AI is steering the wheel toward a new era of autonomous vehicles (AVs) as the global automotive industry pivots towards Connected, Autonomous, Shared, and Electrified (CASE) vehicles, and with a heightened focus on achieving net-zero carbon emissions. Reflecting this momentum, private investment in automotive AI surged by USD 2.66 billion between 2017 and 2023.^[1] Yet, this rapid growth has also placed autonomous vehicle (AV) development at the heart of an intensifying U.S.-China technology rivalry—one shaped by advanced AI capabilities, high-performance semiconductors, and massive data requirements. As a result, the AV sector has become a new strategic battleground, with stakeholders on both sides racing to secure leadership in the next era of mobility innovation. 

Accelerating Demand for AI Chips in Autonomous Vehicles

The significance of AI chips in autonomous vehicle development cannot be overstated. The operation of autonomous vehicles hinges on robust computing power, with AI chips serving as the cornerstone of this capability. Key functions such as Advanced Driver Assistance Systems (ADAS), in-vehicle entertainment, environmental sensing, and driver monitoring depend on AI chips to provide the necessary computational power.

As the industry shifts from traditional distributed systems to zonal architectures, which demand greater functional integration and higher computing capabilities, centralized computing platforms equipped with high-performance AI chips become indispensable. These platforms are vital for smart vehicle control, autonomous driving technologies, and smart cockpits, all of which require AI chips with substantial computing power to manage the requisite calculations and integration.

With its conservative stance on security, the automotive industry has traditionally partnered with established suppliers holding robust security credentials and recognized industry certifications. Consequently, many automakers have relied on AI chips from U.S. technology powerhouses such as NVIDIA, AMD, and Intel (Mobileye). However, the global chip shortage of 2020 to 2023 underscored the need to diversify and de-risk supply chains—prompting a reassessment of longstanding global partnerships in search of more resilient and adaptable sourcing strategies.

Facing a Computing Power Shortfall: The Rise of In-House AI Chips

Beyond their reliance on standard GPUs, both Tesla and Chinese automakers are accelerating investments in custom AI chips to strengthen their competitive edge. These specialized processors provide more than just high-performance training capabilities; they address the escalating demand for substantial computing power as autonomous driving technologies advance, ensuring the capacity to handle increasingly complex algorithms and data-intensive workloads.

Tesla pioneered this approach to enhance adaptability, becoming the first electric vehicle maker to design its AI chips in-house. Alongside its automotive systems, Tesla has embraced the development of AI chip technology, spanning from cloud training chips to ground-level inference chips, thus controlling the foundational capabilities across the entire ecosystem, from upstream to downstream. The company focuses on AI software and computing chip technology breakthroughs while vertically integrating the whole electric vehicle supply chain to incorporate its products and software services.

From the geopolitical perspective, as competition in AI intensifies between China and the United States, Chinese automakers have identified a critical bottleneck in computing power. At the Global Intelligent Vehicle Summit 2024 (GIV2024) hosted by China EV 100, industry data revealed that Chinese automakers possess less than 10 EFLOPS of total computing capacity. Meanwhile, public computing infrastructure providers—China Mobile, China Telecom, and China Unicom—are on course to achieve 17 EFLOPS, 21 EFLOPS, and 15 EFLOPS, respectively, by the end of 2024, totaling 53 EFLOPS. In contrast, Tesla alone is expected to reach 100 EFLOPS in the fourth quarter of 2024, highlighting a significant performance gap that Chinese automakers must address to remain competitive.[2]

Domestic Support Accelerates Chinese Automakers’ AI Chip Investments

China is actively scaling its advanced semiconductor, software, and AI capabilities, bolstered by the 14th Five-Year Plan and complementary industrial policies aimed at next-generation technologies. To reduce regulatory uncertainty, policymakers have paved the way for Level 3 (L3) and Level 4 (L4) autonomy through regulations permitting on-road testing. Notably, the 2024 release of the Intelligent and Connected Vehicle—General Technical Requirements for Automated Driving Systems (GB/T 44721-2024) established China’s first national standard for autonomous driving systems, defining protocols for dynamic driving tasks, backup mechanisms, and human-machine interaction at L3 and above.

Given the heightened computing needs for L3 autonomous functionality, automakers have accelerated the evolution of in-vehicle computing platforms, with AI chips at the core of this competitive landscape.

In July of 2024, NIO introduced the Shenji NX9031, the world’s first 5-nanometer automotive AI chip. Soon after, XPeng unveiled its Turing AI Chip, geared not only for vehicles but also for robotics and flying cars. Meanwhile, as Li Auto, BYD, and Xiaomi continue collaborating with NVIDIA, BYD is notably launching its 4nm BYD 9000 chip for smart cockpits. Li Auto—while maintaining a partnership with Chinese AI chip manufacturer Horizon Robotics, exemplified in its L9 Pro featuring the Horizon Journey 5 solution—is also reportedly developing an in-house SoC chip, “Shu Ma Ke,” for autonomous driving applications. Collectively, these moves underscore how Chinese automakers are accelerating their pursuit of self-designed AI chips.

At CES 2025, NVIDIA introduced “Thor,” its next-generation automotive chip, announcing new partnerships with Tesla, BYD, Mercedes-Benz, Li Auto, Toyota, Jaguar, and Zeekr. Conspicuously absent were XPeng and NIO, signaling that, amid an escalating US-China tech standoff, Chinese automakers are coalescing around two strategic pathways for AI chip development. On one side, Li Auto and BYD continue to partner with NVIDIA while simultaneously designing their own chips—a dual-track approach that pairs the reliability of an established supplier with in-house innovation. On the other side, XPeng and NIO are leaning more decisively toward proprietary solutions, positioning themselves to secure a larger share of the domestic market if heightened geopolitical tensions restrict access to foreign technologies. Both strategies underscore the pressing need for robust and scalable AI chip capabilities to maintain leadership in autonomous vehicle technologies.

There Are No Eternal Friends—Only Eternal and Perpetual Interests

For over a century, the automotive sector has been a key driver of economic growth, technological innovation, and employment in major economies such as the United States, the European Union, Japan, and the United Kingdom. However, recent developments—most notably EV tariffs and stricter U.S. AI export controls—underscore the risks of relying too heavily on a single technological source. In December 2024, four Chinese industry associations representing the Internet, automotive, semiconductor, and telecommunications sectors declared that U.S.-made chips are no longer secure, urging domestic firms to adopt locally produced alternatives. Chinese automakers, already facing supply chain vulnerabilities and rising geopolitical tensions, have responded by designing their own AI chips and embracing a “de-Americanization” strategy.

In both business and international relations, no partnership is guaranteed to last indefinitely. The United States has introduced a three-tier AI export framework that effectively divides the European Union into varying levels of “like-minded partners,” potentially constraining certain countries’ ambitions for AI sovereignty.[3] As economic security concerns come to the fore, the EU faces a stark reminder that longstanding alliances can be redefined by shifting policy priorities and trade strategies.

The Path Forward: Taiwan's Role

With economic security concerns driving Tesla and Chinese automakers to design their own AI chips, European automakers have been relatively cautious—hindered by legacy structures and a slower pace of adopting next-generation AI solutions. In this context, Taiwan's ICT and semiconductor ecosystem offer a compelling partnership opportunity for global automakers. TSMC’s recent investment in Germany underscores the deepening ties between European automakers and Taiwan’s semiconductor sector. Taiwanese fabless firms have already played a critical role in Tesla’s AI chip development.[4]

Moreover, Taiwanese companies now offer more than just robust chip manufacturing; they are moving up the value chain with advanced sub-system solutions that streamline vehicle development. Recent initiatives underscore this evolution: AUO’s acquisition of the German firm BHTC in 2024 expanded its smart cockpit portfolio, Pegatron’s collaboration with NXP propelled software-defined vehicle (SDV) capabilities, and Clientron introduced its own proprietary smart control system. By leveraging these strengths, Taiwan stands out as a strategic partner for automakers seeking to diversify their supply chains and bolster technological resiliency.