– Geely integrates 6 autonomous driving teams (3,000 employees) into Chongqing Qianli Zhijia Technology
– Combines Megvii’s AI algorithms with former Huawei executives’ hardware expertise
– Aims to overcome fragmentation in R&D, data systems, and technical standards
– Short-term goal: Launch competitive L2+ systems; long-term: Build Tesla-like tech ecosystem
Geely’s Bold Consolidation Move
Geely Holding Group’s surprise integration of six autonomous driving units into Chongqing Qianli Zhijia Technology represents a seismic shift in China’s auto industry. Affecting nearly 3,000 engineers across Geely’s research divisions and partner Megvii Technology, this consolidation directly addresses what founder Li Shufu (李书福) identified as Tesla’s core advantage: “Fundamentally, Tesla appears as a carmaker but actually builds online technology platforms—using vehicles as R&D laboratories.” By merging teams from Zeekr, Lotus, ECARX, Freetech, and Megvii’s Mach Drive, Li Shufu follows Musk’s blueprint for integrated innovation.
The Fragmented Landscape
Before consolidation, Geely’s autonomous driving efforts suffered from significant fragmentation:
– Different brands used incompatible systems: Zeekr developed Haojan AI, Lynk & Co relied on ECARX, while Geometry used Mach Drive
– Hardware variations created compatibility headaches with divergent chip architectures
– Data silos prevented cross-platform learning—performance vehicle data couldn’t improve family SUV algorithms
– R&D priorities conflicted between cutting-edge innovation and cost-effective solutions
Resource Drain and Duplication
The redundancy became economically unsustainable. In 2024 alone, Geely invested ¥10.4 billion (US$1.4B) in R&D—4% of total revenue—yet efficiency suffered. Multiple teams developed parallel solutions for identical problems, delaying time-to-market. When Zeekr owners protested delayed autonomous feature updates and Geometry models launched without any self-driving options, the operational costs of fragmentation became undeniable.
Engineering the ‘Software + Hardware’ Solution
Qianli Zhijia’s ownership structure reveals Li Shufu’s integration strategy. Established on June 27 with ¥200 million capital, its shareholders include:
– 30% Megvii (via Mach Drive)
– 30% Geely (via Zhejiang Geely)
– 30% Chongqing government fund
– 5% Lotus
– 5% Megvii-affiliated entity
This tripartite structure balances AI expertise, automotive experience, and regional support—a deliberate approach as Li Shufu follows Musk’s blueprint for cross-industry convergence.
The Algorithm Architects
Megvii’s involvement brings crucial AI capabilities:
– Co-founder Yin Qi (印奇), heading Qianli Technology, pioneered facial recognition systems
– Mach Drive leverages Megvii’s ‘AI Four Dragons’ computer vision expertise
– Their end-to-end model development focuses on perception accuracy and scenario adaptation
However, Megvii’s limited hardware experience created a capability gap—prompting Geely’s complementary hire.
The Hardware Specialists
Former Huawei Auto BU president Wang Jun (王军) anchors the engineering side with:
– Proven ability to transition prototypes to mass production
– Expertise in automotive-grade validation and cost control
– Shortened development cycles from lab to roadway
He’s joined by Huawei veteran Chen Qi, whose sensor integration experience completes the technical puzzle. This ‘algorithm + engineering’ duality enables Li Shufu follows Musk’s blueprint for vertical integration.
Closing the Autonomous Driving Gap
Geely faces urgent market pressure. While competitors advanced:
– XPeng rolled out XNGP to 227 Chinese cities
– Huawei’s ADS 2.0 achieved 1,000 km zero-intervention drives
– BYD deployed ‘Eye of God’ autonomy in ¥70,000 models
Geely’s Geometry lineup lacked any autonomous options—a glaring omission in China’s EV wars. Qianli Zhijia must deliver immediate results while building long-term advantages.
Short-Term Game Plan
Within 12 months, expect:
– L2+ systems on new Zeekr models
– Highway navigation assist for Geometry EVs
– Unified interface across brands
Priority is matching competitors’ core user experiences to eliminate Geely’s ‘autonomy lag’ perception.
Building the Moat
Long-term, Qianli Zhijia aims for:
– Proprietary L3/L4 architecture by 2026
– Scaled sensor fusion technology
– Data lake aggregating 5 million km monthly test data
Targeting Huawei ADS-like capabilities will require overcoming current algorithm training limitations through consolidated data resources.
Beyond Vehicles: Musk’s Multi-Platform Playbook
Li Shufu’s ambitions extend far beyond self-driving cars. Like Tesla, Geely is constructing an integrated tech ecosystem:
Satellite Networks
Geely’s GeeSpace has launched 72 low-orbit satellites—direct counter to SpaceX’s Starlink—enabling centimeter-level positioning for autonomous navigation beyond 5G coverage zones.
Robotics Revolution
Lotus Technology’s robotics division mirrors Tesla’s Optimus, developing humanoid robots for manufacturing and logistics applications. Initial deployments are planned at Geely’s Wuhan plants.
Mobility Services
CaoCao Mobility (Geely’s ride-hailing arm) prepares autonomous taxi services to challenge Tesla’s planned Cybercab network. Pilot programs launch in Hangzhou in Q4 2024.
AI Research Convergence
Mach Drive’s algorithm team collaborates with Megvii’s core AI researchers on large vision models—creating an xAI equivalent within Geely’s ecosystem. This cross-pollination accelerates innovation as Li Shufu follows Musk’s blueprint.
The Integration Obstacle Course
Merging six teams presents monumental challenges beyond technical alignment. Cultural integration poses equal difficulty:
Workflow Clashes
Legacy teams operate at different velocities:
– Zeekr’s agile development contrasts with ECARX’s staged validation
– Megvii’s research-first approach conflicts with Lotus’ performance orientation
Standardizing development processes requires careful change management.
Compensation Disparities
Salary gaps between former Megvii AI engineers (average ¥850,000/year) and Geely automotive developers (¥450,000) risk morale issues. Harmonizing incentive structures is crucial for retention.
Leadership Alignment
With Wang Jun (王军) handling engineering and Yin Qi (印奇) leading algorithms, maintaining decision-making coherence will test the dual-leadership model. Early reports suggest weekly integration committees oversee collaboration.
The Roadmap to Tesla-Level Integration
Geely’s success hinges on executing Li Shufu’s vision where others failed. Great Wall Motors’ Haomo AI struggled with similar consolidation before scaling back ambitions. By contrast, Qianli Zhijia benefits from:
– Chongqing’s robotics manufacturing cluster
– Zhejiang province’s semiconductor subsidies
– State-backed R&D tax incentives
Still, catching Tesla requires overcoming fundamental gaps in data volume—Tesla’s fleet gathers 3 billion miles of real-world driving data annually versus Geely’s estimated 120 million.
Phase-Based Deployment
Insiders reveal a three-stage rollout:
1. 2024: Unified software architecture across brands
2. 2025: Sensor standardization and data platform integration
3. 2026: Proprietary chip development
This incremental approach prevents disruption while progressing toward Musk-level vertical integration.
Investor Expectations
Redefining Mobility’s FutureLi Shufu’s gamble transcends corporate restructuring—it’s a philosophical bet that future mobility winners will be tech integrators, not mere vehicle assemblers. By combining Megvii’s AI prowess, Huawei’s engineering rigor, and Geely’s manufacturing scale under one banner, he’s constructing what internal documents call “an innovation foundry.” As the industry watches Qianli Zhijia’s progress, remember Li Shufu’s own words: “Tesla’s essence is creating laboratories on wheels.” For automakers and investors alike, the imperative is clear: Track integration milestones quarterly, monitor talent retention metrics, and watch for cross-platform technology transfers. The company that masters ecosystem-scale innovation will dominate the next automotive era—and Li Shufu just positioned Geely at the starting line.