What Nvidia’s B30 Chip Means for China-Based Manufacturers

Nvidia is developing a new AI chip called the B30, designed specifically for the Chinese market. This chip is being created to comply with updated U.S. export restrictions, which have blocked the sale of Nvidia’s more powerful processors, such as the H100 and even the H20, to China. By tuning the B30 to stay just under the U.S. regulatory performance thresholds, Nvidia aims to continue serving a market that made up approximately 14% of its revenue last year, states the South China Morning Post.

This development isn’t just about one chip, it reflects how companies are adapting to a regulatory environment where access to key technologies depends on government policy as much as market demand. For companies operating in China, the B30 could play a critical role in enabling local AI development, data center infrastructure, and advanced computing, without triggering compliance red flags from Washington.

In this blog, we’ll look at what the B30 chip is, the export controls it was designed to navigate, and why this matters for businesses developing or deploying AI systems in China.

The Genesis of the B30 Chip

Nvidia's development of the B30 chip is a strategic response to escalating U.S. export restrictions that have significantly impacted its ability to supply advanced AI processors to China. Previously, Nvidia attempted to navigate these restrictions with the H20 chip, a modified version of its high-performance H100, designed to comply with earlier export controls. However, in April 2025, the U.S. government imposed stricter regulations, requiring export licenses for the H20, effectively barring its sale to China and leading Nvidia to take a $5.5 billion charge due to unsold inventory and purchase commitments.

In response, Nvidia initiated the development of the B30 chip, tailored specifically for the Chinese market and engineered to meet the latest U.S. export control thresholds. The B30 is expected to be based on Nvidia's Blackwell architecture and will utilize conventional GDDR7 memory instead of the more advanced high-bandwidth memory (HBM), which is subject to export restrictions. This design choice not only ensures compliance but also simplifies manufacturing by avoiding the need for advanced packaging technologies like TSMC's Chip-on-Wafer-on-Substrate (CoWoS) states Reuters.

Despite its reduced specifications compared to the H20, the B30 is anticipated to support multi-GPU configurations, enabling the formation of high-performance computing clusters. This capability is particularly significant for Chinese tech giants such as ByteDance, Alibaba, and Tencent, who have expressed interest in the B30 to support their AI development needs (IFP).

Nvidia plans to produce over one million units of the B30 chip in 2025, underscoring its commitment to maintaining a presence in the Chinese market, which accounted for approximately 14% of its revenue last year . The development of the B30 exemplifies Nvidia's adaptive strategy in navigating the complex landscape of international trade regulations while striving to meet the demands of a critical market.

U.S. Export Controls: An Overview

U.S. export controls on advanced semiconductors have become a defining factor in the global tech landscape, particularly in the race for AI dominance. These regulations, administered by the U.S. Department of Commerce’s Bureau of Industry and Security (BIS), are aimed at limiting China’s access to high-performance computing chips that could be used for military or surveillance purposes.

The key mechanism behind these restrictions is the Export Administration Regulations (EAR), which control the transfer of U.S.-origin technology based on performance thresholds. For AI chips, these thresholds focus on two main parameters: computing performance, measured in FLOPS (floating point operations per second), and interconnect bandwidth, which determines how well multiple chips can be networked together.

In October 2023 and again in 2024, the U.S. tightened these rules, restricting not just top-tier chips like Nvidia’s A100 and H100, but also their downgraded variants such as the H800 and H20. The new rules apply to both direct sales and indirect shipments via third-party countries, significantly narrowing the options for semiconductor exports to China.

The regulations also target advanced packaging technologies, such as those involving high-bandwidth memory (HBM) and advanced substrates, making it difficult for U.S. companies to supply chips that can be easily clustered for large-scale AI model training.

To comply, companies like Nvidia have been forced to redesign chips specifically to fall under the performance caps which has resulted in new models like the B30, which balance reduced performance with continued market access. These export controls are not static; they continue to evolve, often without much notice, creating a challenging environment for companies trying to serve both the U.S. and Chinese markets.

The Role of the Nvidia B30 Chip in U.S.-China Tech Relations

The development of Nvidia’s B30 chip is more than a workaround, it’s a symbol of the ongoing tension and interdependence between the U.S. and China in the global technology race. On one hand, it demonstrates how U.S. export controls are reshaping the semiconductor landscape by limiting China’s access to cutting-edge chips. On the other, it highlights the extent to which American companies like Nvidia still rely on the Chinese market. 

By creating the B30 specifically for China, Nvidia is walking a fine line by staying within the boundaries of U.S. policy while maintaining a foothold in one of the world’s largest and fastest-growing markets for AI infrastructure. The chip’s reduced specifications help it meet export rules, but it still offers Chinese firms the ability to build multi-GPU computing clusters, enabling them to train and deploy AI models at scale.

This balancing act has broader geopolitical implications. The B30 could be seen as a concession to economic realities. It also signals that U.S. firms are finding ways to serve Chinese customers without crossing regulatory lines, potentially setting a precedent for others in the semiconductor space.

Ultimately, the B30 is a case study in the new era of tech diplomacy, where chip design is no longer just about performance, but it’s about policy, compliance, and global strategy.

Conclusion

The introduction of Nvidia’s B30 chip underscores the increasingly complex intersection of technology, policy, and global business strategy. By designing a chip that complies with U.S. export restrictions while still serving the needs of Chinese customers, Nvidia is not only protecting a key revenue stream, but it’s setting a model for how multinational tech firms may navigate this new regulatory landscape.

For companies operating in China, particularly those building AI infrastructure or relying on high-performance GPUs, the B30 offers a lifeline by providing access to capable hardware without triggering export violations. But it also serves as a reminder that the rules are shifting quickly, and long-term strategies must account for regulatory risk as much as technical capability.

As U.S.-China tech relations continue to evolve, manufacturers, developers, and policymakers alike will be watching how chips like the B30 perform in the market and whether they mark the beginning of a new equilibrium or just a temporary fix in a long-term standoff.