OpenAI supports U.S. bills to restrict Chinese access to AI chips

OpenAI joins bipartisan push

OpenAI has officially joined a bipartisan congressional effort to strengthen United States limits on advanced AI chips. The company is supporting four specific legislative proposals aimed at restricting China's access to high-end computing hardware and the equipment used to make it. This move signals a significant alignment between one of the world's leading AI developers and U.S. lawmakers on national security concerns.

The primary goal of these measures is to ensure that advanced technology does not reach restricted entities. By backing these bills, OpenAI is participating in a broader strategy to manage the flow of semiconductors. The company's involvement highlights the growing intersection between private AI development and government oversight of critical infrastructure.

Chris Lehane, the Global Affairs Chief at OpenAI, stated that stronger controls would provide the U.S. and its allies more time to establish international coordination. He noted that these measures are necessary to create oversight systems and safety standards for AI. This suggests that the company views hardware limitations as a prerequisite for a more stable and regulated global AI environment.

Closing the remote rental loophole

One of the most specific proposals, the Remote Access Security Act, targets a common method used to bypass export rules. Currently, Chinese AI companies can use remote connections to access rented GPUs—graphics processing units used for heavy computing—without actually purchasing the physical chips. These chips are often housed in data centers located in countries like Singapore or Malaysia.

The proposed legislation would authorize the Department of Commerce to oversee these types of remote rentals. By monitoring how computing capacity is accessed from abroad, the government hopes to prevent Chinese firms from using overseas infrastructure as a proxy. This addresses a technical workaround where the physical location of the hardware does not match the location of the user.

If these rules are enacted, they could significantly impact customers of major chipmakers like NVIDIA. Companies that currently rely on overseas data centers to provide high-performance computing for their clients might face new restrictions. The policy aims to ensure that high-end computing power remains unavailable to restricted users, regardless of where the server is physically plugged in.

Tracking and equipment controls

Beyond remote access, the other three proposals address physical hardware and manufacturing. The Chip Security Act would mandate that manufacturers track advanced processors and alert the government if they suspect the chips are being diverted to restricted regions. This creates a chain of custody for high-end hardware to prevent it from ending up in the wrong hands through secondary markets.

The MATCH Act focuses on the manufacturing side by pressing allies like Japan and the Netherlands to restrict the sale and maintenance of semiconductor equipment. This means that the machines used to build the chips would be subject to tighter controls. By limiting the tools available to manufacture advanced components, the U.S. aims to slow the development of domestic chip production in China.

Finally, the AI Overwatch Act would expand the role of lawmakers in the export process. It would give Congress more authority to weigh in when export rules for advanced chips are being changed. Together, these four bills create a multi-layered approach that covers manufacturing, physical distribution, and remote digital access.

The debate over trade and security

The push for tighter controls reflects a deep internal debate over the long-term effects of trade restrictions. Some argue that these limits are essential because advanced chips can be used for military and intelligence purposes. They believe that preventing China from accessing these tools is a necessary step for national security and the protection of global AI safety.

Conversely, there is a concern that these restrictions could harm American business interests. Former NVIDIA CEO argued that continuing sales to China would protect American companies from losing their market share. There is a fear that if U.S. companies are blocked from selling to China, Chinese firms will simply turn to domestic suppliers, accelerating China's own technological independence.

The current proposals represent a shift toward prioritizing security over these commercial concerns. By backing these bills, OpenAI is siding with the view that the risks of advanced AI capabilities outweigh the potential for business growth in the Chinese market. The outcome of these bills will determine how the U.S. balances its economic goals with its security priorities.

OpenAI's hardware strategy

While OpenAI is currently a major customer of NVIDIA's GPUs, the company is also moving toward vertical integration. It is currently developing its own custom AI chip, which it has named Jalapeño. This project is intended for deployment within the United States and other allied countries, rather than for the general global market.

OpenAI has explicitly stated that it will not sell the Jalapeño chip to China. This commitment aligns with the goals of the new legislative proposals by ensuring that the company's own hardware innovations remain within a trusted sphere. It shows a proactive approach to hardware security where the company manages its own supply chain to meet government expectations.

The development of Jalapeño suggests that OpenAI wants more control over its computing infrastructure. By designing its own chips, the company can optimize them for its specific AI models while simultaneously adhering to strict export controls. This dual strategy of buying from NVIDIA while building its own hardware may define the company's future infrastructure roadmap.