đź“ť Notes
Though its position has weakened dangerously China now spends more money each year importing chips than it spends on oil. Beijing is now more worried about a blockade measured in bytes rather than barrels. During most years of the 2000s and 2010s, China spent more money importing semiconductors than oil.
Fairchild cofounder Gordon Moore noticed in 1965 that the number of components that could be fit on each chip was doubling annually as engineers learned to fabricate ever smaller transistors. This prediction that the computing power of chips would grow exponentially came to be called Moore’s Law.
Big tech wouldn’t exist if the cost of processing and remembering 1s and 0s hadn’t fallen by a billionfold in the past half century.
Unlike oil, which can be bought from many countries, our production of computing power depends fundamentally on a series of choke points of tools chemicals and software that often are produced by a handful of companies and sometimes only by one. No other facet of the economy is so dependent on so few firms.
Chips from Taiwan provide 37 percent of the world’s new computing power each year. Two Korean companies produce 44 percent of the world’s memory chips. ASML builds 100 percent of the world’s extreme ultraviolet lithography machines without which cutting edge chips are simply impossible to make.
Texas Instruments had been founded to produce equipment using seismic waves to help oilmen decide where to drill. During WW2, the company had been drafted by the US navy to build sonar devices to track enemy submarines (Page 14).
Hardly anyone realized Vietnam had been a successful testing ground for weapons that married microelectronics and explosives in ways that would revolutionize warfare and transform American military power (Page 61).
Semiconductors were at the center of this plan Li knew. There were plenty of Taiwanese-American semiconductor engineers willing to help in Dallas. Morris Chang urged his colleagues at TI to set up a facility in Taiwan (Page 64).
Semiconductors recast the economies and politics of America’s friends in the region. Cities that had been breeding grounds for political radicalism were transformed by diligent assembly line workers happy to trade unemployment or subsistence farming for better paying jobs in factories. By the early 1980s, the electronics industry accounted for 7 percent of Singapore’s GNP and a quarter of its manufacturing jobs. Of electronics production, 60 percent was semiconductor devices and much of the rest was goods that couldn’t work without semiconductors (Page 66).
Noyce and Moore abandoned Fairchild as quickly as they’d left Shockley’s startup a decade earlier and founded Intel, which stood for Integrated Electronics (Page 67).
He proposed coupling a tiny transistor with a capacitor, a miniature storage device that is either charged (1) or not (0). Capacitors leak over time, so Dennard envisioned repeatedly charging the capacitor via the transistor. The chip would be called a dynamic random access memory or DRAM. These chips form the core of computer memory up to the present (Page 68).
Intel, however, launched a chip called the 4004 and described it as the world’s first microprocessor (Page 70).
The U.S. military lost the war in Vietnam, but the chip industry won the peace that followed, binding the rest of Asia, from Singapore to Taiwan to Japan, more closely to the U.S. via rapidly expanding investment links and supply chains (Page 78).
Sony sold 385 million units worldwide, making the Walkman one of the most popular consumer devices in history (Page 83).
U.S. firms, with GCA as the leader, controlled 85 percent of the global market for semiconductor lithography equipment in 1978. A decade later, this figure had dropped to 50 percent. GCA had no plan to turn things around (Page 96).
Being ahead of your time is good for scientists, but not necessarily for manufacturing firms seeking sales. Customers had already gotten comfortable with equipment from competitors like Nikon, Canon, and ASML and didn’t want to take a risk on new and unfamiliar tools from a company whose future was uncertain. If GCA went bankrupt, customers might struggle to get spare parts. Unless a big customer could be convinced to sign a major contract with GCA, the company would spiral toward collapse (Page 108).
In 1990, Noyce, GCA’s greatest supporter at Sematech, died of a heart attack after his morning swim. He’d built Fairchild and Intel, invented the integrated circuit, and commercialized the DRAM chips and microprocessors that undergird all modern computing (Page 108).
“The United States has been busy creating lawyers,” Morita lectured, “while Japan has been busier creating engineers. Moreover, American executives were too focused on this year’s profit, in contrast to Japanese management, which was long-range. American labor relations were hierarchical and old-style, without enough training or motivation for shop floor employees. Americans should stop complaining about Japan’s success,” Morita believed. “It was time to tell his American friends, Japan’s system simply worked better” (Page 110).
By the mid-1980s, Micron used far fewer production steps than its competitors, letting the company use less equipment, cutting costs further. They tweaked the lithography machines they bought from Perkin Elmer and ASML to make them more accurate than the manufacturers themselves thought possible. Furnaces were modified to bake 250 silicon wafers per load rather than the 150 wafers that was industry standard. Every step of the fabrication process that could handle more wafers or reduce production times meant lower prices. “We were figuring it out on the fly,” one early employee explained, “so unlike other chipmakers, we were prepared to do things that hadn’t been written in a paper before” (Page 121).
At one point in 1981, the company’s cash balances fell so low, it could cover only two weeks of payroll. Micron scraped through that crisis, but amid another downturn a few years later, it had to lay off half of its employees and cut salaries for the remainder (Page 121).
Finally, Intel decided to leave memories, surrendering the DRAM market to the Japanese and focusing on microprocessors for PCs. It was a gutsy gamble for a company that had been built on DRAMs (Page 125).
“If HP could grow from a Palo Alto garage to a tech behemoth, surely a fish and vegetables shop like Samsung could too. It’s all thanks to semiconductors,” one HP employee told him (Page 131).
The KGB began stealing semiconductor manufacturing equipment too (Page 143).
Soviet chipmaking taught the country’s weapons designers to limit use of complex electronics whenever possible. This was a viable approach in the 1960s, but by the 1980s, this unwillingness to keep pace with advances in microelectronics guaranteed Soviet systems would remain dumb, even as American weapons were learning to think. The U.S. had put a guidance computer powered by Texas Instruments chips onboard the Minuteman II missile in the early 1960s, but the Soviets’ first missile guidance computer using integrated circuits wasn’t tested until 1971 (Page 146).
By the early 1980s, it was publicly admitted that the U.S. had plugged its submarine sensors into the Illiac IV, one of the most powerful supercomputers and the first using semiconductor memory chips, which were built by Fairchild (Page 147).
In 1993, the U.S. retook first place in semiconductor shipments. In 1998, South Korean firms had overtaken Japan as the world’s largest producers of DRAM, while Japan’s market share fell from 90 percent in the late 1980s to 20 percent by 1998 (Page 157).
As early as 1983, Ogarkov had gone so far as to tell American journalist Les Gelb off the record that the Cold War is over and you have won. The Soviet Union’s rockets were as powerful as ever, it had the world’s largest nuclear arsenal, but its semiconductor production couldn’t keep up. Its computer industry fell behind, its communications and surveillance technologies lagged, and the military consequences were disastrous (Page 159).
Chang had toyed with the idea of creating a semiconductor company that would manufacture chips designed by customers. At the time, chip firms like TI, Intel, and Motorola mostly manufactured chips they had designed in-house. Chang pitched this new business model to fellow TI executives in March 1976. The low cost of computing power, he explained to his TI colleagues, will open up a wealth of applications that are not now served by semiconductors, creating new sources of demand for chips, which would soon be used in everything from phones to cars to dishwashers. The firms that made these goods lacked the expertise to produce semiconductors, so they’d prefer to outsource fabrication to a specialist, he reasoned (Page 166).
When one businessman declined to invest after three meetings with Chang, Taiwan’s Prime Minister called the stingy executive and reminded him, “The government has been very good to you for the last twenty years; you better do something for the government now.” A check for Chang’s chip foundry arrived soon after. The government also provided generous tax benefits for TSMC, ensuring the company had plenty of money to invest from day one. TSMC wasn’t really a private business; it was a project of the Taiwanese state (Page 167).
TSMC’s business boomed during the 1990s, and its manufacturing processes improved relentlessly. Morris Chang wanted to become the Gutenberg of the digital era (Page 168).
But the People’s Republic had spent the 1960s denouncing capitalists while its neighbors were trying desperately to attract them. A study in 1979 found that China had hardly any commercially viable semiconductor production and only fifteen hundred computers in the entire country (Page 175).
In the early days of chipmaking, transistors were so big that the size of the light waves used by lithography tools barely mattered. But Moore’s law had progressed to the point where the scale of light waves, a couple hundred nanometers depending on the color, impacted the precision with which circuits could be etched (Page 184).
However, the manufacturing of EUV wasn’t globalized; it was monopolized. A single supply chain managed by a single company would control the future of lithography (Page 189).
The x86 architecture dominated PCs, not because it was the best, but because IBM’s first personal computer happened to use it. Like Microsoft, which provided the operating system for PCs, Intel controlled this crucial building block for the PC ecosystem (Page 192).
Today, nearly every major data center uses x86 chips from either Intel or AMD (Page 193).
Intel turned down the iPhone contract. Apple looked elsewhere for its phone chips. Jobs turned to ARM’s architecture, which, unlike x86, was optimized for mobile devices that had to economize on power consumption (Page 195).
Just a handful of years after Intel turned down the iPhone contract, Apple was making more money in smartphones than Intel was selling PC processors. Intel tried several times to scale the walls of Apple’s castle but had already lost first-mover advantage (Page 196).
A fixation on hitting short-term margin targets began to replace long-term technology leadership. The shift in power from engineers to managers accelerated this process (Page 196).
The software capable of laying out these transistors was provided by three American firms: Cadence, Synopsis, and Mentor, which controlled around three-quarters of the market (Page 199).
Logic refers to the processors that run smartphones, computers, and servers. Memory refers to DRAM, which provides the short-term memory computers need to operate, and flash, also called NAND, which remembers data over time. The third category of chips is more diffuse, including analog chips like sensors that convert visual or audio signals into digital data, radio frequency chips that communicate with cell phone networks, and semiconductors that manage how devices use electricity (Page 206).
The company that eventually came to dominate the market for graphics chips, Nvidia, had its humble beginnings not in a trendy Palo Alto coffeehouse but in a Denny’s in a rough part of San Jose (Page 210).
Nvidia’s GPUs can render images quickly because, unlike Intel’s microprocessors or other general-purpose CPUs, they’re structured to conduct lots of simple calculations, like shading pixels, simultaneously (Page 210).
For each generation of cell phone technology after 2G, Qualcomm contributed key ideas about how to transmit more data via the radio spectrum and sold specialized chips with the computing power capable of deciphering this cacophony of signals (Page 212).
This reduced power consumption per transistor because smaller transistors needed fewer electrons to flow through them. Around the early 2010s, it became unfeasible to pack transistors more densely by shrinking them two-dimensionally. One challenge was that as transistors were shrunk according to Moore’s law, the narrow length of the conductor channel occasionally caused power to leak through the circuit even when the switch was off. On top of this, the layer of silicon dioxide atop each transistor became so thin that quantum effects, like tunneling (jumping through barriers that classical physics said should be insurmountable), began seriously impacting transistor performance (Page 217).
ASML’s EUV lithography tool is the most expensive mass-produced machine tool in history, so complex it’s impossible to use without extensive training from ASML personnel, who remain on-site for the tool’s entire lifespan. Each EUV scanner has an ASML logo on its side, but ASML’s expertise, the company readily admits, lies in orchestrating a far-flung network of optics experts, software designers, laser companies, and many others whose capabilities were needed to make the dream of EUV a reality. (Page 230)
In the U.S., Chiang said, if something broke at 1 a.m., the engineer would fix it the next morning. At TSMC, they’d fix it by 2 a.m. “They do not complain,” he explained, “and their spouse does not complain either.” (Page 232)
By 2020, half of all EUV lithography tools funded and nurtured by Intel were installed at TSMC. By contrast, Intel had only barely begun to use EUV in its manufacturing process. As the decade ended, only two companies could manufacture the most cutting-edge processors: TSMC and Samsung. (Page 240)
Yet demand for chips was exploding. China’s leaders realized, driven by cloud computing, the Internet of Things, and big data, these trends were dangerous. Chips were becoming even more important, yet the design and production of the most advanced chips were monopolized by a handful of companies, all located outside of China. (Page 249)
Across the entire semiconductor supply chain, aggregating the impact of chip design, intellectual property tools, fabrication, and other steps, Chinese firms have a 6% market share compared to America’s 39%, South Korea’s 16%, or Taiwan’s 12%, according to Georgetown researchers. (Page 249)
For advanced logic, memory, and analog chips, however, China is crucially dependent on American software and designs, American, Dutch, and Japanese machinery, and South Korean and Taiwanese manufacturing. (Page 249)
China was disadvantaged, however, by the government’s desire not to build connections with Silicon Valley but to break free of it. Japan, South Korea, the Netherlands, and Taiwan had come to dominate important steps of the semiconductor production process by integrating deeply with the U.S. chip industry. Taiwan’s foundry industry only grew rich thanks to America’s fabless firms, while ASML’s most advanced lithography tools only work thanks to specialized light sources produced at the company’s San Diego subsidiary. (Page 252)
China spends more money buying chips each year than the entire global trade in aircraft. No product is more central to international trade than semiconductors. It wasn’t only Silicon Valley’s profits that were at risk. If China’s drive for self-sufficiency in semiconductors succeeded, its neighbors, most of whom had export-dependent economies, would suffer even more. Integrated circuits made up 15% of South Korea’s exports in 2017, 17% of Singapore’s, 19% of Malaysia’s, 21% of the Philippines’, and 36% of Taiwan’s. (Page 253)
Chipmakers jealously guard their critical technologies, of course, but almost every chip firm has non-core technology in subsectors that they don’t lead, that they’d be happy to share for a price. When companies are losing market share or in need of financing, moreover, they don’t have the luxury of focusing on the long term. This gives China powerful levers to induce foreign chip firms to transfer technology, open production facilities, or license intellectual property. Even when foreign companies realize they’re helping develop competitors for chip firms, it’s often easier to raise funds in China than on Wall Street. Accepting Chinese capital can be an implicit requirement for doing business in the country. Viewed on their own terms, the deals that IBM, AMD, and Arm struck in China were driven by reasonable business logic. Collectively, they risk technology leakage. U.S. and UK chip architectures and designs, as well as Taiwanese foundries, have played a central role in the development of China’s supercomputer programs. Compared to a decade ago, though its capabilities still meaningfully lag the cutting edge, China is substantially less reliant on foreigners to design and produce chips needed in data centers. (Page 260)
China’s leaders had begun implementing economic reforms about a decade earlier, experimenting with letting individuals form private companies as a means of spurring economic growth. Shenzhen was one of several cities selected as a special economic zone where restrictive laws were canceled and foreign investment was encouraged. The city boomed as Hong Kong money flowed in and as China’s would-be entrepreneurs flocked to the city in search of freedom from regulation. (Page 271)
Huawei’s equipment now plays an important—and in many countries, crucial—role in transmitting the world’s data. Today, it is one of the world’s three biggest providers of equipment on cell towers, alongside Finland’s Nokia and Sweden’s Ericsson. (Page 271)
Huawei executives say they invest in R&D because they’ve learned from Silicon Valley. Ren reportedly brought a group of Huawei executives to tour the U.S. in 1997, visiting companies like HP, IBM, and Bell Labs. They left convinced of the importance not only of R&D but also of effective management processes. Starting in 1999, Huawei hired IBM’s consulting arm to teach it to operate like a world-class company. One former IBM consultant said Huawei spent $50 million in 1999 on consulting fees at a time when its entire revenue was less than a billion dollars. At one point, it employed one hundred IBM staff to redo business processes. (Page 272)
Huawei coupled this with a militaristic ethos that the company celebrates as “wolf culture.” Calligraphy on the wall of one of the company’s research labs reads, “Sacrifice is a soldier’s highest cause. Victory is a soldier’s greatest contribution.” (Page 273)
By the end of the 2010s, Huawei’s HiSilicon unit was designing some of the world’s most complex chips for smartphones and had become TSMC’s second-largest customer. (Page 275)
Tesla is also a leading chip designer. The company hired star semiconductor designers like Jim Keller to build a chip specialized for its automated driving needs, which is fabricated using leading-edge technology. As early as 2014, some analysts were noting that Tesla cars resemble a smartphone. The company has often been compared to Apple, which also designs its own semiconductors. Like Apple’s products, Tesla’s finely tuned user experience and its seemingly effortless integration of advanced computing into a twentieth-century product—a car—are only possible because of custom-designed chips. (Page 279)
From swarms of autonomous drones to invisible battles in cyberspace and across the electromagnetic spectrum, the future of war will be defined by computing power. The U.S. military is no longer the unchallenged leader. Long gone are the days when the U.S. had unrivaled access to the world’s seas and airspace, guaranteed by precision missiles and all-seeing sensors. (Page 283)
The Soviet Union could match the U.S. missile for missile but not byte for byte. China thinks it can do both. The fate of China’s semiconductor industry isn’t simply a question of commerce. Whichever country can produce more 1s and 0s will have a serious military advantage too. (Page 284)
DARPA is researching alternative navigation systems that aren’t reliant on GPS signals or satellites to enable American missiles to hit their targets even if GPS systems are down. The battle for the electromagnetic spectrum will be an invisible struggle conducted by semiconductors. Radar jamming and communications are all managed by complex radio frequency chips and digital-analog converters, which modulate signals to take advantage of open spectrum space, send signals in a specific direction, and try to confuse adversaries. Sensors simultaneously powerful digital chips will run complex algorithms inside a radar or jammer that assess the signals received and decide what signals to send out in a matter of milliseconds. At stake is a military’s ability to see and to communicate. Autonomous drones won’t be worth much if the devices can’t determine where they are or where they’re heading. (Page 288)
DARPA is investing in technology that can guarantee chips are tamper-free or verify they’re manufactured exactly as intended. Long gone are the days when the military could count on firms like TI to design, manufacture, and assemble cutting-edge analog and digital electronics all onshore. Today, there’s simply no way to avoid buying some things from abroad and buying many from Taiwan. So DARPA’s betting on technology to enable a zero-trust approach to microelectronics: trust nothing and verify everything via technologies like tiny sensors implanted on a chip that can detect efforts to modify it. (Page 290)
By around 2015, deep in the U.S. government, gears slowly began to shift. The government’s trade negotiators saw China’s chip subsidies as a flagrant violation of international agreements. The Pentagon nervously watched China’s efforts to apply computing power to new weapons systems. The intelligence agencies and Justice Department unearthed more evidence of collusion between China’s government and its industries to push out American chip firms. (Page 296)
A laissez-faire system works if every country agrees to it. Many governments, especially in Asia, were deeply involved in supporting their chip industries. However, U.S. officials found it easier to ignore other countries’ efforts to grab valuable chunks of the chip industry, instead choosing to parrot platitudes about free trade and open competition. Meanwhile, America’s position was eroding. (Page 298)
The entire chip industry depended on sales to China, be it chipmakers like Intel, fabless designers like Qualcomm, or equipment manufacturers like Applied Materials. One U.S. semiconductor executive wryly summed things up to a White House official: “Our fundamental problem is that our number one customer is our number one competitor.” (Page 301)
Huawei’s expansion was a threat. Congress wanted a tougher, more combative policy too. “The United States needs to strangle Huawei,” Republican Senator Ben Sasse declared in 2020. “Modern wars are fought with semiconductors, and we were letting Huawei use our American designs.” 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. The more advanced electronics the country produced, the more cutting-edge chips it would buy, and the more the world’s semiconductor ecosystem would rely on China at the expense of the United States. (Page 314)
Thanks to U.S. pressure, China’s government may provide Chinese chipmakers more support than they’d otherwise have received. (Page 320)
The world produced more chips in 2021 than ever before—over 1.1 trillion semiconductor devices, according to research firm IC Insights. This was a 13% increase compared to 2020. (Page 328)
Meanwhile, U.S. restrictions on China’s access to chip technology demonstrate just how powerful the chip industry’s choke points are. The rise of China’s semiconductor industry over the past decade, however, is a reminder that these choke points are not infinitely durable. (Page 329)
The first is to regain manufacturing leadership, overtaking Samsung and TSMC. To do this, Gelsinger has cut a deal with ASML to let Intel acquire the first next-generation EUV machine, which is expected to be ready in 2025. If Intel can learn how to use these new tools before rivals, it could provide a technological edge. (Page 333)
As it began to reckon with the concentration of advanced chipmaking in East Asia, the U.S. government convinced both TSMC and Samsung to open new facilities in the U.S., with TSMC planning a new fab in Arizona and Samsung expanding a facility near Austin, Texas. These fabs are partially intended to appease American politicians. (Page 334)
TSMC’s chairman is certainly right that no one wants to disrupt the semiconductor supply chains that crisscross the Taiwan Strait, but both Washington and Beijing would like more control over them. (Page 337)
The stronger the PLA gets, the less likely the U.S. is to risk war to defend Taiwan. If China were to try a campaign of limited military pressure on Taiwan, it’s more likely than ever that the U.S. might look at the correlation of forces and conclude that pushing back isn’t worth the risk. (Page 339)
Taiwan produces 11 percent of the world’s memory chips. More importantly, it fabricates 37 percent of the world’s logic chips. Computers, phones, data centers, and most other electronic devices simply can’t work without them. So, if Taiwan’s fabs were knocked offline, we’d produce 37 percent less computing power during the following year. The impact on the world economy would be catastrophic. The post-COVID semiconductor shortage was a reminder that chips aren’t only needed in phones and computers—airplanes, autos, microwaves, and manufacturing equipment of all types would face devastating delays. Around one-third of PC processor production, including chips designed by Apple and AMD, would be knocked offline until new fabs could be built elsewhere. (Page 340)
Other data infrastructure would be hit harder. New 5G radio units, for example, require chips from several different firms, many of which are made in Taiwan. There’d be an almost complete halt to the rollout of 5G networks. It would make sense to halt cell phone network upgrades because it would be extremely difficult to buy a new phone too. Most smartphone processors are fabricated in Taiwan, as are many of the ten or more chips that go into a typical phone. Autos often need hundreds of chips to work, so we’d face delays far more severe than the shortages of 2021. Of course, if a war broke out, we’d need to think about a lot more than chips. China’s vast electronics assembly infrastructure could be cut off. We’d have to find other people to screw together whatever phones and computers we had components for. (Page 340)
After a disaster in Taiwan, in other words, the total costs would be measured in the trillions. Losing 37 percent of our production of computing power each year could well be more costly than the COVID pandemic and its economically disastrous lockdowns. It would take at least half a decade to rebuild the lost chipmaking capacity. These days, when we look five years out, we hope to be building 5G networks and metaverses, but if Taiwan were taken offline, we might find ourselves struggling to acquire dishwashers. (Page 341)
Looking at the role of semiconductors in the Russia-Ukraine war, Chinese government analysts have publicly argued that if tensions between the U.S. and China intensify, “we must seize TSMC.” (Page 343)
Aircraft from the Longtian and Huian airbases on the Chinese side of the strait, from which it’s only a seven-minute flight to Taiwan, not coincidentally in 2021, upgraded these airbases with new bunkers, runway extensions, and missile defenses. (Page 343)
With Soviet Russia and Communist China building industrial-scale militaries, the U.S. couldn’t count on fielding bigger armies or more tanks. It could build more transistors, more precise sensors, and more effective communications equipment, all of which would eventually make American weapons far more capable. (Page 346)
Today, even the Pentagon’s $700 billion budget isn’t big enough to afford facilities for building cutting-edge chips for defense purposes on U.S. soil. The Defense Department has dedicated shipyards for billion-dollar submarines and ten-billion-dollar aircraft carriers, but it buys many of the chips it uses from commercial suppliers, often in Taiwan. Even the cost of designing a leading-edge chip, which can exceed $100 million, is getting too expensive for the Pentagon. A facility to fabricate the most advanced logic chips costs twice as much as an aircraft carrier but will only be cutting-edge for a couple of years. (Page 347)
Industry luminaries from NVIDIA CEO Jensen Huang to former Stanford president and Alphabet chairman John Hennessy have declared Moore’s Law dead. At some point, the laws of physics will make it impossible to shrink transistors further. Even before then, it could become too costly to manufacture them. The rate of cost declines has already significantly slowed. The tools needed to make ever smaller chips are staggeringly expensive—none more so than the EUV lithography machines that cost more than $100 million each. The end of Moore’s Law would be devastating for the semiconductor industry and for the world. We produce more transistors each year only because it’s economically viable to do so. (Page 348)