Taiwan Semiconductor Manufacturing Company fabricates the majority of the world’s most advanced logic chips. A single facility in Hsinchu holds more leverage over American military readiness, consumer electronics, automotive production, and artificial intelligence than any tariff schedule Congress has debated in a generation. The United States does not own that facility. It does not control it. And it sits ninety miles from the coast of a power that has made its acquisition a stated national objective. Call that a supply chain vulnerability if you want a polite word for it. I prefer the honest one: strategic exposure.
A republic that cannot produce the physical substrate of its own computation is not an industrial nation in any meaningful sense. It is a client state wearing the clothes of a great power. Semiconductors are not a consumer product that can be reshored at leisure after demand signals justify the investment. They are the precondition for every other productive capacity the United States wishes to maintain. Defense electronics, precision guidance, advanced manufacturing, energy grid control, communication infrastructure, and the entirety of what we now call artificial intelligence run on chips. The country that designs them but cannot fabricate them at scale has handed its adversaries a single point of failure in American civilization.
There is a coherent tradition in American political economy, forged in the earliest years of the republic, that understood this problem before the word semiconductor existed. That tradition held that a nation’s productive independence was the precondition for its political independence. Dependence on a foreign power for essential manufactures was not a market outcome to be celebrated; it was a strategic liability to be corrected by deliberate national policy. The instruments were tariffs, credit directed toward productive enterprise, and the organized will of government working alongside private capital to build what the market alone would not build fast enough or in the right place. The tradition was not hostile to commerce. It was hostile to the kind of commerce that left the nation unable to defend or reproduce itself. It remains right.
The CHIPS and Science Act of 2022 was a partial acknowledgment of this. Federal commitments of roughly $52 billion in direct incentives and an investment tax credit aimed at pulling advanced fabrication back to American soil were significant by the standards of recent industrial policy. TSMC has broken ground in Arizona. Intel has announced expansion in Ohio and Oregon. Samsung is building in Texas. These are real. They are also, measured against the scale of the problem, inadequate without follow-through that Washington has so far failed to guarantee. TSMC’s Arizona facility, when complete, will produce chips at a 3-nanometer node, competitive with leading-edge global production. But completion timelines have slipped, skilled labor is scarce, and the full second and third phases of that investment depend on conditions, including continued federal support and a stable permitting environment, that are not secured. The diagnosis is not that the policy was wrong. The diagnosis is that the republic has not yet decided whether it is serious.
The named enemy here is not China alone, though China’s semiconductor ambitions under its own national programs, including billions directed through state funds toward domestic chip production, constitute a direct challenge to American technological primacy. The deeper enemy is the financialized worldview that governed American corporate strategy for forty years and treated manufacturing as a cost center to be optimized away. American semiconductor firms became design houses. Fabrication was offshored to Asia because Asian governments subsidized it and American quarterly earnings models could not absorb the capital cost of fabs that take four to five years to build and ten years to pay back. The market was not wrong by its own logic. The market’s logic was simply indifferent to national security. That indifference has a price, and the bill is coming due.
The prescriptions follow from the diagnosis. First, the federal government must complete what it started. CHIPS Act funds must move quickly to facilities that are actually building, not to corporate announcements. Disbursement tied to construction milestones, domestic content requirements, and restrictions on recipients expanding advanced capacity in China should be enforced without the waivers that lobbyists are already seeking. Second, the United States needs a sustained workforce development pipeline specific to semiconductor fabrication. Technicians who run deposition equipment and photolithography tools do not emerge from four-year liberal arts programs. Community colleges, technical institutes, and direct partnerships between funded fabs and regional training programs require federal coordination and money. Arizona State University’s partnership with TSMC is a model, not yet a system. Third, the permitting and environmental review process for semiconductor fabrication plants must be reformed to match the pace of geopolitical competition. A fab that takes seven years to permit and three years to build is not a national security asset; it is a national security aspiration. Congress should establish a dedicated fast-track authority for critical industrial facilities analogous to what it has done in the past for energy and defense infrastructure. Fourth, export controls on advanced chip-making equipment, which the Biden administration tightened in 2022 and 2023 and which the current administration must sustain and extend, are not sufficient by themselves but are necessary. Preventing ASML extreme ultraviolet lithography machines from reaching China buys time. It does not build American capacity.
Time bought is only valuable if it is spent building. The republic has a window. TSMC in Arizona, if fully realized across its planned phases, represents a capability that would have been unthinkable under the offshoring consensus of twenty years ago. But windows close. The semiconductor industry moves on capital cycles that do not pause for political hesitation, and the moment American policy signals ambivalence, the investment calculus tilts back toward Asia. History does not reward nations that identified their strategic vulnerability, announced a correction, and then lost patience before the concrete dried.
A country that outsources the fabrication of its own logic to a contested island and calls that globalization has confused efficiency with sovereignty, and it will learn the difference at the worst possible moment.