Introduction:
Black Sesame Technologies is systematically expanding its identity from an intelligent driving chip supplier into a comprehensive computing power platform company covering embodied intelligence and edge AI inference.
This "first AI chip stock for intelligent vehicles," listed on the Hong Kong Stock Exchange in 2024, recently laid out its clear "three-step" strategic roadmap: progressively leaping from the tens-of-billions-of-dollars intelligent driving market to the hundreds-of-billions-of-dollars embodied intelligence market, and ultimately to the trillions-of-dollars "physical AI" edge inference market.
Strategic Leap: Three Moves from Intelligent Driving to Physical AI
Black Sesame Technologies was founded in 2016. Founder and CEO Shan Jizhang graduated from Tsinghua University and spent over thirty years working in Silicon Valley. The core team comes from well-known technology companies such as Microsoft, Arm, and Qualcomm. The company is headquartered in Wuhan, with R&D bases in Shanghai, Shenzhen, Beijing, and Chengdu, employing over a thousand people, and maintains overseas operations in Singapore and Silicon Valley.
The company's development history can be broadly divided into five stages: from 2016 to 2018 was the technology validation period, focusing on refining two core IPs—ISP and NPU; starting in 2020, it launched the A1000 chip and achieved commercial deployment, with customers covering mainstream automakers such as Geely, Dongfeng, BYD, GAC, and FAW; it then iterated to the A2000, boosting computing power from tens of TOPS to hundreds of TOPS; and this year it entered a phase of overall horizontal expansion.
Black Sesame Technologies believes that the shift in inference paradigms over the past two years has expanded single-computation workloads by more than ten thousand times and usage volume by more than a hundred times. According to China Insights Consultancy, the edge AI market will continue to expand through 2029, with the compound growth rate of AI device connections reaching 44%, and the share of edge AI inference chips in scenarios such as robotics, intelligent vehicles, and visual monitoring will rise significantly.
Based on this judgment, the company has extended its strategic chessboard from the tens-of-billions-of-dollars automotive intelligent driving market to the hundreds-of-billions embodied intelligence market, and ultimately aims at the trillions-of-dollars physical AI full-scenario edge inference market.
Three Major Scenarios: Vehicle, Robotics, and Edge AI Inference
Intelligent driving has been the company's main battlefield since 2019, covering passenger vehicles, commercial vehicles, low-speed terminal logistics vehicles, and open Robotaxi scenarios, with an established customer system.
Embodied intelligence is the second major scenario the company is currently focusing on expanding. In November 2025, the company launched the SesameX multi-dimensional embodied intelligence computing platform, migrating the vehicle-side perception-decision-control-safety closed loop to the robotics track, spawning three major platforms: the most mature and cost-effective Kalos universal robotics hub for commercial service robots; the Aura multi-task execution general computing platform for complex operation scenarios such as multi-legged robots, industrial inspection robots, and collaborative robotic arms (which already has deployment cases including COSCO Shipping Holdings' embodied intelligence inspection); and the Liora embodied intelligence all-purpose computing platform for advanced scenarios such as humanoid robots and multi-robot collaborative systems. Black Sesame Technologies stated that the underlying middleware and base software have reusable modules, making the logic of migrating from vehicle to robotics smooth and capable of improving R&D efficiency.
In terms of commercial rhythm, Black Sesame Technologies has chosen to first spend two to three years building scale and capturing customer demand, then enter mid-to-high-end complex scenarios with a stable supply chain and a domestic AI chip positioning when the industry scale explodes.
Edge AI inference is the third and most aggressively expanding scenario, laid out in three directions: first, vehicle-mounted edge computing boxes and vehicle-road collaboration, relying on intelligent vision algorithms, having previously provided algorithm deployment for Li Auto's AI glasses; second, Eeasy Tech, acquired in June 2026—this Zhuhai-based company founded in 2016 focuses on AI machine vision algorithms plus SoC system solutions, covering three major business areas: intelligent vehicle, intelligent hardware, and intelligent security, with intelligent hardware including smart desk lamps, smart gimbals, action cameras, AI glasses, and other categories. After the acquisition, it formed a complete chip coverage with Black Sesame Technologies ranging from "a fraction of a TOPS to 1000 TOPS"; third, AI inference servers, with the company recently announcing a nearly 300-million-yuan AI computing server sales framework contract, of which approximately 250 million yuan is expected for complete machine procurement within one year, marking the company's first batch of orders in the AI computing server field. The company stated that it will continue to conduct AI computing services business with more leading companies in the future, jointly grasping the new round of industry opportunities brought by the expansion of AI computing from the cloud to the edge and local deployment.
Competitive Landscape: A Replay of the Smartphone Era
Black Sesame Technologies believes that the wave of automakers developing their own chips will eventually converge in the medium to long term. Its logic is analogous to the smartphone era: in the early days, manufacturers widely developed their own chips, but the landscape eventually converged into an oligopoly dominated by a few leading third-party suppliers such as Qualcomm and MediaTek, while phone manufacturers focused on technological improvements like algorithms to deliver differentiated customer experiences. Vehicle-side intelligent driving chips are expected to undergo a similar evolution, and Black Sesame Technologies' goal is to occupy a favorable position in this convergence process, adhering to the core logic of specialized division of labor and an open ecosystem.
The company's differentiated support lies in the continuous accumulation of self-developed IP and cross-scenario reuse capability—the same core perception and inference architecture can be migrated at low cost across the three markets of automobiles, robotics, and general edge AI, which constitutes the underlying logic of its horizontal expansion "three-step" strategy.
Dialogue with Black Sesame Technologies
[Q] How do we view Black Sesame Technologies' strategic direction in vehicles? From our perspective, what changes are happening in the industry?
[A] We believe that intelligent vehicles are the first scenario for the large-scale deployment of embodied intelligence and one of the most important application markets for edge AI. Automotive intelligence is still continuously evolving: from the demand side, intelligent assisted driving is accelerating in penetration, placing higher demands on computing performance, cost, and energy efficiency; from the technology side, intelligent driving and cockpit AI models continue to evolve, cockpit-driving integration is accelerating, and vehicle electronic and electrical architectures are advancing toward central computing; from the supply chain side, the importance of independent and controllable supply chains and supply security continues to rise. Facing current and future industry demands, providing high-performance, high-reliability, and high-efficiency intelligent computing solutions, and continuously improving software-hardware collaboration and system-level innovation capabilities, remains an important mission for domestic automotive chip companies and our ongoing direction of effort.
[Q] What is the company's approach to embodied intelligence? How do you see the future growth of embodied intelligence?
[A] Embodied intelligence is the company's second growth curve and one of its important long-term development strategies. We have already reached strategic cooperation with multiple leading robotics companies in humanoid, industrial inspection, and port scenarios and launched joint development, continuously enriching our scenario deployment reserves. In the next 2-3 years, the company will adopt a differentiated competitive strategy to capture core deployment scenarios. In terms of delivery format, we will mainly promote an integrated package of modules plus underlying software plus toolchain to address customers' underlying development pain points. We expect embodied intelligence to continue growing at a high speed, and its share of the company's revenue will continue to increase in the future.
[Q] What market demand is the recently announced 300-million-yuan order mainly targeting?
[A] We see that with Agentic AI bringing rapid growth in model invocation frequency and Token demand, the industry is forming a layered inference system from MAX models to Flash and mid-sized models. These cost-effective models are undertaking a large volume of high-frequency, large-scale AI inference demand by virtue of a better balance among capability, cost, latency, and energy efficiency. The diverse market structure of AI computing gives us more opportunities to expand. We will focus on these cost-effective AI inference solutions and look forward to collaborating with more leading manufacturers.
Comments