Mastering the Software-Defined Vehicle: A Step-by-Step Guide Inspired by Geely’s Auto China 2026 Strategy

Overview

At Auto China 2026, Geely made a bold statement: the future of mobility is not just electric—it’s software-defined. While other automakers focused on flashy exteriors or novel battery chemistries, Geely dedicated its entire booth to demonstrating how a car’s value increasingly comes from its code, not its chassis. This tutorial breaks down Geely’s approach into actionable steps for automakers, suppliers, and tech entrepreneurs who want to build a true software-defined vehicle (SDV) program. You’ll learn the foundational elements, the integration pitfalls, and the mindset shifts required—all illustrated with real examples from the Beijing show floor.

Prerequisites

Before diving into the step-by-step process, ensure you have:

• Familiarity with automotive electronic architectures (ECUs, CAN bus, domain controllers, zonal architectures).
• Basic understanding of cloud and edge computing (OTA updates, telemetry, microservices).
• Access to industry data platforms (e.g., McKinsey SDV reports, IHS Markit telematics forecasts).
• Certification or coursework in functional safety (ISO 26262) is helpful but not mandatory.
• A cross-functional team including software engineers, hardware architects, UX designers, and cybersecurity experts.

Step-by-Step Guide: Implementing an SDV Strategy Like Geely

Step 1: Redefine the Vehicle Architecture (Hardware Abstraction)

Geely’s first move was to decouple hardware from software. They introduced the SEA (Sustainable Experience Architecture)—a modular platform that separates the physical components (motors, batteries, sensors) from the computing layer. To replicate this:

1. Replace legacy distributed ECUs with a centralized compute unit (high-performance SoC like Qualcomm Snapdragon Ride or Nvidia Drive Orin).
2. Implement a service-oriented architecture (SOA) where software functions (e.g., brake-by-wire, infotainment) are exposed as reusable services.
3. Use a hypervisor to run multiple operating systems (QNX for safety, Android for infotainment) on the same hardware.

Code snippet (pseudocode for service definition): Service VehicleBrake { setBrakePercentage(0..100); getBrakeStatus(); }

Step 2: Build an Open Software Platform (Geely’s Galaxy OS)

Geely showcased its Galaxy OS—an open ecosystem that allows third-party developers to create apps for the vehicle. Follow these steps:

1. Choose a base operating system (Linux, Android Automotive, or a derivative).
2. Define API contracts for vehicle functions (doors, climate, ADAS). Use RESTful APIs over Ethernet.
3. Set up a developer portal with SDKs, sandbox environment, and revenue-sharing models.
4. Enable over-the-air (OTA) updates for both firmware and applications.

Example API endpoint: GET /vehicle/climate/temperature?zone=driver returns current setpoint.

Step 3: Rethink the User Experience (Continuous Improvement)

At Auto China, Geely’s booth had a “Living Room” mockup—the car adapts based on driver profile and context. To achieve this:

1. Collect anonymized data on user behavior (seat preferences, navigation history, climate usage).
2. Train on-device ML models for low-latency personalization (e.g., automatic route suggestion based on calendar).
3. Release monthly OTA updates that enhance UX (new UI themes, voice command improvements).
4. Use A/B testing on a subset of vehicles to validate changes before full roll-out.

Step 4: Establish a Cybersecurity Framework

Geely partnered with JD.com and other security firms to ensure the SDV is secure. Your plan should include:

• Secure boot and signed firmware updates.
• Network segmentation between safety-critical (ADAS, braking) and non-critical (infotainment) domains.
• Intrusion detection systems (IDS) monitoring CAN and Ethernet traffic.
• Regular penetration testing and vulnerability disclosure programs.

Step 5: Evolve the Business Model (Revenue Beyond the Sale)

Geely demonstrated subscription services for advanced autonomous driving (L3+) and premium infotainment. To monetize SDV:

1. Offer a feature-as-a-service model (e.g., $9/month for remote park assist).
2. Create an in-car app store with 30/70 revenue split with developers.
3. License your software platform to other OEMs (as Geely does with SEA to Polestar, Lotus, etc.).

Common Mistakes

Mistake 1: Skipping the Middleware Layer

Many teams jump directly from hardware to applications, leading to brittle code. Always implement a vehicle abstraction layer that shields app developers from hardware changes.

Mistake 2: Not Planning for Bandwidth

Geely’s demo required 1 Gbps Ethernet in the vehicle. Older CAN-based architectures cannot handle software-defined features—invest in 5G and automotive Ethernet early.

Mistake 3: Treating OTA as an Afterthought

OTA must be designed from day one. Common failures: insufficient storage for dual partitions, poor update failure recovery, and lack of delta compression.

Mistake 4: Ignoring Functional Safety in SDVs

When software updates modify safety-critical functions (e.g., braking), you must re-validate against ISO 26262. Many startups overlook this and face recalls.

Mistake 5: Underestimating Developer Ecosystem Costs

Building an open platform requires continuous SDK updates, bug bounties, and legal agreements. Budget for a dedicated platform team.

Summary

Geely’s forceful pivot to a software-defined vehicle at Auto China 2026 is more than a marketing stunt—it is a blue print for the industry. By abstracting hardware, opening software APIs, personalizing UX with data, hardening cybersecurity, and monetizing beyond the initial sale, automakers can survive the transition from mechanical to digital value. Begin with the five steps above, avoid the common pitfalls, and treat the vehicle as a platform that can improve every month, not every model year.