With the news this week of the US Government’s latest rounds of restrictions on China’s semiconductor industry it has hard to escape the conclusion that the two countries have now moved beyond a trade war into a full-blown economic war. We have been reluctant to use the word ‘war’ because it carries such significant weight, but at the same time there are many who think the notion of economic war is not heavy enough and hold that the two countries have been engaged in some new form of Cold War for a long time already. Wherever we stand now on that very hazy spectrum, this week’s restriction signal a marked escalation. Here, we want to think through what the means and try to think thought what may happen next.
We think it is important to take this high level approach because we are increasingly uncertain that anyone in the US government is doing so.
By way of background, the US Department of Commerce published this document this week “Implementation of Additional Export Controls: Certain Advanced Computing and Semiconductor Manufacturing Items: Supercomputer and Semiconductor End Use: Entity List Modification”. This version (which may reflect a preliminary draft) is 139 pages long and contains dozens of restrictions and export control mechanisms. We will not detail them here. They have been well covered in the press already, Dylan Patel’s is probably the best.
Even a casual read throws off some big questions, these is a lot that is either left open to regulatory interpretation or just not considered. And a thorough reading shows a lack of coherent strategic intent.
The adage goes that every military enters war prepared to fight the last one. And while this is not a military matter, that sentiment still holds. The document is full of mentions of Weapons of Mass Destruction and Terrorism, strongly reminiscent of the Global War on Terror. But more to the point, these restrictions are largely redolent of the US-Soviet Cold War. As handy as that term may be for writing headlines and attention-grabbing Twitter threads, we find it grossly misleading.
The US and the Soviet Union operated largely siloed economies for 70 years, with very limited inter-dependencies. By contrast, the US’s and China’s economy are co-dependent. In the 1980’s our favorite arcade game was banned from export to the Soviet Union because its proto-ray-tracing graphics processors were believed to be more advanced than anything the Soviets then produced. Today that game would have to be assembled in Shenzhen. The latest restrictions do not seem to fully conform to, or even recognize, that reality.
Despite being issued by the Department of Commerce, these rules seem to be largely focused on military matters. And while we are not experts on US Policy making, we have to think that the Department of Defense had a large hand in authoring the document and defining the targets. To quote another adage, to the man with a hammer, every problem looks like a nail.
That is the root of the problem. While we accept the right of the US government to determine what is in its national security interests, these measures will have far reaching commercial and economic impact, and we would argue, will only be a small hindrance to curbing the People’s Liberation Army’s (PLA) access to high tech components. The only way to truly cut off the PLA’s ability to procure chips is to completely cripple China’s semis industry. It is not clear to us that the US government fully grasps this, but we are certain that the Chinese government views it this way.
Take the effectiveness of these restrictions. Military systems like missiles and ‘smart’ munitions need a fair amount of semis for advanced guidance, but the total quantity of chips is measured in millions, a drop in the bucket of the billions of chips produced every quarter. Procuring these chips in the sprawling semis supply chain may get a bit more cumbersome for the PLA, but there are enough loopholes and grey markets that they will still be able to get close to the quantities they want.
Another major focus of the restrictions is “supercomputers”. Again, we recognize the importance of these systems especially for designing nuclear weapons. The latest restrictions seem to want to limit the ability of the PLA to acquire the tens of thousands of CPUs and GPUs these systems require (again, not a big number in the grand scheme of things). But China already has several dozen start-ups designing these chips, and many of these products are approaching globally competitive performance levels. The only way to eliminate those chips is to cut off those start-ups’ access to the foundries at TSMC and Samsung. This certainly seems to be in the cards, but how exactly will that compliance regime work? Will TSMC have to vet every customer or every order? Given the whac-a-mole of shell companies already in use in China and the US Government’s clear struggle to trace ultimate owners how effective will that compliance process be?
And then there is the question of Intellectual Property (IP). This is a major element of semis today. Going back to supercomputers, this week Chinese supercomputer maker Phytium announced a new chip built on an advanced set of Arm IP. Will that licensing be allowed to continue? In all 139 pages of restrictions the topic of IP appears exactly zero times. So no one knows. But even if Phytium gets cut off from Arm, it could switch to open sourced RISC V for its next chip.
All of this leads to what is likely to be the biggest problem for the new restrictions – US and Allied companies. Arm has gone through a lot to be able to do business in China. Is that now at risk? How compliant will they be? Arm is a UK company, US companies will likely prove even more recalcitrant when interacting with the new restrictions, complying up to and no further than the letter of the law, which their lawyers are already parsing very closely. For most chip companies, China is their second largest market, and they will not be keen to lose that.
To put it simply, these regulations are a dusted off version of 1970’s Cold War measures, but the situation is so different we have a hard time seeing them proving effective and highly likely to spur strong counter reactions in China.
That all being said, what can be done differently? This depends entirely on the outcome the US government is seeking. If the goal is to hobble the PLA’s effectiveness, then we need to accept the fact that these sanctions may work, but only for a limited time period until China finds workarounds and accelerates its own semis capabilities. And it will come at the high price of greatly increased tensions between the two countries as well as serious side effects for the entire global semis industry.
On the other hand, if instead the goal is to maintain the US’s lead in technology, military or otherwise, we would like to see more constructive efforts in the US. The US should focus on leading the way in advanced compute (AI, quantum, etc.), energy systems, biotech and all the rest of the things that the US does really well. The CHIPs does not come anywhere close to accomplishing this, but is at least pointed in the right direction. The goal should have been to force China to open its market more, not to wall it off. From a purely military strategic perspective, if your adversary is heavily reliant on your technology, better to find ways to make them more dependent on your technology rather than force them to wildly accelerate their own capabilities.
As students of history, we often find ourselves wondering how conflicts go from small disputes to truly horrible wars, with participants often seemingly surprised to find themselves so deeply enmeshed. In these latest restrictions we see a glimpse of this process. Recall that this all started in 2017, when the Trump administration started a trade war, focused on tariffs on unfair trade practice. Somewhere along the line (hint: it happened when Huawei became a target), that trade war became something different, losing the focus on tariffs and instead on broader targets. These latest export curbs are a dramatic further escalation of that poorly defined process. It would give us some modicum of comfort to know that the people putting them in place recognize the extent of this escalation and had some clearer picture of what the ultimate outcome is supposed to be. There is significant risk that both sides under estimate the other side’s resolve and ability to respond, and thus drag us all into far worse outcomes.