Kyodo via AP Images
A Taiwan Semiconductor Manufacturing Company facility in Hsinchu, Taiwan
For months, executives at Taiwan Semiconductor Manufacturing Company (TSMC) insisted that skilled American workers weren’t available to build their chip factories, that this shortage was delaying construction of their first U.S. fabrication plant in Arizona, and that this exposed the key weakness of the Biden administration’s industrial-policy agenda, too concerned with creating win-win scenarios and not focused enough on removing barriers to building. The usual suspects in the media gleefully parroted TSMC’s claims as proof of their belief system.
Within a few days, this entire edifice fell down.
TSMC abruptly announced that pilot production would actually begin this month, and chips would start rolling out in Arizona by the end of the year, several months ahead of schedule. The timeline, former Prospect scribe Lee Harris quipped, was being “preponed.” A few days later, TSMC and the Commerce Department announced a deal under the CHIPS Act that would deliver $6.6 billion in grants to the manufacturer and another $5 billion in loans, not only to stand up the two Phoenix plants already promised, but to build a third that will make two-nanometer chips, some of the most technologically advanced in the world.
The carping from TSMC, in other words, was a very practiced lie, a way to gain leverage in negotiations with the government over the final terms of the deal. As soon as it was struck, the delays and problems magically vanished. Even the most strident narrative-setters about the administration messing up industrial policy had to admit this. Noah Smith, who a year ago used TSMC’s alleged challenges in Arizona to confidently call America a “build-nothing country,” sheepishly admitted that “we always need to be prepared to adjust our beliefs quickly as new information comes in.”
I’m not convinced he will, considering that TSMC’s gambit was rather obvious from the start. As the Prospect reported nearly a year ago, TSMC’s use of thousands of non-union subcontractors and foreign labor was actually creating the delays it claimed came from U.S. workers. A deal between unions and the company was struck in January, and within a matter of months the timeline accelerated, which we only learned about after TSMC’s double game to extract subsidies.
The $39 billion CHIPS Act, of course, was created to get companies to invest in semiconductor capacity in the U.S., and to bring those high-value jobs, critical manufacturing infrastructure, and supply chain diversity to our shores. TSMC’s promised investment in the U.S. has risen to $65 billion, far outstripping the grants and loans they’re receiving. Joelle Gamble, former chief economist at the Department of Labor, said at a conference in Washington convened by the journal Democracy on Tuesday that this demonstrated how public investment can actually crowd in private capital rather than crowding it out. “To meet the scale of the moment, you need to shape new markets,” Gamble said.
As soon as the deal was struck, the delays and problems magically vanished.
Overall, the various grants and loans to TSMC and others put the U.S. on track to produce 20 percent of advanced logic chips by 2030. Contrary to the ideology that America cannot build, it turns out that a little seed money goes a long way.
“I think that this is an example of a supply-side liberalism that builds power,” said the Roosevelt Institute’s Todd Tucker, referencing an earlier piece from myself about the goals of industrial policy. He pointed out that $50 million of the loans will go toward workforce development and training, and that there are commitments embedded in it requiring TSMC to end its anti-union activities at the Arizona plants. This shows how domestic union jobs can be an asset, not a liability, to U.S. semiconductor production.
“You had [TSMC founder] Morris Chang saying it was sad that Biden was talking about unions, and that it’s not how they do business,” Tucker said. “Now you have joint press releases with the building trades saying they will work together as partners. It’s clear they had a workforce shortage problem and the union’s going to help them solve it.”
That’s not to say this transition will be perfectly smooth. In particular, boosting TSMC’s dominance in the global semiconductor sector, and by extension its biggest customer Apple, could have negative ramifications, as the Prospect has reported. The grants, loans, and tax breaks TSMC can expect to receive for the three plants could total up to $20 billion, going to a very profitable market leader that critics say abuses its power.
In response to the TSMC announcement, Todd Achilles, who authored a recent American Economic Liberties Project report on the industry, explained that the deal “changes none of the anti-competitive practices that TSMC and Apple use to block new entrants,” which include exclusive deals, rollups of competitors, and patent abuse. “Leading-edge chip supply will remain brittle, mostly offshore, and controlled by one dominant firm.”
It would certainly have been better to address those practices in the context of a CHIPS Act grant. The problem, of course, is that semiconductor markets evolved through conscious choices made for decades. The U.S. has less leverage with a mature and consolidated industry than with emerging cleantech companies that really depend on government largesse for their survival. “This is the difference between trying to onshore a global oligopoly versus standing up infant industries,” Tucker said. “But if you onshore part of the production, you’ve solved a big part of the problem.”
Related to this, TSMC is a Taiwanese firm, and the question arises of why the U.S. should be subsidizing it. Again, the company’s dominance in the marketplace is a major factor here: If you want the state-of-the-art fabs in America right now, you are going to have to work with TSMC. And it’s better to have them working in America than elsewhere.
Fortunately, TSMC isn’t the only recipient of CHIPS money; Intel actually received a larger grant, other companies have taken funds, and independent foundries could be beneficiaries in the future. Only $16 billion of the $39 billion fund has been drained, with the major players out of the way. As Ronnie Chatterji, former White House CHIPS coordinator for implementation, explained at the Democracy conference, “We have to make sure that we don’t use the industrial policy, in all its good, to reinforce current market dynamics.”
THE USE OF GRANTS AND LOANS TO STAND UP domestic semiconductor production offers an alternative path to relying on private financing burdened too heavily by high interest rates set by the Federal Reserve. Nonresidential construction has actually fallen the past two months, despite an early construction boom from the Inflation Reduction Act (IRA). The Fed has become a major impediment to industrial policy, as Chatterji acknowledged.
That’s why the CHIPS-related loans, nearly doubling the public investment, are a critical way to circumvent the dampening power of the Fed. Another industrial-policy announcement last week stood up the Greenhouse Gas Reduction Fund, a kind of “green bank” embedded into the IRA, using $20 billion for a national financing network to loan to cleantech companies. The lower-cost loans these green banks will provide are a manifestation of French President Emmanuel Macron’s musings about a “green interest rate” that’s lower than the going rate.
“These are opportunities to make a huge difference,” Chatterji said. But he didn’t go as far as to call it a substitute for good central-bank policy. “You need the Fed to play a huge role in these projects getting done. Without them, things will move a lot slower.”
In effect, the people looking to the administration’s so-called “everything bagel” approach to industrial policy as the source of delay, whether through promoting unionization or child care or other policies, were looking in the wrong place. Monetary policy is a greater concern, and federal credit programs can only go so far to counteract it.
But when it does, the benefits, as we’re seeing in Arizona, are profound. And while TSMC’s bargaining strategy was somewhat infuriating, in the end they realized that building industrial capacity and creating good union jobs were not in conflict whatsoever.
