Gordon Moore, one of the founding fathers of Silicon Valley, died on March 24 at the age of ninety-four. He co-founded Intel in 1968 and grew it into one of the most powerful companies on the planet as its chief executive and chairman. Moore’s Law—his prediction that the number of transistors in a circuit would double every twenty-four months—was pivotal in putting chips in every electronic device and laid the groundwork for personal computers, smartphones, and artificial intelligence.
But according to Intel’s January earnings call, the company is far from its heyday under Moore. Intel’s profits fell by 60 percent in 2022 while revenue fell by 20 percent, leading Moody’s, Fitch, and S&P to downgrade its credit rating. The company has started the process of laying off thousands of employees and has cut its employee base pay by 5 to 10 percent. Intel CEO Pat Gelsinger blamed “persistent macro headwinds” for the collapse, though the company’s inability to deliver new products on time and its severe overestimate of demand for personal computers are also responsible. Intel’s recovery plan is largely dependent on it receiving billions in federal subsidies from the CHIPS and Science Act, which is likely as Intel was the principal lobbyist for the bill.
But why should Washington save Intel? The answer most frequently provided by policymakers is that the U.S. produces no advanced chips, putting it at risk of losing a conflict with China if Taiwan is unable to ship chips across the Pacific. In reality, this is a canard. Peter Wennink, CEO of Dutch lithography giant ASML, dispelled this myth in December, saying that “it is common knowledge that chip technology for purely military applications is usually 10, 15 years old.” America’s warfighting capability is not meaningfully undermined by its lack of production of advanced chips as even platforms like the F-35 fighter jet use only legacy chips.
Another justification for directing subsidies to Intel is the idea that American firms should regain leadership in chip manufacturing to promote U.S. economic competitiveness. As President Joe Biden said when he signed the CHIPS Act into law, “the future of the chip industry is going to be made in America” by “American companies.” This helps explain why Intel may receive upwards of $3 billion from the Commerce Department for its two plants in Ohio, while U.S. factories of foreign firms that have met ambitious technology targets like the Korean memory chip maker SK Hynix will receive less.
Though industrial policy will undoubtedly increase U.S. production of chips, the future of the semiconductor industry is going to be made in East Asia—not America. Taiwan Semiconductor Manufacturing Co. receives over half of all global orders to produce chips, which has enabled it to perfect its manufacturing processes and mass-produce logic chips that are two generations ahead of Intel’s chips. Samsung, SK Hynix, and Tokyo-based Kioxia control 70 percent of the market for memory chips; Intel sold much of its memory business to SK Hynix in 2021. The CHIPS Act will likely increase U.S. semiconductor production from 10 percent of global output to just 14 percent.
Even with billions in government subsidies, Intel is unlikely to regain the commanding position Moore helped it gain. According to semiconductor industry analyst Jonathan Goldberg, Intel will not be able to increase production sufficiently to poach clients from TSMC until the end of the decade, by which time its cash reserves will be depleted. Gelsinger has sold six of Intel’s non-core businesses since he became CEO in 2021, a shrewd move to save cash in the near term but a long-term gamble that Intel’s few remaining products will be dominant even as it innovates at a slower pace than TSMC and Samsung.
Washington’s emphasis on subsidizing American firms to produce on American soil represents a misunderstanding of semiconductor industry dynamics. American companies already capture half of global semiconductor industry revenue and are leaders in key areas including machine tools, design automation software, and process control. Though U.S. firms produce fewer chips than TSMC and Samsung, they still control many choke points in the semiconductor supply chain, allowing Washington to slow China’s production of advanced chips through export controls. Whether a company is headquartered in the United States is a red herring in an industry as globalized as semiconductors; Idaho-based Micron, for instance, has 75 percent of its employees outside of the United States.
A better alternative would be to Americanize foreign companies by enticing them to make good on their promises to invest in the United States. While Samsung has said it may invest $195 billion in Texas over the next two decades, companies rarely deliver on lofty headline investment numbers. TSMC chief financial officer Wendell Huang explained why on a recent earnings call, stating that the costs associated with constructing factories “can be 4 to 5x greater for U.S. fab[s] versus a fab in Taiwan.” This is part of the reason why TSMC intends to continue producing its most advanced chips exclusively in Taiwan. As former Obama administration official Kevin Xu has persuasively argued, Intel’s “core is rotting” while companies like SK Hynix and Samsung are ready to make the United States a leader in semiconductor production again. The Commerce Department should allocate funding to persuade foreign industry heavyweights to make chips in America rather than trying to bail out Intel.
In his 2023 State of the Union address, Biden remarked that “outside of Columbus, Ohio, Intel is building semiconductor factories on a thousand acres—a literal field of dreams.” But a closer look reveals that its field of dreams is fallow.
Kevin Klyman is a technology researcher at Harvard’s Belfer Center and a former fellow at the United Nations Foundation and the AI lab of the UN Secretary-General. Twitter: @kevin_klyman.
Image: Nor Gal / Shutterstock.com