“Choking out” China
It is no secret that US-China relations have soured over recent years. One of the most divisive issues underlying the tension between the two global powers is semiconductors, those tiny little microchips that go into just about every electronic device including smartphones, PCs, cars and even ordinary household appliances. Over the past few years, the US government has increasingly restricted the sale, be it directly or indirectly, of certain semiconductors and semiconductor equipment to China. Reasons for the effective ban include concerns that devices made by Chinese companies such as Huawei, have been used by the Chinese government for espionage purposes. In addition, security officials within the US government are concerned that semiconductors designed for purposes of artificial intelligence (AI) are being utilised by the Chinese military in areas such as autonomous weapons.
What is less talked about but still an underlying reason for the Chinese chip ban is the growing influence of China in the global semiconductor industry and the threat that this poses to the US. America is the current de facto leader of the semiconductor industry owing to its dominance in the technologies that underlie semiconductor design and manufacturing. In his book, Chip Wars, author Chris Miller highlights a number of “choke points” within the semiconductor industry and how the US is able to exert its influence through these areas.
To begin with, nearly every chip in the world is designed using software from one of three US-based firms (Synopsys, Cadence, Mentor – which is owned by Siemens but based in Oregon). It is almost impossible to design a semiconductor without software from one of these companies and many of the smaller firms that offer design software are also US-based making it incredibly difficult to sidestep the US in this regard.
Another segment where the US is able to apply pressure is on the manufacturing side. There are five companies that dominate the market when it comes to semiconductor manufacturing equipment, three of which are based in the US (Applied Materials, KLA and Lam Research), one in Japan (Tokyo Electron) and the other in the Netherlands (ASML).
What provides the US with even more power in this space is that when it comes to the fabrication of advanced semiconductors, excluding those that are manufactured inhouse by Intel within the US, all leading-edge chips are manufactured by just two companies, Taiwan Semiconductor Manufacturing Company (“TSMC”) and Samsung. While these are not US-based companies, they use equipment that is made using US technology. In particular, there is only one company in the world that can produce tools and systems capable of fabricating the most advanced semiconductors and that is ASML. The company may be based in Europe but much of the technology that it utilises can be traced back to California. Through its control of the intellectual property, the US government is still able to dictate who can access the most advanced chips regardless of whether a non-US firm is selling the chips or even the equipment to fabricate them.
The increasingly restrictive approach by the US towards China has fostered a growing concern over the potential invasion of Taiwan by China. The Chinese Communist Party still considers Taiwan as territory that belongs to China and there are a host of reasons as to why China would consider invading the island. Of relevance to this conversation would be the opportunity to gain control of Taiwan’s semiconductor industry, specifically the world’s leading manufacturer of semiconductors, TSMC.
China and Taiwan are separated by a body of water named the Taiwan Strait and at its narrowest the distance between the two reaches only 130 kilometres. TSMC’s fabrication plants in Taiwan are also located on the West coast of the island placing them right on the battle line should any conflict arise. The close proximity of TSMC facilities as well as China’s growing military power, both naval and air force, could allow one to conclude that an invasion is both likely as well as feasible.
While an invasion remains possible and the probability thereof has increased, Chris Miller believes, with quite rational and logical justification, that it is still unlikely China will invade, at least with regards to gaining control of Taiwan’s semiconductor operations.
Firstly, it would only take a handful of TSMC employees to sabotage the entire operation of TSMC and Chinese forces may not be able to capture TSMC employees with critical skills and operational knowledge. Secondly, because TSMC still requires crucial materials and tools, including software updates from firms such as ASML, the fabrication plants would not be able to function effectively especially once the likes of ASML stopped providing access. Finally, even if China was able to successfully capture TSMC or even pressure it into providing China with preferential access, the US along with Europe and Japan would likely crank up the extent of their restrictions, effectively shutting off a Chinese-controlled TSMC from its customers as well as critical new technologies required to advance to smaller, more powerful chips. The end result could see China suffering more than anyone else as the rest of the world would still have access to facilities from Intel and Samsung which would be able to produce more advanced chips compared to a critically handicapped TSMC.
The potential invasion of Taiwan aside, the influence that the US is able to exert over the semiconductor industry is not something that Chinese President XI Jinping is willing to accept as he would like to see China become the world leader in semiconductors. To this end, China has been pouring billions of dollars into its domestic semiconductor industry with the goal of creating an independent ecosystem that is not critically reliant on another other country or foreign supplier. Given the size of the Chinese market and the resources at its disposal on face value, one could easily believe that such an objective is not out of the realms of possibility.
However, Mr. Miller notes that there are a number of obstacles that China faces, some of which may not be possible to overcome. From a big picture perspective, to achieve its “technological independence”, China would need to obtain or develop the following: software to design semiconductors; skills, expertise and equipment to manufacture leading-edge semiconductors at scale; design capabilities that can match leading US firms; and advanced materials used in the semiconductor process. China does not possess the above, at least not at a competitive level.
To put into perspective how difficult it would be for China to become independent, consider the EUV tool that is made by ASML. In order to produce the world’s smallest, most powerful chips, manufacturers such as TSMC and Samsung require EUV tools from ASML. As previously mentioned, ASML is the only company in the world that can produce machinery capable of fabricating leading-edge chips at scale. It took ASML many years of R&D as well as billions of dollars of investment to develop the technologies that go into ASML’s equipment. Such technologies include lasers that can heat up tin to a temperature greater than the surface of the sun or specially designed mirrors that are effectively the smoothest objects ever made. On the assumption that China was able to replicate these incredibly advanced technologies in the short space of a few years, it would still lag behind the curve because ASML would have already launched a better, more advanced tool capable of producing even smaller chips.
Ultimately, the biggest argument against China achieving its objective in totality is that no country in the world possesses such technological independence, not even the US. This is because the semiconductor industry is interconnected on a global scale. A typical chip that you would find in your smartphone has likely been designed using the architecture from UK-based firm ARM by a team of engineers sitting in California or Israel that specialise in designing semiconductors using software from a US-based firm. Once the design is complete, it is sent to a facility in Taiwan to be manufactured on mass scale using silicon wafers, chemicals and gases imported from Japan. Machinery and tools produced by ASML, as well as a handful of other US firms, will then “carve” incredibly precise designs into the silicon, effectively printing wires at a scale so small it is measured in nanometres (one billionth of a metre). Once complete, the silicon wafer which has a diameter equivalent to a standard ruler, will contain billions of microchips made up of trillions of transistors. The chips on the silicon wafter are then packed and tested in
Southeast Asia before being sent to China for assembly into your phone.
As a more likely scenario, it is Miller’s view that China recognises the power that the US wields and instead will pursue a path of increasing its own leverage at certain choke points thereby reducing the overall influence of the US. Smartphones, PCs and data centers require the most advanced chips but cars and other consumer devices use old technologies or lagging-edge nodes (larger, less powerful chips). Much of the investment into manufacturing capacity within China is for lagging-edge nodes, an area that has not been targeted by US restrictions. In the event that the US once again increases the restrictions of chip sales into China, they will likely apply to more advanced chips as opposed to older technologies. Thus China will still be able to increase its influence over the supply of lagging-edge chips. It is worth remembering that the recent chip shortages were predominantly on the lagging-edge side and the effect was felt on a global scale. The US would be reluctant to place restrictions that impact the sale of these chips given what transpired during COVID and the issues that still plague corporates to this day.
A scenario that is more likely to unfold in one where China’s share of global manufacturing capacity, in terms of volume, could see it overtake the likes of Taiwan and South Korea, providing it with significant influence and a choke point of its own. China can also increase its influence over the materials used to make semiconductors, especially emerging materials, which may play a bigger part in the future.
The ramp up in restrictions by the US and its allies are likely to have a negative impact on company sales for semiconductor firms that are impacted. As a reminder, the latest bout of restrictions implemented in the second half of 2022, were aimed at leading-edge technologies with particular focus on equipment used in the fabrication process of advanced semiconductor chips as well as chips that are used specifically for artificial intelligence purposes. Compared to other equipment firms, the impact of the latest restrictions were not as severe for ASML as it was already restricted from selling its EUV tools to China and companies that do use these tools were already prohibited from using them to make chips for Chinese end users.
However, for the other equipment firms such as Applied Materials, KLA and Lam Research, the restrictions may be somewhat more impactful. Certain of their products and services (those used in the production of leading-edge chips) can no longer be sold to Chinese customers and when dissecting their revenue bases it is quite obvious that Chinese customers constitute a material portion of the overall base.
That said, it should be noted that a sizeable portion of the tools sold by these companies to China are for lagging-edge purposes – for use in the fabrication of older, larger chips. This is not an area the US government is currently targeting and therefore the financial impact felt by these equipment firms may not be as severe as some might think. Given the impact from the chip shortage that is still being felt today, US authorities would certainly have reason to pause in applying restrictions on equipment focused on producing older technology chips.
The ban on the sale of chips used for AI purposes will be most impactful on companies such as Nvidia and AMD. The restrictions put in place are quite specific in their wording and seem to be constructed such that they are specifically targeting the most advanced high-end chips which are used for AI training. In a similar scenario to the equipment firms above, this opens the door for sales of chips that are not at the top of the spectrum. Nvidia’s CEO Jensen Huang had remarked in an interview following the introduction of the new restrictions that “It’s excellent to know what the law is and it’s excellent to know what our limitations are. The limitations and the restrictions are very specific to a combination of computation level and multi-chip interconnection level. That restriction gives us plenty of envelope to go and run our business and for the vast majority of our customers in China, this is either unaffected because we have so many alternative products…”
What he was alluding to is that Nvidia had already begun to redesign certain of its high-end chips such that they would comply to the letter with the new restrictions, thereby mitigating the negative financial impact.
While the US holds a dominant position in the global semiconductor industry through its influence at certain choke points, it has become increasingly reliant on facilities located in Taiwan, South Korea and even China for the purposes of manufacturing semiconductors. Two decades ago, the US contributed over a third to global manufacturing capacity, today this figure is closer to 10%.
To bolster its manufacturing capabilities as well as apply further pressure to China, last year the US signed into law the CHIPS Act (Creating Helpful Incentives to Produce Semiconductors and Science Act of 2022) which will see significant investment flow into the US semiconductor industry with regards to R&D as well as manufacturing capacity. Along with incentivising the commercialisation and R&D of leading-edge technologies such as AI, clean energy and quantum computing to name but a few, the intent of this piece of legislation is to alleviate supply concerns, especially for the defence sector, that arose during COVID and its aftermath as well as increase the competitiveness of the US in terms of semiconductor manufacturing.
It is difficult to say how big a boost the Act will ultimately be for America’s semiconductor industry but it seems less likely that it will help the US achieve manufacturing independence. While the introduction of the legislation has been met with a positive response from the likes of TSMC and Samsung who intend to invest billions in production capacity in the US, the impact may not be as meaningful as one might think.
In the case of TSMC, the facility it intends to build in Arizona will produce cutting edge chips but they will not be at the most advanced node (two nanometres or less), the fabrication of which will remain back in Taiwan. It is likely that the investment by these foreign firms is more to placate US senators, in so far as they will produce chips for the US defence industry, rather than to effectively address capacity constraints facing the US. Ultimately, the construction of new or improved facilities by foreign firms will increase production capacity in the US but it will not be enough to sufficiently reduce the country’s reliance on Southeast Asian firms for manufacturing purposes. Instead, TSMC and Samsung are more likely to keep the majority of their production capacity along with their most advanced facilities within the confines of their borders.
For the US to achieve true manufacturing independence or something close thereto, it will have to look to local firms, specifically Intel and GlobalFoundries. This in of itself presents its own set of challenges that may be insurmountable despite the $280 billion worth of incentives contained in the CHIPS Act.
Starting with Intel, current CEO Pat Gelsinger lobbied aggressively for the CHIPS Act as he, more so than most, realises what is at stake. Following a series of poor managerial decisions by previous management regimes where profits were favoured over progress, Intel has fallen behind the likes of TSMC when it comes to manufacturing capabilities. To revitalise its business as well as fulfil a role as the leading manufacturer of leading-edge chips for the US, Mr. Gelsinger has adopted a three-pronged approach.
Supported by the subsidies and incentives provided for in the CHIPS Act, Intel will need to firstly cut a deal with ASML to acquire their latest EUV tools before TSMC is able to obtain access. Secondly, Intel will need to launch a contract manufacturing business or foundry whereby it will manufacture chips on behalf of other semiconductor designers placing it in direct competition with the likes of TSMC and Samsung. Given the competitive strength of TSMC, specifically the skills and IP it has built up in developing and enhancing its manufacturing processes, Intel’s strategy already appears to be on shaky ground. Ironically, Intel’s third prong is outsourcing the manufacturing of its most advanced inhouse chips to none other than TSMC as Intel tries to fix its own inhouse manufacturing issues. This may already sound like a bit of a stretch but assuming Intel is successful in all three parts of its strategy and it is able to provide the US with a local source of manufacturing that is at a level comparable to the likes of TSMC and Samsung, it will only be for leading-edge chips.
To address trailing-edge technologies, the US will have to turn to another US-based firm, GlobalFoundries. It is in this area where the US faces a far greater challenge from China. GlobalFoundries is a medium-sized foundry and though it is planning to build a new facility in Vermont aided by financial support from the CHIPS Act, it faces significant competition from Chinese competitors. One also needs to consider that these Chinese companies are also subsidised by their own government. The CHIPS Act may help bolster domestic manufacturing of trailing-edge chips but from a US perspective, more would have to be done to prevent China from increasing its influence such that it would possess its own choke point in regards to the manufacture of lagging-edge chips.
Note: For those interested in learning about the history of the semiconductor industry, Chris Miller’s book “Chip Wars” is an invaluable source of information as well as an engaging read.
Get an In-depth view of the Semiconductor Industry
Our article on semiconductor investment provides insights into the industry's history, market trends, and potential risks and rewards. Whether you are a seasoned investor or just starting, we can help you make informed decisions about investing in semiconductors.