Ahead of Trump–Xi Summit, former US national security advisor warns on rare earth, AI rivalry, and China’s potential “theft”

Taiwan Current News (TCN), in partnership with the German Marshall Fund of the United States (GMF), has launched a new podcast series, Taiwan Frontlines, to track global trends and explore how Taiwan connects with the world.

In the latest episode, former US National Security Advisor Jake Sullivan sat down with Bonnie Glaser and Jason Hsu to discuss US–China strategic competition spanning rare earth supply chains, export controls, and the escalating race in artificial intelligence (AI).

He offered a candid assessment of missed opportunities, calibrated statecraft, and a rapidly evolving technological contest that could redefine global power as Washington and Beijing edge toward a pivotal summit.

A choke point the US saw — but moved too slowly on

Sullivan confirmed that Washington had long identified China’s dominance in rare earths as a critical vulnerability — but failed to act with sufficient urgency.

“We did look at rare earths,” he said, noting efforts to counterbalance China’s grip through policy tools such as the Defense Production Act and targeted investments, including support for domestic firms like MP Materials.

Yet the verdict, in hindsight, was clear: the US responses should have been “faster and more ambitious with greater alacrity,” Sullivan said.

He invoked the pandemic-era “Operation Warp Speed” as a model for what he said should have been a similarly aggressive industrial mobilization.

Despite incremental progress under successive administrations, Sullivan suggested that neither the Biden nor Trump teams have fully matched the urgency required to restructure such a strategically vital supply chain.

“Small yard, high fence” and the art of managing retaliation

Sullivan offered a rare insider perspective on the logic behind US export controls, particularly the “small yard, high fence” doctrine designed to limit China’s access to advanced semiconductor technologies.

Central to that strategy was not only restriction, but restraint, he revealed.

By signaling that controls would remain targeted rather than expansive, Washington sought to shape Beijing’s response. According to Sullivan, that approach worked — at least initially.

China refrained from deploying its most potent countermeasures, opting instead for limited restrictions on materials such as gallium and germanium. These were disruptions the US could manage, Sullivan said.

In contrast, he criticized the Trump administration’s sweeping tariff escalation, calling it a “failure of statecraft.”

When the US fired the bazooka, China felt the need to fire the bazooka back, Sullivan said, describing the resulting tit-for-tat response something the US couldn’t manage.

The lesson, he argued, lies in balance: effective competition requires a calibrated mix of pressure and restraint to avoid uncontrollable escalation.

The real objective of export controls: computing supremacy

As export controls evolve beyond chips to encompass the broader semiconductor ecosystem, Sullivan pushed back against the notion that US strategy should focus on embedding global dependence on its “tech stack.”

Such arguments, he warned, misunderstand the stakes. Even if US-made chips underpin global systems, the decisive layer is the AI model itself — who builds it, trains it, and shapes its outputs, Sullivan said.

“The world could be built on Chinese models on top of US chips,” he cautioned.

Instead, the core objective has been to preserve America’s advantage in compute — the foundational resource powering advanced AI.

By restricting China’s access to cutting-edge chips and manufacturing tools, Washington has, for now, secured a significant lead. Whether that advantage proves durable, however, remains uncertain. “Whether that translates into a durable lead on AI or not, we will have to see.”

AI on the battlefield: defense and intelligence

Sullivan outlined a dual-front competition in artificial intelligence: military integration and global diffusion.

On the security front, he stated that late-stage policy efforts within the US government acknowledged a critical gap — not in AI innovation, but in its adoption across defense and intelligence systems.

“We got to move. We got to do more.”

Sullivan pointed out that China, by contrast, is rapidly embedding advanced AI across its military apparatus, from surveillance and targeting to autonomous weapons and command structures.

“This runs from logistics all the way to the tip of the spear,” he said.

For Taiwan’s defense capabilities, he added, the implications are critical: strategic investments must align with this shifting technological paradigm.

Diffusion, distillation — and a new kind of risk

Beyond military applications, Sullivan highlighted a subtler battleground: the global spread of AI models.

China may be making a tactical bet that it can't get ahead at the frontier, Sullivan said. “So the best American closed model is going to be ahead of the best Chinese model.”

China, he suggested, is in turn leveraging open-source strategies to compensate for lagging at the technological frontier, enabling broader global adoption of its systems.

At the same time, a more disquieting development is emerging.

Citing recent disclosures, Sullivan pointed to the possibility that Chinese firms are using “distillation” techniques to replicate advanced US models at a fraction of the cost — effectively learning from them.

“If that’s the case, it changes the game,” he said, framing the issue as a potential form of “theft,” and intellectual appropriation that existing trade frameworks are ill-equipped to address.

As Washington and Beijing prepare for high-level summit, Sullivan stated that this unresolved tension — spanning technology, security, and economic norms — may soon take center stage.

For more in-depth coverage, tune in to Taiwan Frontlines on the NOWNEWS official YouTube channel.