David P. Goldman
Chris Miller’s Chip Wars tries to combine a chatty history of the chip industry with a strategic assessment of Sino-American tech rivalry and fails at both. Absent from both facets of Prof. Miller’s book is an appreciation of what made chips so important in the first place. His account draws on more than a hundred interviews with Silicon Valley types and incorporates their bias towards futuristic AI applications rather than practicable industrial and military applications.
The book’s publication coincided with the October 2022 announcement of the most comprehensive ban on US technology exports since the Cold War, and in some ways the most radical. Sales to China account for about 30 percent of the revenue of US makers of semiconductor manufacturing equipment, and the Biden Administration’s attempt to de-globalize the industry has the makings of a Wagnerian tragedy—a Twilight of the Gods (Götterdämmerung) of free trade. But a Chipperdämmerung is not in the offing, despite Miller’s best efforts to dramatize the issue.
“From swarms of autonomous drones to invisible battles in cyberspace and across the electronic spectrum, the future of war will be defined by computing power,” Miller asserts, adding:
China has fielded an array of weapons that systematically undermine U.S. advantages. China’s precision anti-ship missiles make it extremely dangerous for U.S. surface ships to transit the Taiwan Strait in a time of war, holding American naval power at bay. New air defense systems contest America’s ability to dominate the airspace in a conflict. Long-range land attack missiles threaten the network of American military bases from Japan to Guam. China’s anti-satellite weapons threaten to disable communications and GPS networks…. China might try to jam American communications and blind surveillance systems, leaving the U.S. military unable to see enemies or communicate with allies.
This is correct, and the US Defense Department’s 2022 report on China’s military explains it more vividly, noting that the PLA has the “capability to conduct long-range precision strikes against ships, including aircraft carriers, out to the Western Pacific.” China meanwhile has doubled its satellite coverage of the theater since 2018, the Pentagon notes.
Miller jumps from this correct observation to the dodgy assertion that “China is still staggeringly dependent on foreign semiconductor technology—in particular, U.S.-designed, Taiwan-fabricated processors—to undertake complex computation.” He implies that denying advanced chip technology to China can vitiate its military ambitions. That is misleading: older-generation chips power missiles, satellite sensors, and other systems that give China a decisive advantage for defense of its coasts out to 1,000 kilometers or more, as recent RAND Corporation report details.
China concentrates its military resources massively on its coasts, and the US Navy has no effective defense against its conventional missiles, let alone its new hypervelocity missiles. Such a defense would require space-based anti-missile systems, directed energy weapons (e.g., lasers), and other weapons that require breakthroughs in high-energy physics, not just computation. The AI venture community Joneses for government money in the field it knows best, but that has little to do with the military balance in the Taiwan Strait.
Think of the military role of railroads in the nineteenth century. Whether they ran at 40 or 70 miles an hour was less important than their reach. With 2,000 advanced shore-based missiles, 1,000 fourth- or fifth-generation aircraft, sixty submarines, and a layered satellite network, China dominates its own theater with the chips it can manufacture at home. Technology controls against China have no bearing on America’s ability or lack thereof to defend Taiwan in the foreseeable future, as the Defense Department knows all too well.
Prof. Miller suggests that Beijing may assault Taiwan simply because “C-hina’s ruling party has no higher goal than asserting control over Taiwan.” US officials, including Secr-etary of State Antony Blin-ken, have asserted that Ch-ina is “speeding up” plans to seize Taiwan, without explaining why China wo-uld take the risk. It is hard to find an example in Chi-nese history of a government risking war for what it might accomplish without war. Two million of Taiw-an’s 12 million working-age citizens are employed on the mainland, and Taiwanese businesses have invested $200 billion there.
For China’s Communist Party, and indeed for any government that proposes to rule China’s multi-lingual, multi-ethnic people, the prospect of a renegade province supported by a foreign power represents an existential threat. China has broken up into warring provinces too many times in its long and tragic history, and Beijing will go to war to forestall Taiwanese sovereignty. Absent that threat, Beijing will maintain the status quo. With the world’s lowest birth rate (next to South Korea), Taiwan’s workforce will shrink by two-thirds in the course of the present century, and the issue of independence will become moot.
What, then, is the object of America’s chip war against China? In his report on Washington’s campaign to suppress Huawei, China’s national champion in telecommunications, Miller gets the story right: the issue was not Huawei’s alleged spying, but American chagrin at China’s preeminence in a decisive field of technology.
Theft of intellectual property may well have benefitted [Huawei], but it can’t explain its success. No quantity of intellectual property or trade secrets is enough to build a business as big as Huawei. The company has developed efficient manufacturing processes that have driven down costs and built products that consumers see as high-quality. Huawei’s spending on R&D, meanwhile, is world leading.
The issue with Huawei went far beyond the debate over whether the company helped tap phones or pilfer data…. the real issue was that a company in the People’s Republic of China had marched up the technology ladder—from, in the late 1980s, simple phone switches to, by the late 2010s, the most advanced telecom and networking gear. Its annual R&D spending now rivaled American tech giants like Microsoft, Google, and Intel. Of all China’s tech firms, it was the most successful exporter, giving it detailed knowledge of foreign markets. It not only produced hardware for cell towers, it also designed cutting-edge smartphone chips…The pressing question was: Could the United States let a Chinese company like this succeed?
Miller adds, “The point was less that Huawei was directly supporting China’s military than that the company was advancing China’s overall level of chip design and microelectronics know-how.” This is a remarkable conclusion. For the past four years, US intelligence officials have insisted in behind-the-scenes briefings that Huawei’s alleged ability to spy on US communications represented an existential threat to national security, despite the fact (as Miller notes) that Britain’s equivalent of the NSA, the GCHQ, concluded that data integrity could be maintained on Huawei’s systems. I reached a conclusion like Miller’s in my 2020 book on China’s technology.
If the Austro-Hungarian empire was a tyranny tempered by incompetence, the Biden Administration’s attempt to limit China’s access to high-end chips and chip-making technology is a provocation mitigated by lobbying. Two years ago, when the Trump Administration banned sales of advanced chips to China’s telecommunications giant Huawei, Harvard’s Graham Allison wrote:
Could the U.S. attempt to enforce that ban become a twenty-first-century equivalent of the oil embargo the United States imposed on Japan in August 1941? While many people may not remember what happened, and while it was certainly not what the United States intended or anticipated, that action precipitated Japan’s attack on Pearl Harbor four months later—and America’s entry into World War II.
Allison, whose 2015 Atlantic essay on the “Thucydides Trap” warned of prospective war with China, raised a false alarm. Unable to make state-of-the-art smartphones, Huawei—then the world’s largest handset producer—ceded the 5G phone market to its rivals, but worked around US export controls to continue its broadband infrastructure business.
But on Oct. 6, the Biden Administration extended the ban to all Chinese firms, a far greater provocation—at least according to the headlines. American semiconductor firms, who sell about a fifth of their wares to China, are still going over the fine print with their lawyers, but the hastily and sloppily written guidelines from the Com-merce Department offer enough loopholes to maintain most business as usual.
Nvidia makes the A100 microprocessor that powers most data centers and Artificial Intelligence applications. On Nov. 7 it announced that by lowering the chip’s clock speed and rebranding the product as the A800, it could avoid losing hundreds of millions of dollars in annual sales to China while staying within the Biden guidelines. According to industry sources, the Commerce Department guidelines were written inexpertly and clumsily, such that a competent lawyer could find a way around most of them.
Miller’s account of the emergence of the chip industry is full of entertaining anecdotes, but suffers from critical omissions. Unmentioned for example is the technology that made fast, small and energy-efficient chips ubiquitous after the 1960s, namely CMOS (complementary metal-oxide semiconductors, invented at Fairchild Semiconductors in 1963 and commercialized by RCA). CMOS made it possible to pack large numbers of transistors onto a chip. With Defense Department funding during the mid-1970s, RCA Labs built CMOS chips that powered lookdown radar on American fighters, decisively shifting the strategic balance at the peak of the Cold War.
Dr. Henry Kressel, who headed RCA Labs during the 1970s, reports that perfecting the CMOS process took a decade’s worth of R&D before it took off. The technology was understood in the 1960s, but dismissed as impractical. Physicists, engineers, and line production personnel combined to prove the skeptics wrong. The industrial-scale collaboration turned a clever idea into a world-beating practicality.
The CMOS example is especially egregious, but it betrays a deeper flaw in Miller’s understanding of why chips are important. Their development depended on the industry’s ability to create new products that needed them. Only in passing does Miller mention the Internet of Things, that is, the use of cheap chips that allow devices of all kinds to communicate. And nowhere does he mention the Fourth Industrial Revolution, namely the application of Artificial Intelligence to manufacturing, logistics, and transportation.
Chips are a means, not an end. They are the railroads of the twenty-first century. In 1870, Germany and France had built roughly the same length of railways, but Germany learned how to use its rail system to move troops and munitions, giving it a decisive advantage in the 1870 Franco-Prussian War. France had more tanks and military aircraft than Germany in 1940, but Germany learned how to coordinate armor with air and infantry, and crushed France in six weeks.
By the same token, Apple can pack 57 billion transistors onto its latest 5-nanometer chip, while Huawei’s handset business has atrophied without access to high-end semiconductors. But the American technology industry has made its money in consumer applications, while China envisions a Fourth Industrial Revolution driven by semiconductors. China already has taxi services running on autonomous vehicles, fully automated factories where robots communicate over 5G networks, mines where remote-controlled machines take the place of workers underground, and ports where robotic cranes pick containers from freighters, place them in autonomous trucks, and deliver them to warehouses to be unpacked and sorted by other robots.
It isn’t the clock speed or the gate width of the chips that will determine who dominates the industrial revolution of the twenty-first century, but the design and execution of large-scale systems—civilian as well as military—that employ the chips. China may be two generations behind in its domestic production of semiconductors, but it is ahead of the United States in semiconductor applications to industry. That’s what US policymakers should worry about.