The trade spat between the US and China has rattled companies making a wide range of products, from aluminum to flamethrowers. But if one product has surfaced again and again in this tussle, it’s one we never see yet use all the time: the semiconductor.
Also known as the integrated circuit, or the computer chip, they’re made of tiny blocks of silicon, cobalt, and copper and sit inside almost every electronic device.
When Singapore-based chip maker Broadcom attempted to acquire US firm Qualcomm in a more than $100 billion deal, the Trump administration vetoed the deal citing ties to Chinese firms. Besides Qualcomm, Trump cut off a planned acquisition of Oregon-based Lattice Semiconductor by a little-known private equity firm with ties to China’s government. Meanwhile, Qualcomm is on track to scrap (paywall) a planned $44 billion acquisition of Netherlands-based chip maker NXP after failing to secure approval from Chinese regulators by its July 25 deadline for the deal—the last of nine nations that had to clear it.
Why are the world’s two most powerful nations sparring over semiconductors?
“Semiconductors are arguably humanity’s greatest achievement to date,” says Jimmy Goodrich, vice president of global policy at the Semiconductor Industry Association, a trade group based in Washington, DC. “They’re still central to everything that’s modern or electronic, whether you’re driving a car, or surfing the internet, or using a supercomputer—everything is ultimately based on a semiconductor, and lots of them.”
Why chips matter
While we often hear about the internet or software companies that are wielding more and more influence over society, those companies make their breakthroughs off the advancements of chip makers. Simply put, innovation in semiconductors leads to innovative products.
Second, chips are an economic boon. The US chip industry directly employs about 250,000 people, and generates $164 billion in sales each year. When clusters of semiconductor factories are built, entire sub-industries can evolve around serving them. A manufacturing facility “needs power, it needs water, it needs chemicals, it needs other raw materials, it needs equipment, it needs service,” says Risto Puhakka, president of VLSI Research, a California-based firm that tracks the semiconductor industry. “There is just a huge amount of value created around that factory.”
Third, a small but critical portion of the semiconductor industry has specific applications in the defense sector, for use in things like missiles and radars. Mastery of the semiconductor technology can help ensure that a country’s military technology remains at the cutting edge.
As a result of these factors, dominance of the chip industry can give governments amazing political leverage over other countries. There’s no better example of this than when the US Department of Commerce ordered a ban on sales to ZTE. Without access to Qualcomm’s processors, the phone maker had to halt production and could have gone out of business, if not for a subsequent reprieve from the Trump administration.
Mark Li, who researches China’s semiconductor industry at Sanford C. Bernstein, says that the ZTE incident likely heightened the sense of urgency felt by policy makers in Beijing to ramp up the country’s semiconductor prowess. “They need to make their own chips, so their companies can function without imported chips,” he says.
The semiconductor industry remains one of the few manufacturing-driven sectors where the US remains the global leader. After Texas Instruments’ Jack Kilby invented the integrated circuit in 1958, the US military embraced the technology as the Cold War ramped up, and consumer electronics companies later followed once the personal computer hit the market. Most of the fabrication for the country’s largest integrated circuit makers takes place within the United States. Meanwhile, the world’s most sophisticated semiconductor firms—which include Qualcomm, Nvidia, Intel, AMD, and Micron—all reside stateside.
China, meanwhile, is by far the world’s largest purchaser of semiconductors—but its share of global sales still remains marginal. According to Bernstein Research, China purchased approximately $160 billion in semiconductors in 2016, but sold only $20 billion. In other words, semiconductors are China’s biggest import by value—putting it ahead even of China’s oil imports, according to the South China Morning Post.
Why China can’t crack semiconductors
Given China’s manufacturing prowess in general—it’s the “world’s factory,” after all—its lag in semiconductors seems surprising on the surface. Three of the world’s top-five smartphone companies are Chinese. Factories in southern China can reverse engineer hoverboards and e-cigarettes, and manufacture millions of them on the fly. Why can’t it simply dismantle Qualcomm’s chips, figure out how they’re made, and start producing them en masse?
Experts say it’s because semiconductors are some of the most complicated products to design and produce on earth.
For one thing, according to Goodrich, of the Semiconductor Industry Association, the R&D necessary to make advancements in chip design can span decades. There are no shortcuts to understanding the recipe of a chip. “It can take an American company years just to design one chip. Then let’s say you’ve got 5,000 engineers that can reverse engineer the chip and look at the chip layout—by the time you’ve done that, the American company’s already two generations ahead of you,” says Goodrich.
Beyond designing chips, there are also barriers to manufacturing them. Setting up a fabrication facility can cost $10 billion to $15 billion. Inside those facilities are complex machines designed to carry out etchings and moldings at a microscopic level. For Intel, TSMC, Micron, and other chip manufacturers, “their processes are highly proprietary, and you’re literally dealing with a process that is one atom at a time. That knowledge base is not something that you can just go and copy overnight,” says Puhakka.
Factories also need to crank out these chips quickly and consistently, with as few defects as possible. Making integrated circuits is “highly complex, massive scale, and high precision.” says Li. “You need to produce every day, every month, to a degree where you can support say 1.5 billion smartphones every year.”
Analysts say that China’s semiconductor makers remain broadly five to 10 years behind world-class companies in the US, Korea and Taiwan. Li, for example, says that China’s manufacturers can easily make chips that power the iPhone’s fingerprint sensor—despite being a relatively new feature, the underlying technology isn’t that different from what powers a smartphone’s camera.
However, he adds, producing systems-on-a-chip (SoCs) that power 4G connectivity inside high-end smartphones at scale—like those designed by Qualcomm– remains out of reach for most of China’s top semiconductor manufacturers.
As a result China’s telecom sector looks set to remain hugely reliant on US chips for several years—while chip makers like Qualcomm are hugely reliant on China for their revenues.
Playing catch-up
Under the “Made in China 2025” directive publicized in 2015, China’s government has unveiled plans for dominance in a variety of tech sectors, including integrated circuits. The plan encourages government bodies at the central and local level to boost R&D, in so China can rely on its own companies for core technologies rather than overseas ones.
For semiconductors, China has attempted to compensate for its lag through three tactics, none of which have been particularly successful. Throughout the past few years, as Chinese semiconductor companies have attempted to purchase overseas counterparts, US regulators or the administration have blocked them due to national security concerns.
China’s government has also encouraged foreign chip companies to form joint ventures with domestic counterparts, in hopes that know-how and intellectual property from the overseas firms will get transferred to the Chinese ones. Earlier this year, for example, Chinese investors acquired a controlling stake in the local venture of SoftBank-owned chip design firm ARM. Qualcomm, Intel, and AMD also have partnerships with state-affiliated Chinese companies. Yet joint ventures have not done much yet to meet expectations, according to analysts, because foreign companies tend to share obsolete technology with their Chinese partners rather than state-of-the-art IP.
The third tactic is simply spending lots of money. In 2014 Beijing launched a 139 billion yuan (then $21.8 billion) fund explicitly for the chip sector, and it’s currently raising an second 300 billion yuan ($47.4 billion) fund for the same purpose. There are also several local government funds directed at improving a city or province’s chip industry.
Given that acquisitions and JVs have not been effective, moving forward, China will likely be forced to rely on self-funding R&D to improve its semiconductor industry. Can money alone change the industry? Experts are split.
Barry Naughton, an economist at the University of California, San Diego, argues that China will need more than just cash to start making competitive semiconductors—it will need better engineers, managers, and bureaucrats capable of catching up in a sector that other countries have amassed decades of knowledge in. “There are a lot of high tech areas out there, and China sort of has ambitious plans in each one,” he says. “They might think semiconductors are more important than anything else. They’re prepared to throw lots more money at it, but I don’t think the correlation with money is that strong.”