The global AI race has reached a critical time. There, the interaction between geopolitical risk and long-term technical control defines strategic investment opportunities. At the heart of this dynamic is access to China's AI markets and the US government's evolving export policies, which creates tensions between national security orders and economic pragmatism. For investors, interests are clear. Understanding this intersection requires analyzing US policies, China's willingness to self-sufficiency, and its broader impact on AI leadership.
US policy tightrope walk: revenue sharing and strategic leverage
The Trump administration's recent decision to allow NVIDIA to resume sales of H20 AI chips to China, albeit a 15% revenue sharing agreement, highlights a readjustment of US export controls. Surrounded as part of broader trade negotiations, this policy shift reflects an act of balancing China's access to advanced technology with maintaining the economic interests of the US. By allowing mid-range chips to be sold while maintaining economic interests, the administration aims to avoid a complete retirement of China's $50 billion AI market relationship while maintaining leverage [1].
However, this approach is not without risk. Critics argue that a 15% fee could blur the boundaries between national security and political convenience and undermine the reliability of US export controls [6]. For Nvidia, uncertainty has already brought concrete consequences. CFO Collett Cress noted that regulatory ambiguity delayed H20 shipments in the second quarter of 2025, causing significant revenue for the company. [3]. The broader semiconductor industry is currently navigating a fragmented policy environment where AMD and other companies face similar revenue sharing arrangements [2].
China's Self-Sufficient Gambit: Innovation and Avoidance
China has two responses to US export restrictions. Avoid control through accelerated domestic chip production and alternatives. The nation plans to triple AI chip power by 2025 and plans to make state-backed investments in manufacturing plants and R&D for advanced processors such as Huawei's Ascend 910 series [1]. These chips are designed to rival performance on Nvidia's H100, but at low cost it shows a shift towards self-reliance. Meanwhile, Chinese companies are optimizing midrange chips like the H800 through software innovations such as FP8 data formats to maximize efficiency [6].
However, China's progress is not without hurdles. Limited access to EUV lithography machines and immature software ecosystems remains a critical bottleneck [2]. Additionally, US export controls have bypassed restrictions by spurring illegal activities such as network and shell companies smuggling. [1]. These tactics highlight China's adaptability, but also reveal supply chain vulnerabilities. The key question for investors is whether China's self-sufficiency drive will provide sustainable alternatives to US leadership or will it slow the inevitable?
Global Alliances and Fragmented AI Ecosystem
The US strategy goes beyond bilateral negotiations with China and is trying to coordinate with allies to shape the world's AI landscape. January 2025 AI Spreading Framework aims to categorize countries into three layers based on their trust level, and maintain control of the US and alliances while restricting access to hostile countries. [6]. Trusted partners like Japan and the Netherlands receive unlimited access to advanced AI chips, but tier 3 countries face strict security measures. However, this layered approach relies on harmonious enforcement of the allies. This is a challenge given the various legal frameworks and economic dependencies. [4].
Meanwhile, the Trump administration's AI Action Plan emphasizes promoting global adoption of US technology while protecting national security. By exporting a full AI stack (hardware, models, software), the US is trying to maintain the edge of computing power and standard [5]. However, the success of this strategy depends on the support of Congress and its ability to counter China's growing influence in regional markets.
Impact on investment: Navigate intersections
For investors, the Nvidia-china-us triangle presents a paradox. Short-term geopolitical risks can erode short-term benefits, but long-term AI domination can result in very large returns. Nvidia's recent pivot on the development of its recent B30 chips (products designed to meet both US and China's regulatory requirements) is eschewing the company's attempts to span this gap [2]. However, the possibility of higher revenue sharing fees on advanced chips like the Blackwell series suggests that the US government is testing its leverage limits. [3].
The broader semiconductor sector faces a similar dilemma. While US export controls have disrupted China's ecosystems, it has also accelerated innovation in domestic alternatives. This duality creates a fragmented global market. US companies need to balance compliance with the risk of losing market share to their Chinese competitors. [6]. For investors, what's important is to assess whether US policies can maintain their technological advantage without stifling the very industries they are aiming to protect.
Conclusion: delicate equilibrium
The US and China are trapped in high-stakes contests for AI hegemony, and Nvidia is at a strategic crossroads. Export control and revenue sharing agreements aim to maintain US leadership, but also run the risk of catalyzing China's self-sufficiency and fragmenting the global AI ecosystem. For investors, the path forward is to understand the interactions of policy, innovation and market dynamics. This is a challenge that requires both geopolitical and technical insights. In this evolving landscape, the ability to navigate uncertainty separates strategic benefits from costly miscalculation.
sauce:
[1] Among US export curbs, triple AI chip output from China by 2025 [https://mexicobusiness.news/cloudanddata/news/china-triple-ai-chip-output-2025-amid-us-export-curbs]
[2] Trump's political tax on nvidia chips to China – aaf [https://www.americanactionforum.org/insight/trumps-political-tax-on-nvidia-chips-to-china/]
[3] Nvidia says it's missing out on Chinese sales that's waiting… [https://www.cnn.com/2025/08/27/tech/nvidia-earnings-china-trump]
[4] Understanding the Artificial Intelligence Spreading Framework [https://www.rand.org/pubs/perspectives/PEA3776-1.html]
[5] AI leadership won't win by hiding the ball – AAF [https://www.americanactionforum.org/daily-dish/ai-leadership-isnt-won-by-hiding-the-ball/]
[6] Trump lifted the ban on AI chips in China and cleaned up nvidia… [https://builtin.com/articles/trump-lifts-ai-chip-ban-china-nvidia]
