10 Ways BYD Is Using AI [Case Studies] [2026]

BYD offers one of the most instructive examples of how artificial intelligence is moving from isolated software features into the operating architecture of a large industrial enterprise. Its AI footprint now spans advanced driver assistance, whole-vehicle intelligence, generative AI, computer vision, semiconductor design, enterprise analytics, autonomous intralogistics, and humanoid robotics.

The business environment makes these investments particularly significant. BYD generated approximately RMB 804 billion ($116 billion) in 2025 revenue and sold about 4.6 million new-energy vehicles, but net profit fell 19% to RMB 32.6 billion, reflecting severe pricing and competitive pressure in China’s EV market. Automotive and related-product gross margin declined by 1.8 percentage points to 20.5%. BYD nevertheless increased annual R&D spending to RMB 63.4 billion, up 17%, meaning R&D expenditure was almost twice reported net profit.

Due to continued pressure, BYD reported first-quarter revenue of RMB 150.2 billion, down 11.82% year over year, while attributable net profit declined 55.38% to RMB 4.08 billion. At the same time, development expenditures rose 38.75% from year-end 2025 levels as BYD increased internal R&D investment. These figures should not be attributed to AI, but they provide important context: BYD is investing aggressively in intelligence while facing meaningful profitability pressure. BYD has also committed to more than RMB 100 billion of continuing investment in intelligent technologies.

What makes BYD particularly relevant for CEOs and Chief AI Officers is therefore not simply how many AI products it has launched. It is how the company is attempting to combine proprietary data, models, chips, sensors, manufacturing systems, robotics, and physical products into reinforcing technology platforms. At DigitalDefynd, this discussion therefore covers case studies of how the company used AI technologies to strengthen its product portfolio and marketing positioning.

 

Index

1. BYD Built XUANJI as a Whole-Vehicle AI Architecture
2. BYD Used AI to Move Advanced Driver Assistance from $30,000 Vehicles into Cars Below $10,000
3. AI Parking Usage Rose From 21% to 93% After BYD Added Financial Risk Coverage
4. BYD Developed a 4nm AI Driving Chip to Control More of Its Compute Economics
5. BYD Is Evolving DiLink from Voice Commands into Generative and Agentic AI
6. BYD Uses AI Driver Monitoring to Support Safety and European Market Compliance
7. BYD Electronics Uses Predictive Analytics to Turn Factory Data into Operational Decisions
8. BYD Uses Computer Vision to Automate Factory Safety Monitoring
9. BYD Uses Autonomous Mobile Robots to Digitize Battery-Plant Material Movement
10. Humanoid-Robot Training at BYD Doubled Efficiency and Improved Stability by 30%

 

10 Ways BYD Is Using AI [Case Studies] [2026]

1. BYD Built XUANJI as a Whole-Vehicle AI Architecture

Challenge

Automotive intelligence traditionally evolved in separate domains. Infotainment, driver assistance, suspension, powertrain control, body electronics, parking, and cabin systems could each have their own sensors, controllers, software, and decision logic.

That fragmentation becomes increasingly inefficient as vehicles become software-defined. A sophisticated maneuver may simultaneously require perception, trajectory planning, propulsion, braking, steering, suspension adjustment, and cabin interaction.

BYD had already been working on reducing software fragmentation before XUANJI. When it introduced e-Platform 3.0 in 2021, the company said its new electrical/electronic architecture and BYD OS could reduce new-function iteration cycles from about two months to two weeks—a reduction of more than 70%. That metric related to the platform architecture rather than AI itself, but it illustrates the development-time problem BYD was attempting to solve.

 

How BYD Implemented AI

Key Technologies: XUANJI Architecture, XUANJI AI Large Model, multimodal AI, centralized vehicle intelligence, millisecond-level information fusion, BYD OS and domain controllers.

In January 2024, BYD introduced its Integrated Vehicle Intelligence strategy and self-developed XUANJI Architecture.

Instead of confining intelligence to the cockpit or ADAS, BYD describes XUANJI as both the vehicle’s “brain” and “neural network.” The system collects information from inside and outside the vehicle, consolidates that information at the millisecond level, sends it to a central decision layer, and coordinates execution across vehicle systems.

At its center is the XUANJI AI Large Model, which BYD describes as a dual-cycle multimodal vehicle model.

According to BYD, it covers more than 300 whole-vehicle scenarios, giving the architecture the ability to coordinate AI across driving, vehicle control, cabin interaction, and other systems.

BYD subsequently upgraded this stack to XUANJI Architecture 2.0, incorporating a centralized “cockpit-driving-electric” computing brain, a new sensor architecture, a physical-AI large model, and a self-evolving data flywheel.

 

Results and Business Impact

The clearest measured benefit in BYD’s architectural journey is software-development speed.

• Earlier e-Platform 3.0 architecture reduced BYD’s stated software-feature iteration cycle from two months to two weeks, a 70%+ reduction.
• XUANJI’s AI model expanded intelligence across 300+ vehicle scenarios.
• Information aggregation and decision-making occur on a millisecond-scale architecture, according to BYD.
• The architecture has since evolved into a centralized computing platform supporting cockpit, driving, and electrical-control functions.

The executive implication is that BYD is building an AI platform rather than a collection of AI projects. A common architecture allows data, compute, models, and vehicle capabilities to be reused across multiple functions.

BYD has not disclosed the engineering-cost reduction attributable specifically to XUANJI, so any dollar ROI would be speculative.

 

Related: Ways Tesla is using AI [Case Study]

 

2. BYD Used AI to Move Advanced Driver Assistance from $30,000 Vehicles into Cars Below $10,000

Challenge

Historically, advanced driver-assistance systems were economically easier to deploy in premium vehicles because the technology requires processors, cameras, radar, sometimes LiDAR, substantial software engineering, continuous data collection, and OTA infrastructure.

That created a price barrier.

Reuters reported that before BYD’s 2025 expansion, comparable advanced driving functions within BYD’s range had generally started on vehicles priced around $30,000.

BYD’s challenge was therefore not simply developing an AI driving model. It was reducing the technology’s cost sufficiently to make advanced assistance economically viable in mass-market cars.

 

How BYD Implemented AI

Key Technologies: God’s Eye/DiPilot, end-to-end driving algorithms, camera perception, millimeter-wave radar, ultrasonic sensing, LiDAR on higher-tier systems, vehicle-cloud data flywheel and OTA model updates.

In February 2025, BYD launched its “Intelligent Driving for All” strategy around three principal God’s Eye configurations:

• God’s Eye A / DiPilot 600 — higher-end architecture initially associated with YANGWANG.
• God’s Eye B / DiPilot 300 — higher-level platform incorporating LiDAR.
• God’s Eye C / DiPilot 100 — the mass-market system for BYD-branded vehicles.

God’s Eye C uses BYD’s 5R12V perception configuration—five millimeter-wave radars and 12 cameras—alongside end-to-end control algorithms. It supports functions including highway navigation assistance, autonomous lane changes, ramp entry and exit, obstacle avoidance, lane keeping, and cruise control under driver supervision.

Some current mass-market implementations use extremely dense sensor packages. BYD’s God’s Eye-equipped Seagull, for example, is marketed with 29 high-precision sensors and 360-degree surround-view capabilities.

 

Results and Business Impact

The first rollout immediately covered 21 models. Most significantly, Reuters reported that God’s Eye reached the Seagull at a Chinese starting price of RMB 69,800, approximately $9,555 at the time.

Compared with the roughly $30,000 entry point Reuters cited for BYD’s previous advanced-driving offerings, BYD brought the minimum vehicle price associated with such functionality down by roughly two-thirds. That is a price-access comparison, not a claim that BYD reduced the underlying AI system’s cost by the same percentage.

BYD’s AI scale has also expanded sharply.

In February 2025, the company said it had cumulatively sold more than 4.4 million vehicles equipped with Level 2 or higher driving assistance, supported by more than 5,000 intelligent-driving engineers.

By May 2026, BYD reported more than 3.15 million vehicles specifically using its current intelligent-driving assistance technology, with the God’s Eye fleet generating more than 200 million kilometers of driving data every day.

At that stated run rate, BYD’s driving systems can potentially observe roughly 6 billion kilometers of new driving data in a 30-day period. This is an arithmetic extrapolation from BYD’s reported daily figure, not a separately reported company KPI.

For AI leaders, the significance is the combination of low-cost distribution and massive data generation. BYD is using product scale as an AI-learning asset.

BYD has not disclosed the incremental gross profit or per-vehicle manufacturing cost associated with God’s Eye.

 

3. AI Parking Usage Rose From 21% to 93% After BYD Added Financial Risk Coverage

Challenge

AI features create little value when customers do not trust them enough to use them.

Automated parking is particularly sensitive because even a relatively minor perception or trajectory error can cause visible vehicle damage, repair costs, insurance claims, and customer dissatisfaction.

BYD therefore faced both an AI problem and a behavioral-adoption problem.

 

How BYD Implemented AI

Key Technologies: multi-camera vision, ultrasonic sensors, millimeter-wave radar, automated trajectory planning, integrated vehicle control, and God’s Eye driving algorithms.

BYD’s automated parking system interprets surrounding objects and available parking geometry through vehicle sensors, then controls steering, braking, acceleration and maneuver planning.

The broader God’s Eye platform has been developed to cover more than 300 parking scenarios, according to BYD. Functions available across different configurations include assisted parking, remote parking and advanced automated parking maneuvers.

BYD then introduced an unusual trust mechanism.

Beginning in July 2025, the company offered damage coverage for eligible use of its intelligent-parking functionality in China, effectively putting BYD’s balance sheet behind the AI experience.

 

Results and Business Impact

BYD reported one of its clearest AI adoption metrics in May 2026.

After implementation of the parking-coverage program:

• Reported God’s Eye intelligent-parking usage increased from 21% to 93%.
• That represents a 72-percentage-point increase.
• Usage became approximately 4.4 times the original level.
• BYD also claimed that the intelligent-parking accident rate had become “almost zero.” That safety statistic is company-reported and has not been independently audited publicly.

BYD then expanded the concept by announcing a one-year damage-coverage program for eligible Urban NOA use. Under stated conditions, BYD said it would directly cover qualifying vehicle repair, third-party property damage and personal-injury losses, with no coverage cap and without affecting the customer’s following-year commercial insurance premium.

For executives, the key insight may be more important than the parking algorithm itself.

BYD appears to have discovered that commercial risk-sharing can be an AI-adoption mechanism.

Enterprises deploying AI into high-consequence workflows may similarly need warranties, human-review commitments, service guarantees or liability policies—not merely better models.

 

Related: Ways Hyundai is using AI [Case Study]

 

4. BYD Developed a 4nm AI Driving Chip to Control More of Its Compute Economics

Challenge

Sophisticated automotive AI consumes enormous computing resources.

As perception models become multimodal and vehicles add more cameras, radar, LiDAR, and real-time planning, automakers face increasing processor cost, power consumption, heat-management requirements, and dependence on external semiconductor suppliers.

The challenge is not only achieving more TOPS. It is obtaining usable computing performance per watt and per dollar while controlling the relationship between software and hardware.

 

How BYD Implemented AI

Key Technologies: XUANJI A3 SoC, 4nm semiconductor process, physical-AI models, hardware-software co-design, centralized vehicle computing and multi-chip parallel processing.

In May 2026, BYD unveiled the XUANJI A3, which it describes as China’s first domestically developed 4nm automotive-grade intelligent-driving SoC.

According to the company:

• A three-chip configuration delivers more than 2,100 TOPS.
• The architecture supports future L3/L4 applications.
• Power consumed per unit of compute is 20% lower than comparable products.
• Software optimization around BYD’s own algorithms increases compute utilization by 100%, effectively doubling utilization.

The semiconductor operation behind the chip is itself substantial.

BYD says its chip R&D organization has more than 7,000 employees, with four R&D bases and five wafer-manufacturing plants. The company says it has developed more than 2,000 chip products overall, including 567 automotive-grade chip products across 13 categories, with products used by 46 domestic and international vehicle brands.

 

Results and Business Impact

The measurable outcome is primarily compute efficiency and vertical integration rather than disclosed financial savings.

Key verified/company-reported metrics:

• 4nm automotive AI process technology.
• 2,100+ TOPS using three coordinated chips.
• 20% lower power per TOPS versus comparable chips, according to BYD.
• 100% increase in computing utilization through BYD-specific software optimization, according to the company.
• 7,000+ chip R&D personnel.
• Five wafer fabs and four R&D bases.

This is an important enterprise-AI strategy lesson.

Models become economically more defensible when companies can optimize the infrastructure underneath them. BYD is pursuing hardware-software co-design in much the same way leading AI cloud companies design custom accelerators around their workloads.

However, chip design does not mean complete semiconductor independence. Advanced fabrication remains connected to broader global semiconductor equipment and manufacturing supply chains.

BYD has not disclosed A3’s manufacturing cost or quantified savings relative to Nvidia, Huawei, Horizon Robotics, or other external computing platforms.

 

5. BYD Is Evolving DiLink from Voice Commands into Generative and Agentic AI

Challenge

Traditional automotive voice systems are essentially command processors.

They can execute requests such as adjusting temperature or opening a window, but usually struggle with contextual conversation, general knowledge, multi-step requests, and ambiguous human intent.

As consumers become accustomed to large language models on smartphones and computers, a fixed-command vehicle interface increasingly feels dated.

 

How BYD Implemented AI

Key Technologies: DiLink 100/150, DeepSeek large models, natural-language processing, four-zone speech recognition, cloud connectivity, multimodal interaction, and the DidiXia AI agent.

BYD’s earlier DiLink systems already supported sizable command libraries. One officially documented implementation offered 70+ AI voice-control functions, alongside cloud-based vehicle control and monitoring capabilities.

The newer DiLink 150 platform adds four-zone AI voice interaction, allowing the system to determine where speech originates in the cabin and support more contextual passenger interaction.

BYD has also embedded DeepSeek models into production vehicles. The current Sea Lion 05 DM-i and Sea Lion 05 EV, for example, explicitly advertise DeepSeek-powered large-model question answering through DiLink 150.

Other BYD vehicles advertise “full” DeepSeek integration alongside smart-cockpit and connected-services functions.

In May 2026, BYD advanced the concept further with its DidiXia super intelligent agent inside the DiLink AI Intelligent Cockpit.

BYD says the system combines natural dialogue, large-model reasoning, intent understanding, task decomposition, and agent execution so that the cockpit can move from answering questions toward proactively completing tasks.

 

Results and Business Impact

The clearest measurable story is the widening functional scope.

BYD has progressed from:

70+ command-based voice functions → four-zone AI interaction → DeepSeek large-model knowledge functions → agentic task execution.

That progression is strategically significant because generative AI is being embedded into an existing control environment rather than delivered as a standalone chatbot.

The agent can potentially become the orchestration layer connecting navigation, entertainment, vehicle settings, external digital services and eventually connected-home ecosystems.

BYD’s November 2025 strategic partnership with Midea, for example, specifically identified AI agents, standardized interfaces, data protocols and vehicle-home collaborative decision-making as cooperation areas.

Public disclosures do not yet provide hard metrics for user-engagement growth, touchscreen time saved, service revenue generated or customer-satisfaction improvement from DeepSeek or DidiXia.

For a C-suite audience, the important lesson is therefore architectural: generative AI creates more value when it is given controlled access to operational systems.

 

Related: Ways Porsche Is Using AI [Case Study]

 

6. BYD Uses AI Driver Monitoring to Support Safety and European Market Compliance

Challenge

Level 2 driving assistance still requires an attentive human driver.

Fatigue, phone use, and visual distraction can therefore undermine even sophisticated ADAS.

The challenge became more important in Europe because driver-monitoring capabilities increasingly form part of mandatory vehicle-safety requirements.

Under EU rules, Advanced Driver Distraction Warning systems generally activate above 20 km/h and may use up to one minute for initial calibration. EU Driver Drowsiness and Attention Warning rules require systems to warn drivers at a defined drowsiness threshold, including the equivalent of Level 8 on the Karolinska Sleepiness Scale, with Level 7 warnings also permitted.

 

How BYD Implemented AI

Key Technologies: computer vision, gaze estimation, drowsiness recognition, behavioral classification, driver-state inference, and human-machine alerts.

BYD’s Driver Monitoring System analyzes driving and driver behavior to identify risks including:

• Fatigue.
• Long eye closure.
• Driver distraction.
• Phone use.
• Attention directed away from the road.

BYD developed both indirect and direct monitoring approaches for different regulatory requirements.

Privacy was also incorporated into the design. BYD states that users can physically close the DMS camera cover, while its broader vehicle privacy framework emphasizes data minimization and local processing where applicable.

 

Results and Business Impact

BYD’s DMS was selected for China’s 2024 national list of outstanding new digital-service cases, following company application, government recommendation and expert evaluation.

More importantly for international expansion, BYD says it became the first Chinese automaker to pass both the European DDAW and ADDW regulatory requirements, with the technology deployed in domestic and overseas production models.

The business value is therefore partly regulatory.

AI driver monitoring does not simply provide another safety feature; it can become an enabler of market access and homologation as safety regulation becomes more software-intensive.

That is relevant for CEOs in every regulated industry. Some AI investments may generate their return not through direct cost savings but by satisfying compliance, qualification, or safety requirements that allow a product to enter or remain in a market.

BYD has not published a verified crash-reduction percentage attributable specifically to DMS.

 

7. BYD Electronics Uses Predictive Analytics to Turn Factory Data into Operational Decisions

Challenge

Large factories produce enormous volumes of data from machines, quality systems, warehouses, production planning, operators, maintenance systems, and supply-chain workflows.

The management problem is often not insufficient data. It is fragmentation.

If those systems remain disconnected, managers may know that a problem occurred without understanding where it started, how serious it is, or what action should happen next.

 

How BYD Implemented AI

Key Technologies: E-BI enterprise intelligence, predictive analytics, MES, WMS, data modeling, data mining, multi-level alerts, visualization and closed-loop workflow management.

BYD Electronics has built an Industry 4.0 architecture in which production systems and analytics are developed together.

Its smart-factory platform uses real-time data monitoring and predictive analysis to identify potential operational issues early.

A core component is the internally developed E-BI enterprise intelligence platform.

E-BI covers:

• Data acquisition.
• Data calculation.
• Data storage.
• Modeling.
• Analysis and mining.
• Prediction.
• Visualization.
• Closed-loop operational warnings.

The platform connects with internally developed systems including E-MES, E-WMS, EOMS, E-OPS and E-PLM.

BYD Electronics states that it has almost 20 years of IT system-development experience, operates with 100% independent R&D capability for this stack, and has built nine information systems centered around MES.

 

Results and Business Impact

BYD Electronics explicitly says predictive monitoring is used to prevent and address problems earlier and to reduce production risk and cost.

The company has not disclosed a current percentage reduction in downtime, defects, or manufacturing cost attributable to E-BI.

Nevertheless, several measurable characteristics demonstrate operating maturity:

• Nine integrated information systems around MES.
• Nearly two decades of system-development experience.
• 100% internally developed capability for the relevant information-system stack.
• Multi-level automated alerts distributed through dashboards, email, and enterprise messaging.

The strategic lesson is that enterprise AI becomes more valuable when connected to the systems that actually operate the business.

Without reliable MES, warehouse, lifecycle, and equipment data, predictive analytics risks becoming another dashboard. BYD’s design connects analytics to operating workflows.

 

Related: AI Salaries in the Middle East

 

8. BYD Uses Computer Vision to Automate Factory Safety Monitoring

Challenge

Traditional factory surveillance is largely passive.

Cameras record events, but humans still need to watch footage, identify noncompliance and initiate corrective action.

That approach becomes difficult to scale in automotive plants characterized by large production areas, dense worker movement, machinery, forklifts, and fast production cycles.

BYD’s manufacturing scale makes the challenge particularly large. The company sold approximately 4.6 million new-energy vehicles during 2025.

 

How BYD Implemented AI

Key Technologies: deep-learning computer vision, PPE recognition, human-behavior detection, smoke/fire recognition, intrusion detection, real-time alerts and production-control integration.

Computer-vision company Extreme Vision publicly documents a named BYD project built around an AI environment, health and safety management platform.

The deployment includes models for:

Worker behavior

• Leaving-post detection.
• Sleeping detection.
• Mobile-phone-use detection.

PPE and compliance

• Safety-helmet recognition.
• Workwear recognition.

Environmental safety

• Flame detection.
• Smoke detection.
• Restricted-area intrusion.
• Traffic/signal-light recognition.

Extreme Vision says its team performed repeated factory-specific data collection, analysis, parameter adjustment and algorithm optimization to compensate for frequent personnel movement, visual interference and difficult lighting conditions.

Alerts are connected directly to BYD’s production dispatch/control workflows rather than being left as passive video footage.

 

Results and Business Impact

Extreme Vision says its customized work significantly improved algorithm accuracy and adaptability in the BYD workshop environment. The supplier also describes the system as protecting a “super production line” operating at approximately one new-energy vehicle per minute. This is partner-reported operating context, not a BYD financial disclosure.

Public sources do not disclose a BYD-specific percentage reduction in:

• Safety incidents.
• Manual inspection labor.
• Insurance expense.
• Lost production hours.
• Regulatory violations.

Those values should not be invented.

The business impact instead lies in moving supervision from human observation to event-driven monitoring.

Rather than asking safety personnel to continuously watch cameras, algorithms identify potentially actionable behavior and route it into the operating workflow.

For industrial companies, that is a useful model for applying AI to existing infrastructure: the value may come from making cameras, sensors, and operational data active decision tools rather than installing entirely new systems.

 

9. BYD Uses Autonomous Mobile Robots to Digitize Battery-Plant Material Movement

Challenge

Material handling is a major hidden source of manufacturing inefficiency.

Battery-production lines require components, large racks and pallets to reach specific workstations precisely when needed. Manual forklifts and carts can create idle time, labor intensity, congestion, and safety exposure.

ForwardX, the robotics supplier involved in BYD’s deployment, says high operating costs and inefficient labor were among the reasons BYD automated this workflow.

 

How BYD Implemented AI

Key Technologies: autonomous mobile robots, autonomous forklifts, LiDAR SLAM, visual-semantic positioning, autonomous obstacle avoidance, fleet-management software, PDA task assignment and manufacturing-system integration.

In the named BYD battery implementation documented by ForwardX, the company deployed:

• 9 Max-series AMRs.
• 6 customized Apex autonomous forklifts.
• At least 15 autonomous material-handling vehicles in the documented configuration.

The system had to handle unusually large materials, including pallets measuring approximately 2.4 × 1.36 meters.

Employees initiate tasks through handheld PDAs. The robot fleet then transports material and communicates with the factory’s existing management system, allowing operations to be visualized and managed in real time.

ForwardX’s Apex platform uses technologies including laser SLAM, visual-semantic positioning, LiDAR sensing, and dynamic route planning.

 

Results and Business Impact

The strongest published BYD-specific outcome is process digitization.

ForwardX reports:

• 100% digitized process management for the automated workflow.
• Autonomous handling of oversized 2.4 × 1.36-meter pallets.
• A documented fleet of nine AMRs plus six autonomous forklifts in the detailed case study.
• Integration with BYD’s existing factory-management system for end-to-end process visibility.

The supplier does not disclose a defensible BYD-specific percentage for labor reduction, cost savings, throughput improvement, or ROI.

That distinction is important.

ForwardX publishes broader manufacturing benchmarks showing that AMR deployments elsewhere can produce large productivity improvements, but those figures cannot legitimately be transferred to BYD.

For COOs, the more important transformation is the closed loop:

digital production request → autonomous task assignment → physical material movement → real-time system visibility.

AI is no longer merely analyzing the factory; it is executing work inside it.

 

10. Humanoid-Robot Training at BYD Doubled Efficiency and Improved Stability by 30%

Challenge

Traditional industrial automation works best in structured environments.

A robot arm can repeatedly weld the same point or move the same component extremely efficiently, but factories still contain tasks designed around human bodies: walking between locations, reaching irregular spaces, manipulating varied items, and adapting to changing environments.

Humanoid robots promise greater flexibility, but their economic value depends on whether they can become reliable enough for continuous industrial operation.

 

How BYD Implemented AI

Key Technologies: UBTECH Walker S1, large language models for task planning, multimodal perception, semantic VSLAM, whole-body motion control, robotic manipulation, autonomous logistics orchestration, and multi-agent coordination.

In 2024, UBTECH deployed its Walker S1 humanoid robot for factory training at BYD.

Walker S1 uses:

• Large-model-based general task planning.
• Natural-language intent understanding.
• Multimodal environmental perception.
• Semantic VSLAM navigation.
• Learning-based whole-body motion control.
• Robotic manipulation.
• Integration with manufacturing and logistics systems.

UBTECH says Walker S1 subsequently participated in a BYD material-handling scenario coordinating humanoid robots with autonomous logistics vehicles, AMRs/AGVs, and intelligent factory-management systems.

 

Results and Business Impact

This case provides one of the strongest quantified manufacturing-AI outcomes publicly available for BYD.

UBTECH’s 2024 annual report states that after the first training stage at a BYD automobile plant:

• Operational efficiency doubled.
• Robot stability improved by 30%.

These are supplier-reported results published in UBTECH’s annual report; they should not be interpreted as evidence that the robot became twice as productive as a human employee.

The improvement concerned the robot’s own performance over the course of industrial training.

That distinction matters because humanoid robotics remains immature.

UBTECH has continued positioning humanoids for automotive manufacturing, but current industrial humanoid systems are still developing toward human-level speed, reliability and economics.

The strategic importance to BYD is therefore not immediate workforce replacement. It is learning-curve acceleration.

By putting embodied-AI systems into live factories, BYD and its robotics partners can collect data about navigation, manipulation, coordination, failure modes and workflow integration.

For manufacturers evaluating physical AI, this is a critical lesson: the first return on humanoid robotics may be operational learning, not immediate labor savings.

 

Conclusion

BYD’s AI strategy shows how artificial intelligence can evolve from a collection of digital features into a core operating capability. By combining intelligent driving, generative AI, proprietary computing, predictive analytics, computer vision, autonomous logistics, and robotics, the company is embedding intelligence across both its products and manufacturing ecosystem.

The broader lesson for business leaders is clear: sustainable AI advantage comes from connecting proprietary data, domain expertise, scalable technology platforms, and real-world execution. BYD’s approach demonstrates how these elements can reinforce one another to improve product intelligence, accelerate innovation, and create new operating efficiencies at scale.

For corporate executives looking to build similar capabilities within their organizations, explore our featured AI Executive and Leadership Programs to strengthen your understanding of AI strategy, governance, transformation, and enterprise implementation.