A 58.7% operating margin is not what a competitive market looks like. It is what a toll booth looks like.
TSMC's sixth straight record quarter is being reported as a triumph of engineering and foresight, and the engineering is real. But when a manufacturer keeps more than 58 cents of operating profit on every dollar of revenue, and its customers are, in the company's own words, in intense competition for access to its capacity, the market is not allocating abundance. It is rationing a chokepoint. Two-thirds of that revenue now comes from high-performance computing, the hardware under the AI boom. Whoever controls that hardware controls who gets to build the future and at what price.
The question the earnings release never asks is who built that chokepoint. TSMC's rise rests on decades of Taiwanese state investment, a public research apparatus, cheap and subsidized land, power and water, and a geopolitical bargain in which an entire island of 23 million people lives under military threat partly because its most valuable export is something Beijing, Washington and Silicon Valley all want. Taiwan's public carries the risk. Shareholders collect the margin.
Now the same logic is being exported to Arizona. The company says it will invest $265 billion there and is eyeing Texas. Governors are preparing ribbon-cuttings. Washington, under the CHIPS Act, committed billions in grants and loans to TSMC's American expansion. I supported the premise: a country that cannot make its own advanced chips is a country at the mercy of someone else's shipping lanes. But supporting industrial policy is not the same as supporting this deal.
Here is where my own side should be embarrassed. Democrats wrote a landmark subsidy and then, too often, handed out the money as though the goal were to be liked by corporations. Public money went out with thin strings: some labor standards, some buyback restrictions, little claim on the upside. Then the Trump administration, hardly a friend of working people, converted CHIPS money into an equity stake in Intel. Whatever you think of that deal, it exposed what we failed to demand. If taxpayers underwrite the risk of building the most strategically valuable factories on earth, they should own a piece of the return, the way any venture investor would insist. The Left used to understand that. Public investment that produces private fortunes with no public share is not industrial policy. It is a gift with a patriotic bow on it.
Then there are the workers. Arizona's chip boom has come with disputes over safety, over the use of imported labor and over whether the communities hosting these fabs get stable, well-paid, union-protected jobs or just construction booms that fade. A company with a record NT$763 billion quarter can afford to answer those questions with binding commitments, not press releases.
The defenders of the status quo will say profits fund the next generation of chips, that margins reflect real scarcity and genuine skill, and that taxing or conditioning this success risks killing the golden goose. It is a serious argument, and nobody should want to wreck the most advanced manufacturing capability in human history. But the goose already has public money in its nest. Scarcity is not a law of nature here. It is the result of policy choices about who builds, who funds and who benefits.
So make the choices explicitly. Take equity or warrants in exchange for subsidies. Tie every dollar to enforceable local hiring, wage and safety standards. Fund competing capacity so no single company on a single contested island is the sole gatekeeper of the AI economy. And stop treating record profits as proof that everything is working. Sometimes they are proof that someone has been allowed to charge whatever the rest of us will pay.
