The Trump administration has oscillated significantly in recent weeks on its policy toward two critical sources of China tech risk: chips and TikTok.
It seemed to oppose Nvidia selling highly capable H20 chips to China, including for national security reasons, before it backtrackednow accelerating their sale in exchange for a cut of the profits. Now, chip sales to Saudi Arabia may grow despite China’s considerable activity in the country. Trump’s first administration also hammered relentlessly to ban TikTok on national security grounds, watched many congressional allies pass a bipartisan bill to effectuate a ban, and then spent months ignoring the Supreme Court-upheld law altogether. Now, a deal will potentially leave China with influence over TikTok’s algorithm.
The Trump administration has oscillated significantly in recent weeks on its policy toward two critical sources of China tech risk: chips and TikTok.
It seemed to oppose Nvidia selling highly capable H20 chips to China, including for national security reasons, before it backtrackednow accelerating their sale in exchange for a cut of the profits. Now, chip sales to Saudi Arabia may grow despite China’s considerable activity in the country. Trump’s first administration also hammered relentlessly to ban TikTok on national security grounds, watched many congressional allies pass a bipartisan bill to effectuate a ban, and then spent months ignoring the Supreme Court-upheld law altogether. Now, a deal will potentially leave China with influence over TikTok’s algorithm.
Flip-flopping on U.S. policy toward Chinese tech is a terrible move. It creates long-term strategic vulnerability for the United States, undermining U.S. credibility on the security risks of Chinese technology. This will come back to bite Washington years down the line, with detrimental ramifications for foreign policy, intelligence relationships, and supply chain cybersecurity. The solution, right now, is to reimpose restrictions on H20 sales to China and tighten sales to nations that could pass chips to China, demand more from China in a TikTok deal, and make clear to allies and adversaries alike that in trade negotiations, core U.S. national security protections are not on the table.
Over the last century but especially the last two decades, the U.S. government has increasingly blurred its rhetoric about economic security and national security. Indeed, these two issues are quite entangled. A robust economy enables the funding of an advanced (and quite expensive) military; critical tech supply chains, owned and operated by the private sector, can become targets for nation-state cyber-espionage that harms those companies’ competitiveness and U.S. national security at once. The list goes on.
But there’s a pivotal lesson from the first Trump administration that policymakers seem to have conveniently forgotten. When national security moves look like self-interested trade policy, they derail the end goal of tackling Chinese tech threats. This happened during the 2018 U.S. campaign against Chinese telecommunication firm Huawei. The White House claimed that Huawei was a security risk while also suggesting it would toss those concerns aside for a trade deal. Far from being seen as a smart negotiating tactic, such willingness to discard a supposedly serious security issue caused even the United Kingdom, a close U.S. ally and member of the Five Eyes intelligence alliance, to distrust its U.S. counterparts—believing instead that Washington was hyping up “security” claims to land itself a better economic deal. It took years for London to get on board with a Huawei 5G ban.
Today, we’re seeing this failure repeated at enormous scale. In January, the Trump administration was reportedly discussing a move that its predecessor agreed with, too: cracking down on sales of Nvidia’s H20 chips (designed for tasks such as AI inference) to China because of the national security costs to the United States. The chips could be used for advancing Chinese AI models, the logic went, and enabling Chinese military and intelligence actors to develop capabilities that could compete or even achieve parity with U.S. ones. Restrictions, in this way, were more than common sense: Beijing has visibly demonstrated its ability to coerce and control the Chinese private sector, and Chinese entities were already smuggling in AI chips as it was. Stronger rules were needed to curtail the flow of sensitive technology to a U.S. adversary.
The Trump administration abandoning that supposedly strong position on China after a meeting with Nvidia signaled that such national security imperatives are, in fact, up for negotiation. Officials had reportedly spent two months crafting stronger chip controls to bolster the U.S. position vis-à-vis competition with China—and limit the advancement of Chinese military and intelligence AI capabilities.
Instead, after U.S. backtracking, companies around the world saw that U.S. security restrictions were malleable. Allies and partners witnessed how quickly security could be compromised for supposed trade wins. And adversaries watched an important export control debate get sidelined. Now, the White House’s November position that Nvidia cannot sell B30A chips to China is far less credible—and everyone from Silicon Valley to London to Beijing is likely waiting for the other shoe to drop.
The United States has similarly flip-flopped on TikTok with harmful effects. After gunning for the ByteDance-owned app in the first Trump administration, senior officials have now decided that TikTok, evidently, does not pose as great a risk as originally claimed. What the first administration would have deemed unthinkable is now considered acceptable in the reported deal: ByteDance and Chinese investors will retain around a 20 percent stake in the company; oversight of the algorithm remains unclear; and few details have been provided about protections for data security, which are unlikely to be robust with insufficient U.S. oversight of the matter. Once again, U.S. credibility suffers as allies and partners in Europe, for instance, continue to lag the United States in believing there are any national security risks from TikTok in their countries.
In the long run, these reversals in policy will come back to bite the United States. The next time the U.S. government identifies a serious national security risk from Chinese technology and brings it to allies and partners—whether labeling DeepSeek a national security threat or calling for multilateral bans on a Chinese-connected vehicle—Washington is more likely than ever to get skeptical looks back. Skeptical looks because allies and partners may not believe that U.S. officials genuinely care about the national security issue. And because allies and partners will assume, perhaps correctly so, that even if the officials in the room are serious, leaders at the top of the political system will suddenly undermine them to get a trade deal or a supposed “win” economically. And also because adversaries will see this flip-flopping, too, and doubt the U.S. resolve to follow through on its tech security commitments, even beyond China.
Tackling security risks in a global, digitally interconnected environment without those allies and partners is a losing game. The United States cannot effectively restrict chip sales, app ownership structures, or technologies from Chinese vehicles to routers if non-U.S. governments, companies, and organizations simply sell their own, comparable versions to China or use those same technologies in their own supply chains—which connect back to the United States. Effective controls depend on multilateral coalitions, on allies and partners’ genuine buy-in. But credibility is the currency that enables robust U.S. mitigation of Chinese tech risks—and flip-flopping burns credibility every day.
