TSMC is the ultimate battleground in a US-China wrangle where no one can win
November 6 2019
by Daniel Bizo
The US government has reportedly applied political pressure on Taiwan, with which it has no official diplomatic ties, to curb exports of high-tech silicon to China. Taiwan is home to the world's largest contract chip manufacturing (foundry) industry, and Taiwan Semiconductor Manufacturing Corporation (TSMC) is the dominant force behind it.
This is why it now finds itself in the crosshairs of the current US administration's effort to restrict the access of Huawei (and potentially other Chinese technology stalwarts) to essential semiconductor technology. However, China is vital to TSMC's financial health to pay for its multibillion-dollar fabrication plants. US and European companies also depend on TSMC for many of their products, which exposes them to any degradation in TSMC's ability to invest in cutting-edge chip capacity, let alone any tit-for-tat retaliation in the semiconductor supply chain.
The 451 Take
The US government has already sanctioned Huawei hard, with technology export controls of US processors and intellectual property around chip designs. But going after TSMC, if successful, will likely hit a nerve in Beijing. We expect TSMC and Taiwan to resist pressure from an administration that may not even be around two years from now – particularly as it is potentially destabilizing to a relationship that is brittle at the best of times. China cannot afford the risk of losing access to cutting-edge chip-making services, which it will likely make clear to the island nation. However, should Taiwan or TSMC ultimately comply with US demands and stop offering their most advanced chip manufacturing technologies to Huawei, there is an elevated risk of consequences.
Semiconductor technology has entered geopolitical prime time over the past couple of years, as the tension between the US and China has escalated over irreconcilable views of trade and power balances.The US government has placed Huawei on a list for strict technology export controls of any product that, by value, contains more than a quarter of US-originated intellectual property.
Because there is no trivial or objective method to calculate the implied value of US-based IP content once it's rolled up into more complex products and services delivered by a global supply chain, some companies, such as UK-based chip design licensing house Arm, have played it safe and complied with the US sanction. This already hit Huawei hard because its products draw heavily on chips, technology licenses and various software design tools from US firms and other suppliers that have a major US presence.
Earlier this year, TSMC reported it does not reach the threshold, and does not need to apply any restrictions, bringing a sigh of relief to Huawei and its Chinese peers. TSMC is not only the world's largest foundry, with $33.9bn in revenue in 2018 – it is also the only one that can take a customer's processor design and manufacture on advanced technology at high volumes (processing hundreds of thousands of silicon wafers a month) that range from smartphone chips and network processors to graphics processing units and server processors. Other foundries simply do not offer a comparable combination of advanced semiconductor technology and vast manufacturing capacity.
In 2018, TSMC generated 17% of its revenue from China – a number expected to rise closer to 20% in 2019, and even further in the coming years. This is not revenue TSMC will be willing to risk if it concedes to US demands. This would undermine its R&D budget and capital expenditure plans; ultimately hurting US and European chip designers and their customers.
China cannot afford to let TSMC go
For Huawei, losing access to Intel's processors, electronic design automation software and Arm intellectual property is an existential threat. It needs key US technologies for a vast array of products it sells globally, and to power its infrastructure services to compete with Alibaba, Baidu and Tencent for cloud supremacy.
For China as a whole, on the other hand, it is a painful prospect; but one that, within some acceptable time frame and effort, it can ride through via transitioning to its own chip designs and ecosystem. It would take years, but China already has a well-established ecosystem of chip design houses to build on, drawing on expertise gained from working with US and European chip design houses such as Arm, MIPS, Qualcomm and IBM.
However, losing access to the most advanced manufacturing technologies of TSMC, and potentially other advanced foundries with close ties to the US such as Samsung Electronics, would likely pose an insurmountable technological (and, in turn, economic) challenge, even for China.
Despite its efforts investing billions of dollars and decades to catch up in semiconductor technology, China continues to rely on imports of integrated circuits – to the tune of tens of billions of dollar every year. It is the single biggest importer of chips. Export controls to global foundry services would frustrate any transition to chip design self-sufficiency because the end products would have no chance of being competitive with what the rest of the world has access to.
It is difficult to overestimate the complexity and cost of developing advanced semiconductor technology and high-volume manufacturing capacity that would be competitive with what TSMC or Samsung has to offer. The history of chip manufacturing is a history of industry consolidation under the relentless pressure of semiconductor economics: technology and fabrication plants become harder and more expensive every generation, reducing the number of players that can fund a competitive development and capital plan. Today, only Intel, TSMC and Samsung can finance the race to develop the most advanced silicon technology – each will have spent over $10bn on their chip fabrication plants in 2019.
The most recent to capitulate is Globalfoundries, the former manufacturing arm of AMD, bought and recapitalized by Abu Dhabi's sovereign fund Mubadala Investment Company, which threw in the towel last year after abandoning plans to develop sub-10nm technologies to compete with TSMC and Samsung. Intel's current 10nm technology suffered severe delays even as the world's largest maker of processors, and the traditional technology leader struggled for years to get it under control and into shape for high-volume manufacturing.
While China could, in theory, fund the billions of dollars it would take to build up its foundry industry, this is unlikely to yield results on a meaningful horizon. China's current largest foundry player, Semiconductor Manufacturing International (SMIC), pales in comparison to TSMC in every metric.
With less than one-tenth of TMSC's sales, its research and development budget is less than a fifth of its Taiwanese counterpart. Its technology is at least five years behind – a gap that will widen in the coming years as TSMC presses ahead with extreme ultraviolet lithography tools, which is a major factory overhaul that requires billions of dollars in capital investment. Customers don't appear to find SMIC's proposition attractive either: the company posted losses in the second quarter of 2019 as its revenue dropped, again, by 11.2% over the same period of 2018.
To put it simply, China's foundry industry has neither the capacity nor the technology to be competitive with what TSMC has to offer. From this point of view, past technology export controls to China, without recent attacks on Huawei and ZTE, have already succeeded: chip manufacturing technology remains beyond the reach of China.
Due to the ever greater complexity and specialty expertise involved, more money alone won't help China build competitively advanced semiconductor fabrication technologies when it has failed so far. But access to global chip manufacturing is a key component to the global strategic competitiveness, not only of its tech sector but to its broader, already highly digitized, economy. A five-year gap would take many more years to close – if the efforts succeed at all. That is a cost too high to pass without broad repercussions.