Chinese authorities have indicated an intent to approve purchases of Nvidia’s RTX PRO 5500 chips by domestic companies, including internet giant Alibaba and tech powerhouse ByteDance. This development, communicated by China’s Ministry of Industry and Information Technology, suggests a nuanced approach to semiconductor imports. While previous export controls have restricted the sale of Nvidia’s most advanced data-center AI accelerators to China, the RTX PRO 5500 series is designed for high-end professional computers rather than massive training clusters. Allowing these purchases could provide Nvidia with a continued foothold in one of the world’s largest technology markets, offering a revenue stream that may not have been possible under stricter interpretations of export regulations.
The practical significance of this approval lies in the specific hardware involved. The RTX PRO 5500 is a workstation-class GPU. By permitting its sale, Beijing may be aiming to support domestic enterprise and professional computing needs without fully reopening the door to the frontier-level AI chips that are central to national security concerns. For Nvidia, this represents a partial mitigation of the market loss experienced in China since export restrictions were tightened.
However, the challenge to Nvidia’s market leadership is no longer limited to geopolitical barriers. On the commercial front, the company faces a formidable rival in Amazon (AMZN), a long-standing customer and partner that is increasingly developing its own semiconductor solutions. While Nvidia designs the chips and TSMC manufactures them, Amazon has moved beyond merely renting these chips to designing its own alternatives to control costs and optimize performance for its cloud infrastructure, Amazon Web Services (AWS).
Amazon has developed its own line of custom silicon, including Graviton central processing units (CPUs) and Trainium AI accelerators. These chips are designed to compete directly with Nvidia’s offerings in specific workloads. According to Amazon’s recent financial disclosures, the annual revenue run rate for its custom chips has reached $25 billion, with both Graviton and Trainium delivering triple-digit growth. Amazon CEO Andy Jassy has noted that while AWS remains a premier destination for running Nvidia chips, the company is simultaneously building a stand-alone chip business capability. This dual strategy allows Amazon to offer customers a choice between Nvidia’s premium hardware and its own cost-optimized alternatives.
The competitive dynamic is shifting toward price-performance efficiency. Amazon claims that its Graviton CPUs offer at least 30% better price-performance than competing products. As the market expands to include “agentic AI” and broader inference tasks, the volume of chips required is expected to grow exponentially. In this environment, customers are increasingly looking for lower-priced options to manage the high costs of AI inference. Amazon’s focus on value positioning puts it in a strong position to capture a larger share of the market, potentially eroding the premium margins Nvidia has historically enjoyed.
Despite this growing competition, the overall demand for AI infrastructure remains robust. Nvidia is also preparing to launch its first stand-alone CPU this fall, aiming to enter the CPU market where customers are currently gravitating toward Amazon’s Graviton. This move highlights the intensifying battle for control over the entire AI stack, from the underlying processors to the top-level accelerators.
Underpinning this entire ecosystem is a critical dependency on ASML Holding N.V., the Dutch company that holds a monopoly on extreme ultraviolet (EUV) lithography machines. These machines are essential for TSMC to manufacture the advanced chips used in complex AI systems. ASML CEO Christophe Fouquet has warned that excessive restrictions on China could inadvertently accelerate the development of domestic Chinese chip technology, as nations seek to reduce reliance on Western-supplied equipment. This supply chain reality means that while Nvidia competes with Amazon for market share, both are ultimately dependent on the same limited manufacturing capabilities provided by ASML.
The next critical period for observing the impact of these shifts will be the release of Nvidia’s new stand-alone CPU this fall. The market reception of this product will serve as a key indicator of whether Nvidia can maintain its leadership position against both the rising tide of custom silicon from cloud giants and the regulatory uncertainties surrounding the Chinese market.
