Greater than the sum of its parts: Inside the US effort to block Chinese AI hardware
Northeastern Global News spoke to Ryan Ellis, an infrastructure politics and cybersecurity expert, about the U.S. effort to check China’s AI industry.
The United States is doubling down on efforts to block Chinese AI components and tech — from recent rules forbidding the import of Chinese humanoid robots, to bans on solar inverters and consumer routers produced by Beijing.
The Federal Communications Commission (FCC) rules are part of a broader effort to keep Chinese technology from becoming deeply embedded in critical U.S. AI infrastructure and supply chains, and safeguarding American domination in this field. The agency has warned that connected power inverters, which manage the flow of electricity between solar panels, batteries and the electric grid, could be used to collect and enable espionage or disruption.
The FCC has said that advanced robots could be exploited to “surveil Americans, enhance the capabilities of foreign intelligence services or to remotely commandeer the robots.”
The issue over U.S. efforts to curb Chinese technological influence surfaced again this week amid reports that the agency is weighing a ban on imports for devices in Chinese data centers.
The national security and cybersecurity threats posed by Chinese parts export markets are real and pressing, said Ryan Ellis, an associate professor at Northeastern University whose research focuses on communication law and policy, infrastructure politics and cybersecurity. But those concerns are intertwined with broader questions of economic competitiveness, industrial policy and the technological rivalry between the United States and China.
Northeastern Global News spoke to Ellis about the U.S. effort to check China’s AI industry. His comments have been edited for brevity and clarity.
The FCC has framed this as a national security issue. Is that the right way to think about what’s happening here, or is there a bigger picture?
Ellis: My response to that, or my thinking about it, would be that when we’re talking about, let’s say, computing infrastructure, particularly AI infrastructure, it’s impossible to disentangle all these different issues. You can’t really disentangle questions of economic competitiveness, economic rivalry, national security concerns and geostrategic concerns. Generally speaking, you can’t really separate them out.
So, is it true? Are there reasonable security concerns around hardware and software components inside AI data centers? Certainly there are. Are there reasonable concerns about different AI models and how they operate? Absolutely. We see these concerns around different AI models: which ones are suitable to be released, how we [the United States] manage open-source models, particularly those coming from China that appear capable of competing with some of our own very significant, complex and sophisticated models.
But we also see debates about how to manage the export and import of components that are vital to the computing power necessary to train and run these models. In the United States, we’re seeing a concerted effort to regulate the exporting of chips and other components that might be useful in advancing computing infrastructure and AI development abroad, particularly in China.
Why has national security become the vehicle through which policymakers seem to be approaching AI infrastructure?
Editor’s Picks #
Ellis: The language of national security is a really powerful rhetorical tool, and invoking concerns around national security is often quite compelling. At the same time, national security enables broad maneuverability for the government. There are statutes that enable officials to take certain steps they couldn’t take if the issue were simply economic competitiveness.
So it’s not only a compelling narrative. It also has certain legal justifications that are persuasive and powerful. It’s not surprising we’re seeing this conversation, and I think the concerns are legitimate, but they’re not the only set of concerns animating this ongoing debate and action.
This is a big-picture question, but all of this is taking place in the context of the broader conversation about AI supremacy between the United States and China — and presumably other players. From where you sit, are these sorts of rules and regulations necessary to ensure that U.S. industry is producing a superior product? Are they necessary to the protectionist approach the the White House is taking?
Ellis: For one, this isn’t particularly novel. The last two administrations — the Biden administration and the Trump administration — have both been very concerned about American dominance in the age of AI.
The question of protectionist measures — or maybe a better way to put it is supportive measures, ways of boosting the industry, developing domestic capacity to manufacture chips, making sure our scientists have the resources they need — those are all part of the conversation. But there’s no single intervention that’s going to determine a winner or loser in this AI arms race.
We may see restrictions on exports, or restrictions on imports. Those things all matter, but no one of them is a silver bullet. You would also want to see robust investment in public scholarship and research, investment in universities, investment not just in the capacity to manufacture, but in the capacity to innovate. It really requires a holistic approach, one that moves beyond simply thinking about restrictions and also considers supporting and buttressing the world-class scientists and researchers we have here.