovation and supply chain integration, but face challenges such as cost pressure and solution compatibility. As a leading enterprise focusing on automotive computing SoC chips, Black Sesame Intelligence is committed to providing a safe and efficient chip base for the industry through platform-based solutions to help smart travel enter a new era. According to the data of the high-tech list, Black Sesame Intelligence ranks third among the suppliers of NOA integrated domain control computing solutions for traditional independent brand passenger cars in the Chinese market in 2024, with a market share of 12.15%.
On July 22, 2025, Zhou Yong, Director of Product Management of Black Sesame Intelligence, said at the 8th Intelligent Assisted Driving Conference in 2025: In the wave of automotive intelligence, enterprises need to be “people-oriented, start with the end”, take safety as the first criterion, and emphasize the strategy of “slow thinking and fast action”, avoid blindly pursuing computing power competition, but should focus on algorithm efficiency and platform innovation, and empower full-scenario travel with cost-effective solutions.
Zhou Yong | Director of Black Sesame Intelligent Product Management
The following is a summary of the speech:
The development of semiconductors promotes a new era of automotive intelligence
The century-old evolution of semiconductor technology has transitioned from the era of Moore’s Law to the era of the law of scale, driving the automotive industry from the first half of electrification to the second half of intelligence. The Moore’s Law era focused on the number of transistors and the improvement of chip processes, which gave birth to the emergence of processors such as CPUs and GPUs; The chip process is iterating from micron to nanometer, from 40nm to 2nm, and the computing speed has achieved rapid growth. In the era of the law of scale, computing has shifted from hardware-driven to algorithm-driven, algorithm architecture has evolved from CNN/LSTM to Transformer, and the emergence of NPU has greatly exceeded computing power, promoting large models to enter device-side reasoning, and empowering thousands of industries to open a new era of intelligent computing.
Source: Speaker material
The current situation and multi-dimensional challenges of automotive intelligence
Automotive intelligence is accelerating its penetration into multiple domains such as cockpit, assisted driving and chassis at the business and application levels, and the application of large models has accelerated this trend. Cockpit domain applications are gradually transitioning from traditional DMS and voice assistants to multimodal perception based on LLM/VLM to improve the interactive experience. The assisted driving domain benefits from BEV+Transformer and end-to-end model evolution to achieve intelligence from perception to decision-making. The chassis domain relies on the modelforecastControl optimizes body attitude management. As an automotive computing SoC chip company, Black Sesame Intelligence focuses on assisted driving and intelligent cockpits.
Source: Speaker material
At present, the industrialization of automobiles is accelerating, and the life cycle of automobiles in the era of fuel vehicles is very long, generally a major change every 8 years and a small change in 4 years. Right nowNew energy vehiclesThe life cycle of each model has been significantly shortened compared with the past, and the new car research and development cycle has been compressed from 3~4 years in the era of fuel vehicles to 18 months now. In the face of the acceleration of automobile industrialization and intelligence, OEMs and Tier-1 customers are facing many challenges, which can be summarized as 4 points: extreme cost, solution compatibility, rapid algorithm iteration and application sinking, these challenges stem from: the difference between the automotive chip solution and the consumer electronics specification level, the long development and verification cycle of the automotive platform, and the phenomenon that each module of the automotive electronic and electrical architecture under the traditional development process needs to be redeveloped and rebuilt every time is very obvious, and the algorithm is iterated and updated too quickly. For OEMs, how to plan high, medium and low-end models, balance performance and cost, and break involution has become a key proposition in the transformation from “scale expansion” to “value deep cultivation”.
Platform-based solutions: Huashan and Wudang series of innovations
As a Tier 2 in the automotive industry, Black Sesame Intelligence focuses on the research and development of automotive SoC chips, launches the Huashan series and Wudang series, and builds platform-based solutions to address industry pain points. The company’s vision is to become a leader in intelligent automotive computing chips, with a strong team background, with more than 1,000 employees accounting for 85% of R&D personnel, including more than 40 doctors, and the core team from top semiconductor, ADAS and AI application companies. Huashan series of positioning assisted driving computing chips, including A1000 and A2000, of which A1000 is the first local automotive-grade high computing power chip; The Wudang series focuses on cross-domain fusion computing chips, including C1236 and C1296, which are the world’s first automotive-grade cross-domain fusion chips.
Source: Speaker material
In response to the hierarchical needs of high-end models and economic models, the company proposes a safe and intelligent base architecture, which realizes the essential computing functions of integrated vehicles through the Wudang C1200 family, similar to Toyota’s TNGA architecture, to provide a long life cycle platform. In this solution, Wudang C1200 can be connected to cameras, millimeter-wave radar, ZCU and ECU, support cockpit safety screen display and control, economy models achieve cabin driving and parking integration with a single chip, and high-end models are equipped with Huashan A2000 to achieve high-end assisted driving. The advantage of platformization is to balance cost and performance, avoid remanufacturing wheels, and accelerate mass production through tool chain optimization.
The key elements of high-level assisted driving are broken through with the A2000
Four key elements are required to achieve high-level assisted driving: high computing power + high bandwidth, friendly and universal tool chain, platform-based serialization and full-stack solutions. High computing power supports high-order iterations and large model bandwidth requirements; The tool chain needs to quickly adapt to the algorithm supplier and support the expansion of new operators and models. Platformization requires chip combinations to cover high, medium and low configurations; full-stack requires simultaneous adaptation of chip + algorithm, providing reference models and data closed-loop tools to shorten the mass production cycle. Black Sesame Intelligence’s Huashan A2000 family will be released at CES in 2025, representing a new generation of breakthroughs: integrated high-security NPU supports redundancy verification, a three-layer design of the new storage and computing architecture balances performance and cost, natively supports Transformer, and the product portfolio covers pilot assisted driving to Robotaxi;
Source: Speaker material
In addition, the ultra-high integration integrates CPU, DSP, GPU, NPU, MCU and ISP/CV six, the new ISP supports 4 frames of exposure and 150dB HDR, and the self-developed RISC-V scheduling core realizes a single-chip data closed loop, and flexibly expands to support multi-chip computing power. The core of the NPU adopts Jiushao architecture to optimize computing efficiency; The toolchain BaRT (Black Sesame acceleration Run Time) improves ease of use. The A2000 not only empowers smart cars, but also expands to embodied intelligence and device-side AI, inheriting the high reliability of automotive specifications by matching the large and small brain applications of Wudang C1200 to meet the different computing power needs of robots.
Technology implementation and industrial collaboration practice
Black Sesame Intelligence accelerates the implementation of technology through open cooperation, and Wudang C1200 has supported urban NOA scene applications, realizing mapless urban auxiliary piloting and high-speed scene processing. The chip optimizes algorithm efficiency, such as handling complex scenarios such as motor vehicles and pedestrians turning left without protection on urban roads.
Enterprise representatives emphasized that automotive intelligence requires the coordination of the entire industry chain, and chip companies need to work closely with algorithm partners to shorten the mass production cycle with a safe and efficient platform. Black Sesame Intelligence adheres to slow thinking and fast action, avoids involution of invalid computing power, and focuses on algorithm optimization and cost control.
Source: Speaker material
Future prospects and challenges
Looking forward to the future, automotive intelligence will continue to deepen, but challenges coexist. On the one hand, the device-side inference of large models will empower full-scenario travel, which is expected to be in 2030New energythe penetration rate of passenger cars exceeds 80%, and the leader of smart cars will dominate the global market; On the other hand, cost pressure, rapid algorithm iteration, and security compliance require industrial chain coordination. As a long-termist, Black Sesame Intelligence will continue to innovate the Huashan and Wudang series, expand embodied intelligence and device-side AI, and respond to challenges through platform-based solutions. Enterprises need to balance innovation and risk, lead with the “core” force, and achieve a new era of smart travel. Black Sesame Intelligence will practice internal skills to empower OEMs and consumers with full-stack AI chips and promote the global export of Chinese technology.
(The above content comes from the keynote speech “”Core” Power Leads the New Era of Smart Travel in All Scenarios” delivered by Zhou Yong, Director of Product Management of Black Sesame Intelligent Intelligence, at the 8th Intelligent Assisted Driving Conference in 2025 on July 22, 2025.) ）
