Why is Chery an AI company? It used AI to rebuild the way it makes cars, turned that capability into a reusable foundation, and let it grow into new species beyond automobiles.
Author | Wang Ruihao Editor | Li Yuchen
"Several leaders told me that Chery's cars feel less and less like cars, and that Chery feels less and less like a car company. That's exactly what I want to hear — Chery is a high-tech innovation enterprise 'wearing the vest of a car company.'"
On October 20, the 2026 Chery Global Innovation Conference opened in Wuhu, Anhui. The man who said this was Yin Tongyue, Chairman of Chery Automobile, who has made cars for more than forty years. A man who has devoted most of his life to automobiles, voluntarily taking off the 'car company' nameplate — this is not modesty, it is a judgment.
His reasoning rests on a metaphor: in 1886, when Benz built the first automobile, those left behind by the times were the horse breeders who had driven horse carriages all their lives. They did nothing wrong; they just failed to see that the road had changed. "The real risk is not that AI is too strong or too fast, but that we cannot keep being the horse breeders of the old days." Yin Tongyue said, "In the future, what AI will truly eliminate are the enterprises that reject AI and dare not use AI."
So Chery's choice is very direct: turn the blade inward, proactively use AI to give itself a genetic recombination — first change itself, then change the world.
Lei Feng Network (WeChat public account: Lei Feng Network) believes that judging whether Chery is an AI company should not depend on how many times it shouts AI at launch events, but on three things: whether AI has restructured its production processes, whether it has precipitated into a reusable unified foundation, and whether it has grown new species beyond automobiles.
PART 1
Chery forged 30 years of car-making experience into an AI foundation
Generic large models on the market are like an encyclopedia — they can chat about anything, but once they enter a factory workshop they fall short: not precise enough, not deep enough, not reliable enough. Chery's choice was to self-develop a vertical foundational large model, CheryMind, which has three major characteristics: a brain that understands the physical world, co-evolution with the business, and deployability across all domains.
What supports it is not scraped internet corpora, but the assets Chery accumulated over nearly 30 years: 700TB of enterprise knowledge covering technical standards, process specifications, design guidelines, and safety standards; 200PB of engineering data, covering the full chain from simulation results to test data, from CAD to real vehicles; tens of millions of R&D documents; 25 million coach-level driving data records; and 5 million robot motion-capture data records.
In the era of physical AI, the moat is not how much data you have, but where the data comes from. Generic large models scrape the internet; CheryMind feeds on the technical standards, process specifications, and simulation and test data accumulated over 30 years of car making. This is a textbook that others cannot buy and cannot scrape away.
Lei Feng Network has learned that it is already running, not just a PowerPoint: 1500+ end-to-end business processes, 400+ enterprise systems integrated, 60,000+ daily employee uses, covering 200+ business scenarios. Gao Xinhua, Executive Vice President and CTO of Chery Automobile, gave a comparison on site: previously, one round of simulation optimization by an engineer required 6 iterations and took over a hundred days; now, based on CheryMind, each person gets results for 100 schemes in half a day, a 42-fold efficiency improvement.
CheryMind walks on two legs. Enterprise-side vertical models make AI better understand car making: CAD vertical model, quality vertical model, manufacturing vertical model, enterprise Agent vertical model, covering the full chain of R&D — manufacturing — quality — supply — marketing — operations; product-side vertical models make cars better understand humans and the physical world: smart cockpit, intelligent driving, in-vehicle intelligence, flight control vertical model, MoAgent, agricultural machinery OS, extending all the way to eVTOL, robots, and agricultural machinery.
The key lies in the closed loop: the new capabilities formed by vertical models in real scenarios continuously flow back into the base model. The two engines of business and products feed each other, and CheryMind grows stronger with use. Self-evolution is the hardest standard for judging whether a company is an AI company — tools do not self-evolve, foundations do.
Beyond CheryMind, Chery released self-developed chips, named after Ziwei — the North Star, around which all stars revolve, pointing the direction. The Ziwei chips come in three series, each handling one thing.
The C series handles thinking. The central computing chip C100 delivers 2400TOPS of whole-vehicle NPU computing power and 600GB/s memory bandwidth, supports on-device deployment of 100B large models, and completes the E2E algorithm closed loop from perception to decision in 80ms — less than the time of a single human blink. Benchmarked against industry flagships: 2.2x bandwidth, 1.2x computing power, 1.5x image processing.
The S series handles energy. The automotive-grade silicon carbide S1500 withstands 1500V, the highest in the industry; it outputs 1450A, with efficiency ≥99.6% and junction temperature of 200℃, and mass-production costs drop by more than 30% thanks to 8-inch wafers.
The M series handles control. Automotive-grade control chips: the MC200 domain controller, MS100 pre-driver, MX100 PMIC, and MC100 electric drive, landing in sequence from 2027.
Behind these three things lies Chery's ledger for the past five years: R&D investment exceeding 110 billion yuan, 313 Yaoguang laboratories, and intelligent driving penetration up 267% in two years. What AI is rewriting is the cost curve of car making itself, not marketing rhetoric.
Even more noteworthy is a layer of synergy: the large model defines what the chip should look like — 100B on-device deployment forces 2400TOPS of computing power and 600GB/s bandwidth; the chip in turn determines how deeply the model can be deployed. This 'model—chip' co-definition is a hallmark of technology companies, not integrators.
A saying of Yin Tongyue fits well here: Chery is "insatiably greedy for technology, but timid as a mouse on safety." AI can replace skills, but it cannot replace humanity, responsibility, and warmth.
PART 2
AWE, Falcon, Lingxi, Robotaxi and Robots
What this conference truly released was not a single technology. Yin Tongyue's summary was: mobility, energy, food, and life are not four isolated directions — they share models, chips, data, and engineering capabilities, while continuously generating new demands in real scenarios. The structure is 'one foundation × N scenarios.'
The foundation was first applied to the hardest scenario: taking intelligent driving overseas. In Southeast Asia, two-wheelers and cars share lanes, intersections with low light see wrong-way driving, and there are forced cut-ins... Solutions that worked in the Chinese market may not hold overseas.
Chery's judgment is: in global competition, what should be compared is not computing power, data, or the number of lidars, but the model foundation. Without a unified foundation, every new market entered and every new car launched would require redoing an entire solution.
Chery's All-World Engine (CHERY AWE) handles this with three models.
The scene-construction model is responsible for "understanding the world": it has built 700,000+ scenes, of which 300,000+ are high-risk and 20,000+ extreme, uniformly encoding five categories of data—multi-view images, LiDAR point clouds, scene structure, descriptive text, and driving trajectories—into a latent space, supporting closed-loop simulation of long-tail scenarios. The language-comprehension model is responsible for "reading semantics": multimodal reasoning plus short-term memory and a language topology graph, reaching destinations by road signs alone like a veteran driver, outputting traceable reasoning chains. The action-policy model is responsible for "producing behavior": an MoT multi-expert architecture shifting from "passively reacting to danger" to "actively anticipating risk."
The Falcon driver-assistance system has been running on this foundation for a year: meeting regulatory entry requirements in more than 130 countries and regions, earning C-ICAP Five-Star+ and IVISTA G+ certifications, and winning six consecutive championships in the China Intelligent Driving Competition.
Falcon intelligent driving comes in three versions: Falcon 500 offers urban NOA right at entry level, with equal intelligence for gasoline and electric; Falcon 700 covers full-scenario parking-spot-to-parking-spot and cross-level valet parking; Falcon 900 covers highway and urban L3 with full redundancy in sensing, control, and power.
On the cockpit side, the Lingxi 8D intelligent sensing cockpit uses an AIOS full-domain architecture to connect five domains—chassis, cockpit, body, intelligent driving, and powertrain—with vehicle-level signal orchestration and millisecond-level cross-domain response. Outside the cabin it recognizes pedestrian intent; inside it senses emotion and fatigue, upgrading from "people finding functions" to "functions finding people"; eight dimensions—hearing, vision, smell, motion, touch, emotion, health, and social—are orchestrated in a unified way, with one-sentence Skill free orchestration for personalization of each individual. Details like the Boya audio system with 33 speakers, the 43.7-inch PHUD, and S2S end-to-end voice all push toward "cars understanding people better."
It is the industry's first native, factory-installed Robotaxi built for the world from the start, with five natives—architecture, factory installation, experience, operations, globalization—and six layers of safety—passengers, behavior, perception, algorithms, execution, and fleet, targeting 100,000 vehicles in global operation by 2030.
The robots share a common origin with vehicles: 30 years of automotive supply chain, hardware architecture, and quality systems are directly reused, with 31 innovation application laboratories laid out across six major fields. Under the Multi-agent intelligent architecture, MoAgent 2.0, developed based on CheryMind, uses a slow brain for understanding and planning and a fast brain for real-time response, with end-edge-cloud collaboration forming a data loop of "execute—feedback—optimize—re-execute."
The product matrix has already spread out: quadruped/wheeled-legged robot dogs, smart police, medical guidance, Mobao family companionship, industrial and humanoid robots; Moja robots have already gone overseas with 2,000+ units, covering more than 60 countries.
In addition, in terms of energy freedom and food freedom, the Xunlong flash charger delivers 500 kilometers of range from a 5-minute charge, and the solid-state battery pilot line will be put into use in the first half of 2027; smart agricultural machinery cuts energy consumption by 74%, raising yield on the same plot of land from 1,600 jin to 2,000 jin, with the goal of serving 1 billion mu of farmland worldwide by 2035.
When these scenarios are put together, Chery's most unique leverage in doing AI becomes apparent: reuse rate is the gross margin of the AI era. The same set of models, chips, and supply chains migrates across vehicles, Robotaxis, robots, and agricultural machinery, with diminishing marginal costs. This is an advantage only a company born in manufacturing can have in doing AI.
Chery also knows it cannot run through all scenarios on its own. It is moving from product manufacturer to ecosystem organizer: opening technology platforms, scenarios, and industrialization capabilities to suppliers, universities, and entrepreneurs, and co-building the 2050 Future Institute with USTC for long-cycle, fault-tolerant frontier exploration.
So why is Chery an AI company? Because it has used AI to remake the way it builds cars, turned that layer of capability into a reusable foundation, and let it grow new species beyond automobiles.
In Yin Tongyue's words: "Only when technology leaves the laboratory and enters roads, homes, farmland, and energy systems, continuously solving real problems, can it truly create freedom." "
Thirty-year-old Chery no longer wants to be just a car company. That sounds like a slogan, but when a carmaker's robots begin to genuinely solve social pain points, and its models begin to determine what its own chips look like, it is no longer just a slogan.
Chery's New Battle: Going Global, Intelligent Driving, and Head-On Clashes with BYD and Peers
Bigger than a Mercedes S: Can the Da Han EV carry BYD's upward ambitions?
This is an original article by Lei Feng Wang; unauthorized reproduction is prohibited. For details, seethe reprint notice。