Fabsolute Advantage

Name-brand US chip companies like Intel, Nvidia, and Advanced Micro Devices (AMD) conduct research for and design the chips that most US consumers ultimately use. But the actual production of computer chips has been outsourced overseas, with the United States producing just 12 percent of chips globally — down from 37 percent in 1990. The companies hired to make the chips in their fabrication (“fab”) facilities are incredibly few in number, with just one making nearly all the fanciest microprocessor chips: the Taiwan Semiconductor Manufacturing Company (TSMC).

TSMC is the eleventh-most-valuable company on the planet, worth around $550 billion. And if you believe the business press, it’s unlikely to lose its title as chip manufacturing king any time soon. The Wall Street Journal reports that TSMC’s

technology is so advanced . . . that it now makes around 92% of the world’s most sophisticated chips, which have transistors that are less than one-thousandth the width of a human hair. . . . Most of the roughly 1.4 billion smartphone processors world-wide are made by TSMC.

So complete is the company’s dominance that observers have noted Taiwan’s “silicon shield,” thought to deter military action by China lest it disrupt the chip supply.

The reason for TSMC’s near-monopoly is that semiconductors, the Journal explains, have become “so complex and capital-intensive that once a producer falls behind, it’s hard to catch up. Companies can spend billions of dollars and years trying, only to see the technological horizon recede further.” The economies of scale have reached truly stunning proportions in the chip industry: a modern semiconductor factory now costs up to $20 billion to build. That stratospheric price tag keeps out everyone but the already-biggest players, even as the gigantic volumes of chips produced drag their per-unit cost below what competitors can match.

None of this is the result of the “invisible hand”: the Taiwanese government, putting the East Asian state-led development model into practice, poured subsidies into the industry for decades, footing the bill for over half its initial funding investment. (Notably, the other oligopolist of the fanciest chips is Samsung, with a similar record of heavy support from the Korean state.)

Efforts by chip designers like Intel to catch up with their own fabs are expected to only bear fruit many years and billions of dollars from now. Even companies making far less sophisticated analog chips, which do simpler tasks like managing phone displays or battery chargers, are struggling to manage the surging demand. Texas Instruments, the market leader for these workaday processors, has a yearlong backlog and is building three new plants in Texas. Apple, one of the world’s largest consumers of chips great and small, can’t keep up either. The Journal reports that

Apple has used its might to invest billions of dollars into suppliers to guarantee space on their assembly lines so that the iPhone has the parts it needs. . . . But even with those steps, there is only so much that can be done.

Bargaining Chips

Products as technical as modern smartphones and electric cars rely on more than semiconductors. But the whole sector of related commodities and industries is still shaped by the power of the giant chip companies downstream. Consider the shortage of substrates, the relatively simple materials made of copper wire compressed into industrial resin, which are essential because the “ultrathin” wiring emerging from microscopic chips can’t handle direct connection to the soldered wiring on the circuit boards that hold the chips in place.

The business press reports that, much as chip fabs like TSMC and Samsung are thriving because chip designers like Intel outsourced fabrication, “chip companies have largely outsourced substrate production to focus on improving chip performance rather than low-cost items with relatively meager returns.” Further, “chip companies have long pressured substrate suppliers to keep prices low. . . . Those dynamics have limited investments in adding substrate production capacity.”

Many substrate makers also got burned in the move to mobile in the last decade after anticipating continued PC market growth. Now, the chip makers are placing orders far earlier and paying in advance, “so that substrate companies have ample cash to build more factories. Some are committing to buying the entire supply of new production lines to give their suppliers confidence to invest.” Despite substrates being far simpler to produce than the chips they hold in place, substrate plants still cost $1 to 2 billion to put up. The business press reports that the small oligopoly of Pacific Rim firms producing them is “in a rare position of strength,” with prices rising.

Even related ancillary industries are seeing similar tightness, partly due to the pandemic but also due to years of industry concentration. Take multilayer ceramic capacitors (MLCCs), which store tiny amounts of energy so chips and other components have it in precisely the right places. Some analysts call MLCCs the “rice” of the electronics industry, due to their staple role as tiny essentials — 5G phones have over a thousand of them. While far easier to manufacture and thus not subject to the dizzying levels of concentration seen in semiconductors, a small enough number of East Asian firms dominate the industry that COVID shutdowns in certain major plants have already tightened this industry too.

Fabricating Consent

In the United States, a bipartisan group of lawmakers have proposed a bill to plow a quarter-trillion dollars into subsidizing domestic chipmaking, advanced research, AI, and quantum computing. The measure’s fate is unclear, but much as with Taiwan’s support for its own chip industry, the New York Times relates that “whether Congress approves billions of dollars in new funding . . . appears likely to determine whether an investment like Intel’s is a one-time occurrence or a trend.” So much for the innovations of the private marketplace.

In the meantime, Samsung and GlobalFoundries have announced new factories in the United States (although as with Foxconn’s heavily subsidized LCD plant in Wisconsin, nice words don’t guarantee investments).

More domestic fabrication could be a boon for US workers if they’re able to organize — particularly in logistically crucial plants, since chips have become as pivotal to the circulation of capital as ports and warehouses.

But sourcing chips domestically instead of from Taiwan or Korea won’t change capitalism’s profit-hungry hunt for scale and efficiencies that leave one or two private companies completely in charge of one of the world’s most important industries. Time for the world’s working class to cash in their chips.