The world of technology is ever-evolving, and few sectors are capturing attention quite like Artificial Intelligence (AI). As AI capabilities expand, the chips that power them – particularly GPUs (Graphics Processing Units) – have become central to both technological advancement and global competition. Recently, we explored the intricate dynamics between the United States and China in this critical area in our video “The AI Chip Race: Has China Already Pulled Ahead?” Let’s delve into the insights shared in that discussion, focusing on what this means as we navigate the financial markets.

Over the past several years, China has made substantial and rapid advancements across several burgeoning sectors, including AI, semiconductors, electric vehicles, and robotics. This growth is unprecedented in the nation, and the technological gap between the East and the West appears to be diminishing at an accelerated rate. From AI models like Deepseek and Alibaba’s Qwen to breakthroughs in semiconductor technology by Huawei and the surge in intelligent robotics, China is not just catching up but is establishing the technological pace in certain areas.

The Current Leaders and Market Dynamics

When we look at the current global AI chip market, our analysis indicates that Nvidia remains one of the leading GPU companies in terms of AI chips. Their products, such as the H100 and the latest Blackwell GPU, have been dominating the AI market, with many large companies purchasing these AI GPUs. AMD is identified as the second player in the GPU segment. Their MI300 series GPU is considered a good alternative to Nvidia’s H100 or Blackwell GPUs. For now, their leadership positions persist, and we don’t currently see new products threatening Nvidia’s worldwide position in the GPU market.

Despite geopolitical tensions and US export restrictions, the Chinese market remains significant. For Nvidia, the Chinese market accounts for about 13% of their total revenue for H20 sales in China. While this might seem small, it experienced a substantial 50% growth from 2024 to 2025, highlighting its importance to Nvidia’s growth trajectory. Nvidia’s CEO visited Beijing after new export controls were imposed on their H20, H200, and H800 AI GPUs designed for China, vowing to serve China with an alternative GPU – the B20, specifically created for the China market alone. This action underscores that Nvidia considers China a very important market. Similarly, AMD’s CEO also visited China, showcasing that AMD’s products can work with Chinese models like Deepseek and Alibaba’s Qwen, indicating that AMD also views China as an important market. In summary, despite geopolitical tensions, the demand from the Chinese market is still strong, and these companies remain very interested in it.

China’s Progress and Remaining Challenges

China has narrowed the divide despite imposed restrictions. Huawei’s Ascend series GPU is highlighted as a key competitor. Their latest series, the 910C, is said to be almost as good as the Nvidia H20, and some even claim it is better. However, it’s important to understand the context. The Nvidia H20 was designed specifically for the Chinese market as a trimmed-down version of higher-end GPUs restricted from import into China. While Huawei’s GPU may match the H20’s performance or even some older generation performance in China, countries outside China still have access to the latest Nvidia GPUs like the Blackwell. Globally, if one seeks the strongest computing power, they will likely still opt for the strongest available GPU. Compared to Blackwell, Huawei’s Ascend still has some distance to cover, estimated to be around 3 to 5 years behind due to differences in semiconductor technology and chip design. Therefore, our analysis suggests Huawei’s Ascend chip does not yet threaten Nvidia’s global sales.

Deepseek, one of China’s AI models mentioned as potentially closing the gap with US AI technology, is noted to be moving away from Nvidia’s CUDA software library. CUDA is an AI software library that enables developers to use Nvidia processors for faster AI processing and provides a strong ecosystem advantage for Nvidia. While Deepseek is reportedly using a lower-level method called PTX assembly (part of Nvidia’s tools), it was still trained by Nvidia GPUs (specifically older H800 models for training and H20 for inference). Using lower-level assembly can make it harder to move away from the specific hardware set, suggesting that for Deepseek to fully move away from Nvidia may take some time.

There are several prerequisites that China must fulfill before it can potentially surpass the United States in this industry. These include:

• Advanced Chip Manufacturing: The U.S., through TSMC and Intel, is already at the 3-nanometer node for mass production and researching towards 1.5-nanometers. In contrast, China’s main foundry (SMIC) is still at 7-nanometer for stable mass production, with 5-nanometer having a low yield. Huawei’s latest chip is around 6-nanometers, seen as an enhancement of the 7-nanometer technology. A key reason for the US advantage is access to ASML’s EUV (Extreme Ultraviolet) machines, critical for fabricating chips at 5-nanometer and below, which China has been restricted from accessing. China is investing heavily in developing its own fabrication machines, with Huawei investing in a company (SI carrier) that has a sub-5-nanometer machine prototype targeting launch around Q3 2025, although performance and mass production scaling remain unproven. Huawei has an ambitious plan – targeting 70% semiconductor value chain self-sufficiency by 2028.
• AI Computing Power: US companies have invested in hundreds of thousands or millions of high-end GPUs (like H100, Blackwell), resulting in very strong computing power. Export controls and limitations in domestic manufacturing have resulted in lower computing power for China. As highlighted in the discussion, using an analogy: even with an optimized model and a strong driver, a powerful engine is needed – without sufficient high-end chips, raw computing performance is limited.
• Investment in R&D: A Significant investment in advanced R&D is needed to design competitive chips. China’s government has reportedly introduced large funds for R&D in the semiconductor space since 2024 or earlier.
• Building its own Software Ecosystem: Nvidia’s CUDA software library is seen as a dominant force and a significant advantage in the AI GPU market, providing a strong moat. For China to compete globally, it needs to develop its own compelling software ecosystem.

In summary, while China is actively working on these areas, our analysis indicated that a significant gap remains, particularly in advanced chip manufacturing and the software ecosystem.

Implications for Investors

As investors, it’s natural to consider how these dynamics impact our portfolios. Li Chye, our market analyst, suggested that despite China’s progress, the US market remains the main focus. Nvidia and AMD are still in strong global positions because their GPUs are the fastest worldwide. Furthermore, countries like those in Europe, the US itself, and allies in Asia (like Japan or Korea) may still prefer US chips for national security or security reasons.

The AI revolution is closely linked to cloud computing, and US-based companies like Microsoft, Google, and Amazon are giants in cloud infrastructure and computing. For companies storing sensitive data, US-based cloud providers may be preferred due to security considerations. This also highlights potential opportunities in cyber security companies, many of which are based in the US.

While there is a risk perception based on market news about China catching up, based on the factors shared (global position, security preferences, technological gaps), the US market is still the primary area of focus.

However, for those concerned about solely focusing on the US market or wanting to navigate geopolitical factors, Li Chye suggested considering diversification. This could involve exploring the Hong Kong/China market if aware of strong companies there, or considering staple stocks or non-AI stocks in the US to create a more stable portfolio amidst the focus on AI.

In Conclusion

The AI chip race is a fascinating and crucial development with significant implications for technology and global markets. While China is making remarkable strides and narrowing the gap in key areas, the United States currently maintains technological superiority in critical aspects like advanced manufacturing and software ecosystems, as outlined in the video discussion. As investors, understanding this dynamic is key. The US market, particularly in AI hardware leaders, cloud computing giants, and potentially cyber security, remains a strong area of focus based on the analysis presented. However, prudent investors may also consider diversification strategies in light of the evolving global landscape.

Remember, staying informed and maintaining a diversified approach are cornerstones of navigating dynamic markets like the one shaped by the AI revolution.