China EV AI Race Enters Deep Waters: Why Legacy Automakers Are Racing to Own the ‘Soul’ of Software

What happens when artificial intelligence analyzes 200 different vehicle infotainment systems and concludes they are 95% identical? For China’s electric vehicle industry, this isn’t a hypothetical—it’s the existential crisis now forcing legacy automakers like Great Wall Motors to bet everything on AI differentiation or face obsolescence.

The China EV AI race is entering its most treacherous phase yet. As Navigate on Autopilot (NOA) user activation rates accelerate beyond industry forecasts, intelligent driving is evolving from a premium option into the defining ‘soul’ of the vehicle—a shift that threatens to expose which automakers possess genuine software capabilities versus those merely assembling hardware.

The Homogeneity Trap: When 95% of Dashboards Look Alike

According to She Shidong, Vice President of Intelligent Products at Great Wall Motors (GWM), the industry has hit a ‘painful situation’ where innovation feels like ‘cooking without rice.’ Speaking at the 2026 Intelligent Electric Vehicle Development Forum, She revealed that when GWM fed 200 different cockpit interface designs into a large language model, the AI determined similarity rates exceeded 95%.

This convergence spans from Huawei’s HarmonyOS Cockpit to BYD’s DiLink and XPeng’s Xmart OS. Whether it’s 3D vehicle models, wallpaper desktops, or fixed Dock bars, the underlying logic has become interchangeable. When competition devolves into ‘whose screen is larger’ or ‘who has more apps,’ the hardware differentiation that once drove margins collapses.

The Three Phases of Automotive AI Evolution

She identifies a critical inflection point where large AI models transition from ‘rear-mounted tools’ to ‘system kernels.’ This evolution occurs in three distinct stages:

• 2022-2023: The Generative Add-On – Early implementations treated AI as a content generation layer for chat responses, wallpaper creation, and route planning—essentially a ‘rear-mounted’ feature.
• 2024-2025: The Voice Agent – Systems developed contextual understanding and memory capabilities, creating persistent ‘voice agents’ capable of multi-turn conversations.
• 2026+: The Conversational Native Entry – By late 2025 and into 2026, the technology shifts toward becoming the primary interface architecture. As Bloomberg reported regarding Tesla’s Grok deployment in North America, the race is now for ‘conversational native’ systems where AI replaces the app-based paradigm entirely.

Why ‘Independent System Capability’ Is the Real Competitive Moat

The brutal truth emerging from the China EV AI race is that the decisive factor isn’t raw computing power, algorithmic sophistication, or parameter scale. Instead, survival depends on establishing independent intelligent system capabilities—a holistic integration that legacy automakers, rooted in mechanical engineering cultures, struggle to build.

Unlike internet companies that iterate software daily, traditional OEMs like GWM admit that large model integration for cockpits has ‘just crossed the starting line.’ The gap between automotive software cycles and internet-speed development creates a structural disadvantage that only vertical integration can solve.

The Edge Computing Bottleneck

Standing between traditional automakers and AI differentiation is a hardware constraint rarely discussed in boardrooms: edge computing limitations. While cloud-based AI offers flexibility, the latency and connectivity requirements of autonomous driving demand sophisticated onboard processing. Current vehicle-grade chips struggle to run large parameter models locally, forcing a dependency on network infrastructure that compromises the ‘intelligent driving as soul’ vision.

Reuters analysis suggests that by 2026, the market will bifurcate between AI-native automakers capable of proprietary system integration and hardware-assemblers forced into commodity pricing.

Investment Implications: Reading the 2026 Tea Leaves

For Western investors and automotive executives, the China EV AI race offers a preview of the global industry’s next consolidation phase. The warning signs are clear: when software homogeneity reaches 95%, brand loyalty evaporates.

Companies like Great Wall Motors are attempting to pivot, but as She Shidong notes, the transition requires rebuilding organizational DNA around AI-native development—a transformation most legacy players will fail to complete. See our analysis on how Huawei is reshaping global autonomous driving standards to understand the competitive landscape facing Western entrants.

The China EV AI race is no longer about who builds the best car. It is about who owns the conversation between driver and machine. As 2026 approaches, the ‘deep waters’ of intelligent driving will separate the souls from the shells—and the global automotive hierarchy may look radically different by the time the winner crosses the finish line.