How the right U.S. chip strategy can keep Taiwan free
Opinion How the right U.S. chip strategy can keep Taiwan free
Dmitri Alperovitch is chairman of the Silverado Policy Accelerator, a national security think tank. This piece is adapted from his new book, “World on the Brink.”
It’s not hyperbole to say that in the 21st century, nothing happens without computer chips — a commodity that has both an important technological component and a geopolitical one. Understanding the geopolitics of the next half-century requires understanding this simple fact: China, as large and economically powerful as it is, does not yet possess the ability or knowledge to build the sophisticated machinery one needs to manufacture the most advanced chips.
That’s why China has been stockpiling as much Western equipment for making chips as it can procure. Largely cut off from acquiring the machinery needed for manufacturing advanced chips by the Biden administration’s export controls, it has been pumping money into older, more established chip designs while engaging in price-dumping to boost Chinese firms’ global market share. But its ambitions do not stop there. China eventually seeks to follow the lead of chipmaking powerhouse Taiwan Semiconductor Manufacturing Co. and build up its domestic capability to manufacture advanced semiconductors.
Adapted from “World on the Brink: How America Can Beat China in the Race for the Twenty-First Century” by Dmitri Alperovitch with Garrett M. Graff. Copyright © 2024. Available from PublicAffairs, an imprint of Hachette Book Group, Inc.
Such a technological breakthrough would be a major geopolitical milestone, eliminating the Chinese economy’s reliance on imported semiconductors from the West — irrevocably altering the balance of power in the new cold war that is unfolding between China and the United States.
Advanced semiconductor manufacturing tools are the most sophisticated equipment that humankind has ever built — the process of etching out a 300-millimeter silicon wafer is enormously complex. Doing so at scale and with a high success rate — what’s known in the industry as “yield” — is even more challenging.
So far, China hasn’t been able to reproduce much of that most sophisticated equipment domestically, despite pouring hundreds of billions of dollars into the industry and enabling rampant economic espionage and intellectual property theft. The systems of these chip-fabrication facilities — “fabs” for short — are among the most delicate manufacturing processes in the world. The chemicals, equipment and clean-room facilities in which semiconductors are made constitute an intricate robotic ballet that will be exquisitely difficult for China to replicate.
Any U.S. strategy to avoid losing global influence and power to China requires ensuring that China neither achieves a tipping point of domestic independence in the means of production of semiconductors nor seizes Taiwan and its valuable fabs, thereby making the rest of the world dependent on China for chips.
Because, as important as chips are to today’s world, they’re going to be even more important tomorrow.
Advanced semiconductors are critical to four main categories of products: advanced weapons systems such as guidance systems for precision munitions; high-end electronics such as iPhones, laptops and computers; gaming systems such as Xboxes and PlayStations; and the massive cloud-computing data centers run by giants such as Amazon Web Services and Google Cloud. This last category is where the entire internet lives and breathes, and where the world’s most sophisticated AI models are being trained on computer systems thousands of times more powerful than a regular desktop.
To maintain leverage over China in the next quarter-century, the United States needs to ensure that China doesn’t achieve its much-desired “chip breakout.” Luckily, the United States has tremendous power here. Preventing China from achieving its chip breakout requires a four-pronged approach:
1. Prevent China from building and operating additional domestic fabs, and sustaining existing ones, via allied export control policies on manufacturing equipment.
The strategy begins with stopping China from accessing the equipment and tools necessary to operate fabs. This is, in some ways, the easiest step the United States can take because the supply chains for the most advanced chip-manufacturing equipment are so tiny.
Making any piece of electronics today requires six sets of companies, and I mean all six. Get five out of six and you’re just as far away from a functioning device, be it an iPhone or a missile, as you would be if you only had one.
Luckily for the West, there are just three countries in the world that produce the sophisticated equipment necessary for a modern fab: the United States, the Netherlands and Japan. A coalition of just these three countries can prevent China from buying the equipment required to build and maintain its own domestic chip fabs.
The process starts with the U.S. companies that make the electronic design automation software needed to design the chips (Cadence Design Systems and Synopsys). Then come those that manufacture critical components and chemicals needed to build the chips (e.g., Japan’s JSR, Shin-Etsu Chemical Co. and Tokyo Ohka Kogyo, or TOK). After that are the companies that manufacture equipment needed to make chips (ASML in the Netherlands; U.S.-based Applied Materials, Lam Research and KLA; and Japanese firms Nikon and Tokyo Electron) and those that actually design the chips (Apple, Nvidia, Qualcomm, etc.). Then come the companies that manufacture chips (Intel, TSMC, Samsung, Micron, etc.).
Finally, production requires the companies that package chips into final products and assemble them into circuit boards for use in electronics. These are companies such as Foxconn, and others in China, Vietnam, Malaysia and elsewhere, as well as the leading so-called printed circuit boards companies, known as PCBs, which include China’s Zhen Ding Tech Groupand DSBJ, Taiwan’s Unimicron and Compeq Manufacturing, and Japan’s Nippon Mektron.
Each of these design and manufacturing stages presents a critical choke point, particularly given how small most of these industries are.
But not every choke point is equal.
The really significant choke point is in the fourth stage. Indeed, the one area where China is unlikely to make significant progress — especially if the United States and its allies ramp up export controls — is equipment manufacturing.
Equipment such as the extreme ultraviolet lithography machines produced by ASML are designed to etch transistors on a piece of silicon wafer that is tens of thousands of times thinner than a human hair — and getting smaller with each new iteration. The complexity of designing, manufacturing and maintaining these machines is immense. In addition, virtually the entire supply chain for building key components is located in the West.
For China to replicate that entire supply chain, as well as design and build these machines, is a task more complex than landing a man on Mars. Perhaps one day, with enormous effort and after at least a decade or more of trying — abetted by a historically unprecedented campaign to steal Western intellectual property electronically as well as via old-fashioned recruitment of industrial spies and thieves — it could get there. But China would catch up only as the United States and its allies move further ahead in the development of even more advanced semiconductor technology.
Fortunately, the U.S. government has recognized this critical element, and in 2022 and 2023 instituted the most sweeping export controls on American semiconductor equipment manufacturers to date. It has also successfully leaned on the Netherlands and Japan to do the same with their companies.
Chips in industry parlance fall into three categories: “legacy,” “mature” and “advanced,” the latter being the preferred term for the ever-evolving cutting edge. While there’s no standard definition, the industry generally considers any chips smaller than 20 nanometers to be “advanced” and anything larger than 40 nanometers to be “legacy.” Everything in between is considered “mature.”
But the terms “mature” and “legacy” underplay how vital even those technologies are to our modern world, making it sound like they are yesterday’s news — a dial-up modem in a fiber-optic world — whereas they remain central to just about every device in our lives. “Foundational” is a better term because these chips will remain as important to our daily digital life as aluminum is a century into the construction of airplanes.
Sure, many wouldn’t consider aluminum to be cutting-edge technology anymore, but try building an airplane without it.
Given China’s propensity both to engage in massive industrial subsidies and to dump products at lower cost on advanced economies to capture market share and decimate competitors — a playbook the country has used repeatedly with rare earth metals, solar panels, steel, textiles and other sectors — it is not surprising that the Chinese are now apparently trying to do the same with foundational chips, attempting to capture the bulk of the market.
If it succeeds, it would be as if we had blocked China from producing advanced carbon-fiber materials but allowed it to corner the world’s market on aluminum — still a disastrous outcome in terms of providing Beijing with enormous leverage over the United States and the global economy.
To prevent this scenario, export controls must be expanded to cover all semiconductor manufacturing equipment to enjoin China from building, operating and maintaining fabs at any scale, a move that would further increase China’s dependence on the West and provide us with the leverage over its economy to, say, deter hostile action in the East China Sea.
2. Diversify chip manufacturing away from Taiwan and invest in U.S., South Korean, Japanese and European operations to reduce U.S. economic vulnerability in the event of a war over Taiwan.
The second component of the strategy involves diversifying chip manufacturing away from Taiwan. The extreme concentration of global chip production on an island roughly the size of Maryland is a critical vulnerability not just for the United States but for the whole world.
The potential that China might blockade or invade Taiwan and hold the United States and the rest of the world hostage by blocking access to TSMC-produced semiconductors is a mortal economic threat that we cannot risk. The answer here — as with reducing U.S. economic reliance on the Chinese chip industry — is greater domestic and allied investment.
The primary goal behind the 2022 Chips Act was to increase the market share of U.S. chips production after it fell from 37 percent in the 1990s to just 12 percent in 2022. By providing government incentives for American and allied companies, such as Samsung and TSMC, to build fabs in the United States, we secure the supply chain in two ways, at both the top and bottom of the process. While we won’t be completely independent of Taiwanese chips anytime soon, the Chips Act helps ensure that our supply of semiconductors is slowly diversified away from vulnerable Taiwan, while providing American chipmaking tool companies a larger domestic market that can offset any loss of access in China caused by U.S. export controls.
In addition, the Chips Act has helped to start a new “chips race,” with countries such as South Korea, Japan, Singapore, Israel, Germany and France, among others, rushing to provide their own subsidies and incentives to global chip companies to build plants within their borders. While the Chips Act’s $76 billion top-line number is a drop in the bucket considering that just one modern advanced fab could cost as much as $10 billion, the total announced and planned investments in this industry through 2030 by the United States and its international partners in both the public and private sectors is estimated to be over $1.2 trillion, according to calculations by my research analysts at Silverado Policy Accelerator.
That level of investment will transform the industry, diversify it and advance it at a pace that was unimaginable just two years ago.
3. Make clear to China that if it attempts to seize Taiwan, it will never gain access to TSMC’s technology and know-how.
The third part of the strategy is making sure the Chinese government understands that Taiwan is not the answer to China’s chip breakout plan. While both the Chinese Communist Party in general and President Xi Jinping personally have numerous historical, political and economic reasons for wanting to invade Taiwan — reasons that predate the age of computers entirely, let alone the rise of TSMC — we need to make sure that the attractiveness of TSMC’s plants, know-how, equipment and people does not become a tipping point that Xi uses to justify an invasion.
It is essential for U.S. and allied diplomats to make clear to Xi quietly that under no circumstance will China be allowed to control TSMC fabs should it use force to invade the island. Regardless of whether the United States chooses to fight China over Taiwan, it would have numerous ways to cripple the fabs in the event of Chinese takeover — from working with the Taiwanese to destroy on-site equipment to working with Western manufacturers to disable TSMC operations remotely or impose an embargo that would prevent Chinese-occupied labs from getting their hands on chemicals and other essential components.
4. Spell out that, much as the United States and the rest of the West banded together quickly and decisively after Russia’s invasion of Ukraine, any move by China against Taiwan would launch a Western blockade of semiconductors to China that would cripple its economy.
Finally, as an additional deterrent, China must be made aware of the consequences to its economy should it decide to invade Taiwan.
This is where the various threads of this strategy combine to deliver even greater force: The United States can communicate that any move against Taiwan will be met with similarly crippling sanctions to those instituted against Russia after its attack on Ukraine. It would be an even more powerful threat if our export control measures were successful in preventing China from building its own chip fabs and China were separately convinced that it could not take Taiwanese fabs by force.
Russia has managed to escape some of the worst impacts of the U.S. semiconductor export ban by procuring black-market chips — including chips manufactured in the United States — from China, but there is no such alternative for China. And its demand for chips is orders of magnitude higher than Russia’s: While Russia imported only $2 billion worth of computer chips in 2021 according to official statistics reported by Global Trade Tracker — a low volume that is comparatively easy to procure on the black market — China is importing nearly 200 times that at $376 billion worth of chips annually. You simply can’t find such a staggering amount of chips falling off the back of a proverbial truck.
Using the powerful foreign direct product rule, which enables the Commerce Department to extend its export control authority to restrict or prohibit the transfer of any chip made anywhere in the world because all are made with a substantial amount of U.S. technology, software or equipment, the United States can block allied countries from supplying chips to China if it invades Taiwan, a move that would devastate the Chinese economy.
Crucially, any realistic threat of a chips blockade following an invasion hinges on China’s continued dependence on the West for chips.
Thus, it is important to limit current U.S. export control measures to equipment and components, not the chips themselves — except, of course, for the small category of very advanced chips necessary for the training and running of AI models, which should continue to be barred from export to China.
The goal is to prevent China from manufacturing its own chips, not to starve it of chips entirely — at least as long as it doesn’t invade Taiwan. In fact, the ultimate objective of this strategy, which I call “unidirectional entanglement,” is to make China more dependent on the West, even as we reduce our own dependence on China and Taiwan.
Played correctly, the West’s leverage with computer chips, small as they are, could be the very thing that keeps Taiwan free and democratic for years to come.
About guest opinion submissions
The Washington Post accepts opinion articles on any topic. We welcome submissions on local, national and international issues. We publish work that varies in length and format, including multimedia. Submit a guest opinion or read our guide to writing an opinion article.
No comments:
Post a Comment