Why the shift from hardware to software matters

For years, Chinese EV makers competed fiercely on price, with some smart EVs dipping below $10,000 . That strategy is losing its edge. As Hsiao notes, "It will be difficult for carmakers to differentiate themselves by introducing new EVs. To do so, they will need to upgrade their autonomous driving AI capabilities" .

The new battleground encompasses everything from in-car AI experiences to advanced driving assistance systems. A recent Wisers 2026 report found that 80% of top trending consumer keywords now relate to cutting-edge technology—including LiDAR, AI smart cabins, and 800V supercharging—rather than price .

Level 3 autonomy: the new baseline

The most immediate consequence of this AI arms race is the acceleration of Level 3 (L3) conditionally autonomous driving. While today's industry standard hovers at L2 or L2+, Hsiao expects L3 "might even soon become the norm" .

Under L3, the vehicle handles all driving tasks under specific conditions such as highway driving, though the human driver must remain ready to take over when requested. This represents a significant leap from current systems that require constant driver supervision.

Carmakers are already positioning beyond L3. BYD, for example, has developed a self-designed 4nm intelligent driving chip that supports both L3 and L4 autonomy—where the car can operate without human intervention in defined conditions. The company has committed over 100 billion yuan (approximately US$14.8 billion) in R&D toward what it calls the pursuit of "zero traffic accidents" .

The bigger autonomous picture

Morgan Stanley's broader analysis paints a transformative picture for global autonomous driving. The bank projects a $200 billion self-driving vehicle market by 2030, with China at its center. Nearly 30% of vehicles sold globally by 2030 are expected to feature Level 2+ smart driving capabilities—and China is forecast to account for 60% of those vehicles .

In more concrete terms, about 60% of passenger cars in China should be equipped with smart driving functions by 2030, representing sales of up to 15 million units, up from an estimated 6 million in 2025 . Hsiao adds that "China is likely to account for half of the global smart-driving market in volume by 2030" .

Robotaxis represent another frontier. Fully automated taxis are projected to make up 8% of China's total taxi and ride-sharing fleet within five years, totaling approximately 360,000 units . Meanwhile, competitors like Xpeng are rapidly closing the gap with Tesla's Full Self-Driving technology, with Morgan Stanley analysts expressing surprise at the pace of improvement during recent tests .

What this means for the industry

The transition from price competition to AI differentiation marks a maturation of China's EV sector. Where manufacturers once competed on who could offer the lowest price, they now compete on who can deliver the most reliable autonomous driving experience and the smartest in-car AI. The winners will be determined not by production scale or subsidy capture, but by software engineering, chip design, and real-world autonomous driving performance.

Why the shift from hardware to software matters

For years, Chinese EV makers competed fiercely on price, with some smart EVs dipping below $10,000 . That strategy is losing its edge. As Hsiao notes, "It will be difficult for carmakers to differentiate themselves by introducing new EVs. To do so, they will need to upgrade their autonomous driving AI capabilities" .

The new battleground encompasses everything from in-car AI experiences to advanced driving assistance systems. A recent Wisers 2026 report found that 80% of top trending consumer keywords now relate to cutting-edge technology—including LiDAR, AI smart cabins, and 800V supercharging—rather than price .

Level 3 autonomy: the new baseline

The most immediate consequence of this AI arms race is the acceleration of Level 3 (L3) conditionally autonomous driving. While today's industry standard hovers at L2 or L2+, Hsiao expects L3 "might even soon become the norm" .

Under L3, the vehicle handles all driving tasks under specific conditions such as highway driving, though the human driver must remain ready to take over when requested. This represents a significant leap from current systems that require constant driver supervision.

Carmakers are already positioning beyond L3. BYD, for example, has developed a self-designed 4nm intelligent driving chip that supports both L3 and L4 autonomy—where the car can operate without human intervention in defined conditions. The company has committed over 100 billion yuan (approximately US$14.8 billion) in R&D toward what it calls the pursuit of "zero traffic accidents" .

The bigger autonomous picture

Morgan Stanley's broader analysis paints a transformative picture for global autonomous driving. The bank projects a $200 billion self-driving vehicle market by 2030, with China at its center. Nearly 30% of vehicles sold globally by 2030 are expected to feature Level 2+ smart driving capabilities—and China is forecast to account for 60% of those vehicles .

In more concrete terms, about 60% of passenger cars in China should be equipped with smart driving functions by 2030, representing sales of up to 15 million units, up from an estimated 6 million in 2025 . Hsiao adds that "China is likely to account for half of the global smart-driving market in volume by 2030" .

Robotaxis represent another frontier. Fully automated taxis are projected to make up 8% of China's total taxi and ride-sharing fleet within five years, totaling approximately 360,000 units . Meanwhile, competitors like Xpeng are rapidly closing the gap with Tesla's Full Self-Driving technology, with Morgan Stanley analysts expressing surprise at the pace of improvement during recent tests .

What this means for the industry

The transition from price competition to AI differentiation marks a maturation of China's EV sector. Where manufacturers once competed on who could offer the lowest price, they now compete on who can deliver the most reliable autonomous driving experience and the smartest in-car AI. The winners will be determined not by production scale or subsidy capture, but by software engineering, chip design, and real-world autonomous driving performance.