It’s time to realize that our industrial policy for semiconductors is failing. It was built on three pillars: subsidies to encourage re-shoring of fabrication, export controls that made trade in chips a function of our political alliances; and an attempt to maintain a competitive advantage in the digital by decoupling from China. None of these are working.
Intel has become the poster child for what this new, mercantilist approach to digital industry really means. The company lost over $4 per share; its Cumulative Annual Revenue Growth for the past 5 years is -6%. Intel’s stock price went from $62 in 2021 to $18 recently. The passage of the CHIPS Act in 2022 did nothing to alter its downward trend. The possibility of the U.S. government acquiring a 10% share of the company, on the other hand, raised it up to around $24, but its share value is still less than half of what it was. The market reaction to a 10% USG stake is a bet that Intel needs a government bailout to survive. It also means it will become an agent of politics and policy, not U.S. competitiveness.
The CHIPS act threw $55 billion at an assortment of semiconductor companies – most not based in the United States – in the hope that this would somehow make the United States the world’s primary fabrication site. Let’s overlook the fact that fabrication plants are on a 10-year build cycle, meaning that any alleged dependency is going to be around for a long time. And let’s forget about the cost disadvantages. The more important fact is that the U.S. already does control the apex of the semiconductor value chain: the designs and intellectual property that account for most of its value. The CHIPS Act was part of a panicky strain of economic nationalism which focuses exclusively on equipment manufacturing and ignores every other part of the value chain. The intent of the Act seems weirdly implausible when stated directly: we want all the same companies and all the same technology to pick itself up and move its fabrication activities to a different, more expensive place on the globe — just because it’s our place.
The drafters of the CHIPS Act must have been in a terrified state of mind. Our national security, they thought, depended on reversing a long-term, highly successful industry tactic of specializing in areas of comparative advantage. They felt this way even though US-based companies account for the largest share of the global market by sales (close to 50%), according to SIA data. The US industry cemented its status as world leader precisely because it specialized in chip design and outsourced the labor-intensive and hardware-intensive parts of the process to other countries. Intel is the company that tried to remain vertically integrated. Its fate – loss of market share, negative profits, lagging production technology, and now begging to become an industrial vassal of the U.S. government – proved that outsourcing was the better way to go. Specialization on design made the Americans better than everyone else in the world. Specializing on fabrication made TSMC better than anyone else in the world. Both countries – and consumers – benefited enormously from this division of labor. Additionally, the fabrication industry spun off specialized segments, such as ASML EUV lithography, that also became concentrated in a particular company or location.
What made America strong was not that we fabricated the chips – it was that we designed them, controlled the intellectual property, outsourced the fabrication and assembly, and sold them to everyone else in the world.
Misguided policy makers started to unravel this successful recipe by telling us three stories: 1) our national security depended on re-shoring, so as to get fabrication as far away from China as possible; 2) we had to block China’s access to advanced chips – one of our most successful exports to one of our biggest markets – else we will lose a war to them; 3) we cannot use or buy Chinese electronics or digital products because they are a military adversary and will only be used to sabotage our critical infrastructure, compromise our privacy and security or to brainwash us. China, China, China; fear, fear, fear.
Our national security mavens proposed to win a technological competition by eliminating the efficiencies of a globally distributed division of labor centered on the U.S., rolling up into a protective ball and moving toward autarky. Which never works. And Intel proves it.
The travails of Intel underscore one of the chief dangers of state-led industrial policy. It inevitably becomes mixed with political favoritism and security nationalism. Our government is betting on a loser, but it’s a politically influential loser. Despite its glorious past, Intel is now just a company that failed to adapt to a very competitive and dynamic market. It is the Kodak of transistors. Unlike Kodak, however, it has chosen to rely on protection and handouts to survive. The news that Masayoshi Son’s SoftBank was investing in Intel was just what one would expect: Son has a long record of picking losers, from Sprint to WeWork to FTX. Intel is going to become the equivalent of British Steel, a state-owned enterprise that we keep alive for political reasons.
Leadership in the chip industry or AI industry is not owned by any one country. The science and technology of ICT benefits from transnational research collaboration and the free flow of ideas and capital. Leading firms must have global markets to achieve economies of scale and to recruit the best talent Every major supplier has a multinational supply chain, and the headquarters of these companies are in different countries. The whole idea of making the chip industry the basis of a geopolitical/military strategy – rather than a market competitor – was brain dead.
Dominance in this industry comes from winning in the marketplace. A global marketplace. There is no way to exclude China from trading in a global market anymore than there is any way for them to exclude us. The biggest beef we should have with the Chinese is access to their home market, which is unduly and unfairly restricted. When the problem is stated that clearly, however, it is obvious that the problem we face is essentially a trade issue, not a geopolitical conflict. So why are we undermining the WTO? American products win in the Chinese market when they are superior and Chinese consumers are allowed to access them. We will never negotiate market access, however, if we deem our trading partner an adversary. Why should any country open their market to us when we insist on telling them that the purchase of digital products from another country is a dire national security threat? Why would they welcome interdependence with our products when we openly use them as choke points (like export controls in chips) to disrupt or cripple Chinese companies or the Chinese economy? If we make trade in digital goods and services a geopolitical power struggle, it’s no wonder an authoritarian regime seeks self-sufficiency and market protection.
President Trump is said to be an astute business bargainer, somebody who knows how to get a good deal. The problem with the current moment is that Trump needs to start bargaining over money, not geopolitical power. The geopolitical game is one the producers and consumers of semiconductors can only lose.
The post Intel’s Glorious Past and Current Funeral appeared first on Internet Governance Project.
