Smart driving chips ramp up production, robot business contributes nearly 100 million yuan, Black Sesame Intelligence's "second curve" and "full-stack ambition"

If you trade stocks, just rely on Golden Kirin analysts’ research reports—authoritative, professional, timely, and comprehensive—helping you uncover promising themes and opportunities!

（Source: 财闻）

The “arms race” in edge AI chips is expanding from a single automotive-grade scenario into intelligence for the broader physical world. While most players are still competing and optimizing within the intelligent driving lane, Black Sesame Intelligence (02533.HK) has quietly completed its strategic leap from “specialized, refined, and innovative” to a “platform-based player.”

On March 31, the 2025 full-year performance report released by Black Sesame Intelligence showed that the company not only delivered a track record of three consecutive years of high revenue growth, but—behind the financial figures—also outlined a clear path of transition from a leading intelligent driving chip provider to an end-to-end edge AI full-stack leader.

A favorable industry tailwind and performance corroboration: revenue up 73.4%, losses narrowed significantly

Currently, edge AI is becoming a core engine of new quality productive forces in the “15th Five-Year Plan period plus the upcoming 5 years” (15th Five-Year Plan’s successor) stage. According to China Industry Research (CINP) and iResearch (In China Research Institute) data, the size of China’s edge AI market surpassed RMB 250 billion in 2025, up 35% year over year. Among them, intelligent driving and embodied intelligence form the dual growth poles. Guoqing Consulting predicts that in 2025 the global intelligent driving chip market size will reach USD 7.6 billion, while the Development Research Center of the State Council has also judged that the embodied intelligence market will exceed RMB 1 trillion by 2035.

By capturing these two tracks, Black Sesame Intelligence’s performance has entered explosive growth. The annual report shows that in 2025, the company achieved operating revenue of RMB 822 million, up 73.4% year over year, maintaining rapid growth for three consecutive years. More importantly, the quality of its growth has improved markedly: gross profit reached RMB 337 million, up 73.1% year over year, and the gross margin stayed steady at a high level of 41.0%. These figures indicate that the company is not relying on price cuts to win market share; while scaling up volume, it has maintained its high product pricing power, and its business model is becoming increasingly healthy.

Of particular note is the transformation on the profit side. Although losses during the year appeared to widen due to non-operating factors such as changes in the fair value of preferred shares, the adjusted net loss—the metric that best reflects actual operating conditions—narrowed significantly by 17.5% year over year. This “scissor-gap” effect suggests that as revenue scale rapidly balloons, the dilution effect of R&D and sales expenses has begun to emerge, and the company is getting closer to the inflection point of breakeven.

Intelligent driving as the core baseline: evolving from “domestic substitution” to a “global breakthrough”

In the core business of intelligent driving, Black Sesame Intelligence’s moat is evolving from “breakthrough in a single point” into “company-wide coordinated operations.”

Faced with pressure from giants such as Qualcomm and NVIDIA, the intelligent driving chip industry in 2025 showed a pattern of “high-end intensifying while mid- and low-end penetrates quickly.” Black Sesame Intelligence adopted a strategy of “pairing high and low, covering all scenarios.” The Huashan A1000 series, a sales mainstay, has a lifecycle of more than 5 years. It has not only been successfully deployed in multiple models from leading domestic automakers such as Geely, BYD, and First Auto Works Group, but has also accompanied China’s vehicle exports to overseas markets, becoming a “cash cow” business supporting the company’s cash flow. Meanwhile, through Desay Xiwai’s unmanned mini-car solutions, Black Sesame Intelligence extends the reach of the A1000 to L4 commercial vehicle scenarios such as unmanned cleaning and unmanned postal logistics vehicles, breaking through the ceiling of passenger vehicles.

As for the Huashan A2000 series, which will determine future voice in the high-end market, Black Sesame Intelligence has demonstrated confidence in challenging global top-tier levels. As the world’s first panoramic general-purpose high-compute chip, A2000—with its industry’s unique big-core NPU architecture, functional safety enabled by hardware isolation MCU, and an in-house ISP (image signal processor)—has already completed deep adaptation of end-to-end and VLA (vision—language—action) algorithms with leading algorithm providers such as Yuanrong Qixing and Nullmax in advanced intelligent driving domains. The annual report reveals that A2000 has already secured designated-project wins from leading automakers, and is expected to see multiple mass-production projects implemented in 2026. This marks that domestic chips are beginning to have the capability to launch substantive challenges against the overseas giants’ dominance in the 100 TOPS-level high-compute market.

The second growth engine: the SesameX platform opening a new paradigm for embodied intelligence

If intelligent driving chips are the company’s “yesterday” and “today,” then the embodied intelligence solution is a key move by Black Sesame Intelligence betting on “tomorrow.”

In 2025, the industry regards it as the “Year of Embodied Intelligence,” but most chipmakers still remain at the stage of providing computing hardware. Black Sesame Intelligence aims to redefine the game rules by launching the SesameX multidimensional embodied intelligence platform. Through three core modules—Kalos, Aura, and Liora—the platform provides “full-stack compute” from bottom-layer foundational control to top-layer advanced cognition. This is not simply a stack of chips, but an “operating system–level” solution optimized for the entire workflow of robot perception, decision-making, and execution.

This “turnkey” model hits the pain points in the current robotics industry chain. Because forms such as humanoid robots and quadruped robots vary greatly and fragmentation is severe, original equipment manufacturers urgently need standardized compute base layers. In 2025, Black Sesame Intelligence quickly allied with key players in the robotics industry chain such as Cloud Deep, Fourier Intelligence, Lenovo, and Zhiping Square, and achieved scaled delivery in quadruped robot and maritime shipping inspection scenarios. The financial report shows that in the starting year, this business contributed RMB 96.3 million in revenue, successfully establishing a dual-business growth pattern of “smart vehicles + embodied intelligence.”

Strategic acquisition: completing the “last mile” of edge AI

Black Sesame Intelligence’s acquisition of Zhuhai Yizhi Electronics initiated at the end of 2025 is an important piece to understand its “edge AI full-stack layout” strategy.

Previously, Black Sesame Intelligence’s advantages were concentrated in automotive-grade high-compute chips, leaving gaps in the vast entry-level AIoT (artificial intelligence Internet of Things) markets such as smart door locks, AI glasses, and smart home appliances. Yizhi Electronics is precisely a leading player in AI SoCs (system-level chips) within China’s intelligent hardware sector.

The strategic logic of this acquisition lies in “connecting the compute spectrum.” By integrating Yizhi Electronics, Black Sesame Intelligence’s existing product matrix achieves coverage of the full high-to-low range of compute—from 2T to nearly 600T. This means that whether it is an autonomous driving vehicle needing large-model inference capability, or AI toys and floor-cleaning robots that require low-power, low-cost edge computing, Black Sesame Intelligence can provide a complete solution from chips to algorithms.

In the intelligent imaging field, this synergy is beginning to show. The company not only reached a cooperation with Li Auto, providing tailored imaging algorithms for its AI glasses Livis, but also achieved scaled deployment across broad AI scenarios such as smart traffic management and parking systems. After filling the shortcomings through acquisition, Black Sesame Intelligence truly has the confidence to compete on the same stage with global giants in the era of everything becoming intelligent and connected.

R&D depth: building a technology moat for the future

Behind high growth is sustained, high-intensity R&D investment. In 2025, the company’s R&D expenditures were RMB 1.417 billion, roughly flat compared with the previous year, but the R&D ratio relative to revenue is returning to rationality as scale effects take hold. This indicates that the company has moved past the early stage of rough, exploratory experimentation, and that R&D spending has entered a more focused, efficient “precise strikes” phase.

Looking ahead to 2026, the company’s R&D roadmap clearly points to “productized deployments.” On one hand, the company will complete the delivery of its next-generation intelligent vision processing IP and NPU (neural network processor) IP, enabling new chip tape-outs, and plans to release the next-generation automotive-grade computing platform. On the other hand, in building the ecosystem of the toolchain, Black Sesame Intelligence is improving customer development efficiency by more than 50% by launching a chip support kit that enables seamless integration with mainstream AUTOSAR (automotive open system architecture).

This shift from “selling chips” to “selling ecosystems” is an important marker of how chip companies move toward maturity. By lowering the threshold for customers’ secondary development, Black Sesame Intelligence is building a developer community around its own hardware ecosystem, thereby locking in customers over a longer time horizon.

As the gradual integration with Yizhi Electronics is completed, Black Sesame Intelligence is no longer a single “autonomous driving chip company,” but an edge AI inference platform giant spanning automobiles, robotics, and AIoT. Driven by “technology + scenarios” on two wheels, this company is trying to become a leader that defines China’s standards in the long-distance race of edge AI.

A massive amount of information and precise analysis—everything is on the Sina Finance app