BYD's In-House Chip Shakes Third-Party Silicon Suppliers

BYD has officially entered the custom silicon race, marking a pivotal shift in the automotive technology landscape. The Chinese EV giant announced the mass production of its first 4nm autonomous driving chip, directly challenging established third-party vendors.

This move validates earlier predictions about original equipment manufacturers (OEMs) taking control of their semiconductor supply chains. It signals a potential decline for companies like Horizon Robotics, which relied heavily on these same automakers for revenue.

Key Facts: The Shift to Vertical Integration

• BYD launched the Xuanji A3: This is China's first 4nm automotive-grade smart driving chip, now in mass production.
• High Performance Specs: The single chip delivers 700 TOPS of computing power, scaling to 2,100 TOPS when three are linked.
• Market Reaction: Horizon Robotics (09660.HK) shares dropped 7.03% the next day, hitting a 52-week low.
• Gartner Prediction Validated: A 2021 report predicted 50% of top 10 carmakers would design chips by 2025; this has come true.
• Revenue Dependency: Estimates suggest BYD accounted for half of Horizon Robotics' monthly revenue.
• Global Trend: Tesla, Nio, XPeng, and Li Auto are also developing or using proprietary silicon solutions.

Gartner’s Prediction Becomes Reality

The automotive industry is undergoing a fundamental structural change driven by necessity and ambition. In December 2021, Gartner released its Predicts 2022: Automotive and Smart Mobility report. The firm forecasted that chip shortages and the rise of electrification would push major manufacturers to design their own semiconductors.

At the time, this seemed like a distant possibility for many traditional automakers. However, the convergence of software-defined vehicles and hardware scarcity forced their hand. Today, that prediction has materialized faster than expected.

BYD's announcement on May 28, 2026, serves as the definitive proof point. During the 'Dare to Be' intelligent strategy launch, Wang Chuanfu revealed the Xuanji A3. This 4nm chip represents a significant technological leap for an automaker venturing into silicon design.

The capability to produce such advanced nodes internally allows BYD to optimize hardware specifically for its vehicle architecture. Unlike generic off-the-shelf solutions, custom chips can be tuned for energy efficiency and specific AI workloads relevant to autonomous driving.

This vertical integration reduces dependency on external suppliers. It also protects margins by keeping intellectual property and manufacturing costs within the company. For the broader market, it confirms that self-sufficiency is no longer optional for top-tier EV makers.

Impact on Horizon Robotics and Third-Party Vendors

The immediate fallout was visible in the stock markets. Horizon Robotics, a leading provider of intelligent driving solutions, saw its share price plummet. On May 29, the stock fell 7.03%, with intraday losses exceeding 10%.

This drop reflects investor anxiety over customer concentration risks. Horizon Robotics has long depended on key accounts like BYD for a substantial portion of its income. Market estimates indicate that BYD alone contributed to half of Horizon's monthly revenue.

When a client representing such a large slice of your business decides to bring operations in-house, the financial impact is severe. This scenario mirrors trends seen in other tech sectors where platform owners eventually compete with their ecosystem partners.

The situation is not unique to BYD. Other major players are following suit:

• Tesla: Has developed its own FSD chips for years, reducing reliance on NVIDIA.
• Nio: Invests heavily in custom silicon to differentiate its autonomous features.
• XPeng: Collaborates closely with chip designers but develops proprietary algorithms and hardware integrations.
• Li Auto: Similar to peers, seeks control over the full stack from sensor to processor.

These five companies represent a significant chunk of the global EV market. Their collective move toward in-house chips creates a shrinking total addressable market for pure-play third-party semiconductor firms.

Technical Breakdown: The Xuanji A3 Capabilities

Understanding the technical specifications of the Xuanji A3 highlights why BYD made this strategic pivot. The chip is built on a 4nm process node, which is currently among the most advanced manufacturing technologies available for automotive applications.

A single Xuanji A3 unit provides 700 TOPS (Tera Operations Per Second). This metric measures the computational throughput available for neural network processing, which is critical for real-time object detection and path planning.

For high-end models requiring redundant safety systems and more complex perception stacks, three chips can be paralleled. This configuration boosts the total computing power to 2,100 TOPS.

Comparison with Competitors

To put these numbers in context, consider the current market leaders. NVIDIA's Orin-X chip, widely used in Western and Chinese EVs, offers approximately 254 TOPS per unit. While multiple Orin units can be combined, the integration overhead is higher.

BYD's approach suggests a focus on efficiency and density. By designing the chip themselves, they can minimize latency between the processor and other vehicle systems. This tight coupling is difficult to achieve with generic third-party components.

Furthermore, the 4nm node implies better power efficiency. In electric vehicles, every watt saved on computation translates to extended range. This is a crucial selling point for consumers who prioritize battery life over raw benchmark scores.

Industry Context: The End of the Golden Age for Chip Sellers?

The rise of OEM-designed silicon marks the end of an era for third-party chip vendors. For years, companies like NVIDIA, Qualcomm, and Horizon Robotics thrived by providing standardized platforms. Automakers simply integrated these solutions into their cars.

However, as autonomous driving becomes the primary differentiator between brands, standardization is no longer sufficient. Car manufacturers want unique capabilities that cannot be replicated by competitors using the same off-the-shelf hardware.

This trend parallels the smartphone industry. Apple designs its A-series and M-series chips to ensure performance leadership. Samsung and Huawei have similarly invested in custom silicon to reduce reliance on Qualcomm.

Now, the automotive sector is catching up. The complexity of AI workloads requires specialized hardware accelerators. Generic GPUs are often inefficient for these specific tasks. Custom ASICs (Application-Specific Integrated Circuits) offer superior performance per watt.

For Western companies like NVIDIA, this presents a challenge. While they remain dominant in training infrastructure, their edge in inference hardware faces increasing competition from in-house designs. The market may fragment, with each major OEM running its own proprietary stack.

What This Means for Developers and Businesses

For software developers working in the automotive space, this fragmentation poses a significant hurdle. Previously, writing code for a standard platform like NVIDIA DRIVE meant compatibility across many vehicles. Now, developers must adapt to multiple proprietary architectures.

Businesses relying on third-party chip ecosystems must diversify. Dependence on a single vendor is risky when that vendor's largest customers become competitors. Strategic partnerships should focus on flexibility and interoperability.

Investors should monitor the cash flow of semiconductor firms exposed to auto clients. Revenue volatility will increase as OEMs transition to internal solutions. Companies that can offer complementary services, such as cloud connectivity or advanced tooling, may survive better than those selling pure silicon.

Looking Ahead: The Future of Auto Silicon

The timeline for this transition is accelerating. With BYD already mass-producing 4nm chips, other OEMs will feel pressure to follow suit. We expect to see announcements from European and American manufacturers within the next 12 to 24 months.

Supply chain dynamics will shift accordingly. Foundries like TSMC and Samsung will see increased demand from non-traditional chip designers. These automakers will need robust design houses to translate their requirements into silicon layouts.

Regulatory bodies may also get involved. Standardization of safety-critical software could be challenged by fragmented hardware. Ensuring consistent safety levels across different proprietary chips will require new testing protocols.

Ultimately, the car of the future will be defined by its brain. As that brain becomes proprietary, the competitive advantage shifts from mechanical engineering to computational intelligence. The winners will be those who master both domains simultaneously.

Gogo's Take

• 🔥 Why This Matters: This is not just a corporate strategy shift; it is a signal that the automotive industry is maturing into a tech-first sector. Vertical integration allows OEMs to capture value previously lost to suppliers. For consumers, this could mean faster innovation cycles and more efficient vehicles, but potentially less interoperability between brands.
• ⚠️ Limitations & Risks: Designing 4nm chips is incredibly expensive and risky. Not every automaker has the capital or talent pool to succeed. Failed projects could drain resources needed for core vehicle development. Additionally, supply chain bottlenecks at foundries could delay launches if capacity is prioritized for larger clients like Apple or NVIDIA.
• 💡 Actionable Advice: Investors should reduce exposure to third-party auto-chip vendors with high customer concentration. Developers should focus on middleware and abstraction layers that can run across heterogeneous hardware. Keep an eye on partnerships between OEMs and EDA (Electronic Design Automation) firms, as this is where the real bottleneck lies.