Stronger Semiconductor Export Controls on China Will Likely Harm Allied Semiconductor Competitiveness
It’s finally clear to most in Washington that the United States faces a major competitor in China in defense, advanced technology industries, and other realms. There also appears to be a growing bipartisan consensus that the United States needs to limit China’s advancement. But that’s where the consensus breaks down. How to limit China’s relative advancement over the United States is a matter of controversy.
Many, including the Biden administration, have landed on export controls on semiconductors and advanced chip-making equipment as a critical weapon to limit China’s advancement, at least militarily. Toward that end, the administration released sweeping export controls designed to limit China’s capabilities in producing advanced semiconductors last October. And they are expected shortly to release updated regulations intended, in part, to tighten “holes” in the last round of rules. China hawks of both parties widely support these restrictions, and the administration is supposedly considering even tighter controls, in part after Huawei recently announced the production of an advanced 7nm chip for its new cell phone using legacy equipment.
It would be one thing if these controls could hamstring China’s production of chips. In that case, they would do less harm to allied-based semiconductor companies because other foreign companies would produce products with advanced chips. But the reality is that these export controls, particularly on semiconductors (rather than on truly choke-point semiconductor manufacturing equipment), will hurt U.S. and allied semiconductor firms. Not only do U.S. export controls limit U.S. chip sales (that’s why they are called export controls), but they will likely be met with a retaliatory response by the Chinese government that will hurt U.S. technology interests. On top of that, they accelerate the time by which China becomes self-sufficient, at least in a wide array of semiconductors. It’s time for the Biden administration to take a strategic pause and ensure that future export controls effectively hamper China but not damage Western firms.
Whether for political reasons or because they honestly believe it, the administration portrays semiconductor export controls not as commercial protectionism but as keeping China from gaining military advantage. Some semiconductors enable this advantage, but China’s military weapons are already sophisticated, and many of them use chips widely available in China. If it were clear that semiconductor export controls would really hamper Chinese military expansion, that would be one thing, but the administration has provided little evidence for this view. As the Rand Corporation wrote last year, the reality is that most weapons systems do not use the kinds of chips the administration seeks to control. The most advanced chips are for advanced computing, not for weapons systems (though there are certainly exceptions to this point, such as chips used in high-performance computers that perform roles such as modeling the impacts of nuclear detonations).
Moreover, limiting sales of semiconductors to China risks significantly reducing allied semiconductor competitiveness. One key reason is that, like many advanced technology industries, semiconductors have high fixed costs relative to marginal costs. In other words, it costs billions to design and make the first chip but far less to make the millionth chip. This is at the heart of the issue of export controls to China. By cutting off U.S. sales, U.S. firms will lose one of the most profitable parts of their market, significantly reducing profits that would otherwise be invested in R&D for the next generation of chips. For example, export controls forced Huawei to produce its own relatively advanced 7nm chip, which at least one study estimates will result in lost sales for Qualcomm of $11.8 billion in 2024. At the same time, losing sales means losing critical feedback from customers about desired product changes. This feedback is central to continued product improvement in advanced products like semiconductor equipment. Taking away Chinese sales reduces this feedback for U.S. firms while increasing it for Chinese (or other foreign) competitors.
None of this would matter much if China were Russia. When the U.S. banned chip exports to Russia, it meant giving up less than 1 percent of the global market, which Chinese firms soon filled. But China is different. It accounts for around one-third of all semiconductor sales. Tell Intel, Micron, Nvidia, Qualcomm, or other major U.S. semiconductor companies that their sales will shrink by one-third and see the result. Ultimately, it can lead to a death spiral: less revenue, less money invested into next-generation semiconductors, less sales of the next-generation chips, leading to another reduction of sales, and so forth. It is certainly contradictory that at a time when the official policy of the U.S. government is to support the U.S. semiconductor industry (through, among other legislative vehicles, the CHIPS Act), it is at the same time working to reduce U.S. chip sales. It’s akin to when the U.S. government subsidized tobacco farmers while taxing cigarettes.
Many who advocate for export controls do so because they can’t think of other steps to impede China technologically. And there are indeed few technologies that we can cut off from China other than chips. In virtually all other technologies, China can either make them themselves or easily import them from non-U.S. sources. But that doesn’t mean that chip exports should be the go-to tool. One reason is that China is making significant progress in building a robust semiconductor ecosystem, with much of that progress stemming from robust efforts not to be reliant on the United States. One key way China does this is to copy and reengineer foreign semiconductor equipment. When it can do this, imposing or threatening to impose export controls on this kind of equipment simply spurs the substitution of Western equipment with Chinese equipment. To the extent the United States should focus on semiconductor export controls, it should be on items that the Chinese are incapable of copying anytime soon. And once they get close to gaining that capability, relevant export controls should be relaxed. At the same time, there is little purpose in controlling U.S. exports if China can gain the needed capabilities from other nations. For example, under the current foreign direct product rule, U.S. companies cannot sell test equipment for making chips, but Japan is not covered by this, and China already has capabilities in this area. This highlights the critical importance of coordinating with allied nations when implementing any type of export controls on China.
Some will argue that there are no real costs to the United States of limiting chip sales to China because, eventually, China will achieve complete self-sufficiency. If that is the case, it’s unclear what national security goals have been achieved. If anything, by forcing China to be more self-sufficient in the semiconductor industry, the United States has given up critical leverage it might use to respond to armed aggression by China. Moreover, why give up many years, if not decades, of Chinese revenue that would fund American technology development? Every dollar of sales made in China is a dollar less the Chinese get and a dollar more U.S. firms get that they can use for innovation.
Similarly, some will argue that China can’t really produce these chips independently and that the controls have a real bite. This is why some reacted to Huawei’s recent announcement of self-sufficiency in 7nm chips with a shrug, arguing, for example, that yield rates are not high. But this is not the point. Low yield rates simply raise prices, something the Chinese government is more than happy to subsidize. Second, yield rates will likely increase with experience and economies of scale.
The idea that the United States can impose export controls on China without Chinese government retaliation is fanciful. We already have seen retaliation in the form of import bans on Micron memory chips and export bans on gallium and germanium metals for use in semiconductors. If the United States ratchets up semi-export controls to “fill gaps,” rest assured, China will be even more aggressive with its retaliation this time.
Finally, some will argue that U.S. semiconductor and equipment firms should just be silent and accept lower sales, especially because most benefit from the CHIPS Act. But this misses the point. If Intel, Micron, and Qualcomm were state-owned, their interests would be the same: sell more to China to maintain and hopefully expand their competitiveness over Chinese firms. In this case, to paraphrase GM’s “Engine” Charlie Wilson, what’s in the interest of Intel is often in the interest of the United States and vice versa.
It might be one thing if the Chinese were like the Soviets: an economy with quite narrow and often weak technological capabilities and one that never threatened the United States from a techno-economic perspective. But China is not the Soviet Union. It is on its way to becoming the most advanced technology economy in the world. Walking away from the Chinese market because of export controls will only speed up that day of reckoning.
The administration wants to pitch its export control regime as a “small garden, high wall,” meaning strict controls on a few products. Historically, this has meant a very limited number of technology products used principally, if not solely, for military applications. Including a wide array of semiconductors used mostly for commercial purposes is now a mid-sized garden and, given China’s capabilities of making their own advanced chips and the inability of the U.S. government to enlist all possible exporting countries in a ban, a mid-sized wall.
Lest the reader think that this is a plea for being soft on China, please read ITIF’s many reports, blogs, and op-eds, dating back to 2008, calling for the U.S. government and allies to be much tougher on Chinese innovation mercantilism. The issue here is not whether the United States works to restrict Chinese innovation advancement; it’s how to do that in a way that is effective without hurting U.S. capabilities.