Special report: Questions swirl over subsidies for chip industry
The US may never catch up in the chip-making race, but the country may be about to spend $50 billion anyway
When a massive container ship called the Ever Given got stuck sideways in the Suez Canal for nearly a week this March, it caused a traffic backup in oceangoing commerce, costing the world about $60 billion, by one estimate.
After days of salvage experts trying to dig the ship out, the Ever Given was ultimately freed on March 29, thanks to a big assist from a higher-than-usual tide. The Ever Given’s saga was the latest reminder of how the world economy can be held hostage by a single point of failure in a global supply chain that is more precarious than most people knew.
The production of tiny semiconductors, one of the most critical products in the world, is likewise a chokepoint because nearly 90 percent of them are made by a handful of companies in Asia. A bipartisan chorus in Washington — including President Joe Biden — says American dependence on just these few sources for its economy and its military is dangerous. An ongoing shortage of chips — due partly to a pandemic-driven surge in computer usage — that has hobbled the U.S. automotive industry had strengthened their case. The Ever Given crisis fortified it further.
“As the global economy grows more interconnected, it is imperative that the U.S. reestablish leadership in the strategically important semiconductor industry,” California Democratic Rep. Doris Matsui said in a statement. “The future of innovation in 5G networks, healthcare, and computing will hinge on the availability of state-of-the-art semiconductors from trusted sources.”
But rebuilding the U.S. chip-making industry will not be easy or cheap. The United States is far behind in the race and may never catch up, some say. And yet Washington is poised to spend as much as $50 billion in the next few years, according to a legislative proposal Biden unveiled Wednesday as part of his infrastructure package — all without first developing a strategy and a plan for achieving it, the critics argue.
Supporters of the new subsidies counter that the issue has been studied to death and that so-called CHIPS Act legislation passed last year in the fiscal 2021 defense authorization act is a great start to solving the problem.
The debate is about to heat up.
“It should be a national priority to level the global playing field and strengthen U.S. chip production and innovation,” David Isaacs, the Semiconductor Industry Association’s vice president of government affairs, said. “That’s what folks on the Hill, in the Pentagon and in the White House are recognizing.”
Demand for chip manufacturing is expected to grow 56 percent in the next decade, according to the association.
“The sole question is: Is that growth going to happen in the United States, or are we going to lose more ground and be more reliant on East Asia?” Isaacs said.
‘Road to hell’
The bipartisan CHIPS Act, written by Matsui and others, authorizes grants, research programs and creation of a public-private consortium. But no money was appropriated for it, and authorities for new tax incentives that could cost nearly half the $50 billion bill have yet to be enacted.
Texas Republican John Cornyn, one of the co-sponsors of the Senate’s version of the CHIPS bill, stressed at a March 16 Finance Committee hearing that the pandemic has only underscored the vulnerability of the U.S. supply chain, from personal protective equipment to semiconductors.
“China’s building 17 foundries as we are thinking about building one or having one built in Arizona,” Cornyn said.
Critics say unanswered questions remain even after the law’s enactment, including: What conditions should and will Washington attach to its money? Is this just the beginning of an ongoing, expensive subsidy program? Might America’s tax money be better spent focusing on helping Americans win the race to build the next generation of chips, not the current one? And should foreign companies benefit from these subsidies?
Most fundamentally, some analysts said, the United States lacks a comprehensive strategy that touches on multiple policy areas.
Victoria Coleman, the Air Force’s chief scientist, said too much emphasis in Washington’s chip debates has been put on inputs of dollars and not enough on desired outputs and strategy.
Coleman used to work for Intel, ran the Defense Advanced Research Projects Agency and has written multiple reports on microelectronics for the Defense Science Board, a Pentagon advisory group.
She said what is needed before big bucks start to flow is, first, a set of measurable goals — more quantifiable than just saying America must regain its dominance in the field — and, secondly, a detailed plan to get there.
“We loosely speak about leadership in semiconductors, but not exactly what this means and how we would know we’ve achieved it,” Coleman said. “This is a classic case of ‘the road to hell is paved with good intentions.’”
Lawmakers realize that a strategy is needed. The CHIPS Act requires that one be developed by a panel of the National Science and Technology Council.
The Biden administration, for its part, has launched reviews of critical supply chain issues, including one focused on key materials such as semiconductors. That review, due June 4, may lay the groundwork for a more comprehensive approach to the issue.
Senate Finance Chairman Ron Wyden, D-Ore., stressed at the March 16 hearing there is bipartisan support for bolstering domestic manufacturing of semiconductors.
“Americans don’t roll out of bed in the morning without flipping some switch or checking some device that relies on semiconductors,” Wyden said. “It is a recipe for trouble when one single pandemic, natural disaster, or terrorist attack can sever brittle supply chains, hobble the economy, threaten our jobs as well as put at risk our national security.”
But the national strategy has to address multiple policy areas, including immigration, export controls, education, diplomacy and more — and many of these areas are contentious with little room for bipartisan compromise.
“It is not a one-shot-and-we’ll-fix-it problem,” said Frank Kendall, a former Pentagon acquisition chief, in an interview.
Long, complicated and painful
The most salient reality about the current challenge is how far behind U.S. manufacturers are, both in capacity and sophistication, in the global semiconductor race.
If U.S. chip makers were to net more subsidies, they would still have to deliver the goods — in every sense. They would have to catch up in a race they are losing badly and create products that customers want to buy in great numbers. Only companies that are earning well can afford a technological arms race in semiconductors.
The basic measure of feature size on a chip is nanometers, or billionths of a meter — an incomprehensibly small measure. A piece of paper is 100,000 nanometers thick. The smaller the features — or the fewer the nanometers — the more powerful the semiconductor.
The Taiwan Semiconductor Manufacturing Company, or TSMC, is reportedly poised to produce 3 nanometer chips later this year.
Samsung of South Korea is at the 5 nanometer level.
Intel, the American pacesetter, has had trouble getting 7 nanometer features, and other U.S. companies do not come close to that.
While the measurements are infinitesimally small, the gap between U.S. producers and rivals is big. Intel is even outsourcing some of its chip fabrication to the more capable TSMC.
Eric Schmidt, the former Google CEO who chairs the congressionally created National Security Commission on Artificial Intelligence, told the Senate Armed Services Committee last month that the road to catching TSMC, let alone surpassing them, will be “long and complicated and painful.”
The CHIPS Act, he said, “is a very good first step but it’s not enough.”
TSMC is the pacing company, he said, because of “year after year of precision and learning and proprietary innovation and so forth — something which is very hard.”
The AI commission, which Schmidt co-chaired, urged that the United States should at least aim to stay two generations ahead of China.
Kevin McGinnis, a director of research and analysis at the commission, said in an interview that its other key goal is ensuring America has multiple semiconductor manufacturers on U.S. soil.
Skating where puck is going
Bryan Clark, a defense technology expert with the conservative Hudson Institute think tank, said in an interview that while he supports federal subsidies for expensive production facilities called “fabs,” he has yet to see a strategy for how to do it wisely.
The U.S. government needs to direct its investments to the next generation of technology, he said. Trying to catch up to TSMC and Samsung and their government backers, who continue to invest enormous sums of funds and all their national pride in the outcome, and relax environmental and labor standards in the process, is not likely to succeed, he said.
“There’s just no way you’re going to win,” he said.
He recommends instead focusing on strengthening a strength — the design of chips — and breaking ahead in some of the newer ways of producing next generation chips, rather than subsidize creation of the current generation.
Similarly, Chris McGuire, another director of research and analysis at the AI commission, said it would be prudent to focus federal funds on a facility optimized for 3 nanometers or lower, not just producing the current models.
“We need to skate to where the puck is going, not where it is now,” McGuire said.
A senior administration official said it is important to ensure U.S. strength in not just leading edge but also more pedestrian types of chips — for instance, the sort that are in short supply in the automotive sector today, which includes vehicles made for the U.S. military.
The CHIPS Act would support all of the above.
Allies, partners and rivals
The CHIPS Act would benefit not just U.S. chip companies but also companies based in countries that are U.S. allies or partners — such as TSMC and Samsung — to the degree they invest on U.S. soil.
TSMC and Samsung are interested in moving some of their factories from Asia to less risky parts of the world, such as the United States, several observers said. Makers of all manner of digital devices are likewise keen to see geographical diversification among the builders of their components.
The market is as global as it gets, and U.S. companies also invest significantly overseas. Intel, for example, has chip factories in Israel and Ireland.
Other questions have been raised about the CHIPS Act.
The law authorizes the Commerce Department to allocate up to $3 billion per project not just to accomplished chip fabricators so they can build expensive fabs, but also potentially to those involved in testing, packaging, workforce development, site development, research and more.
Who can get the grants is also broadly cast: those entities that have a “demonstrated capability” to perform such work.
Several observers wonder if the Commerce Department has the chops to execute billions of dollars in grants and other investments in this area.
The Commerce Department is “going to have to now come up with something they haven’t done very often: come up with a program,” said Jeff Rittener, the chief government affairs officer at Intel, one of the companies that could benefit from any new subsidies. “How are they going to administer this? How are they going to determine what programs or what projects will actually qualify for the grants and so on? So I think there’s a bit of work that the agencies are going to have to do.”
Editor’s note: This is the second in a three-part series looking at the federal government’s race to reinvigorate U.S. chip manufacturing, which would be a boon for the semiconductor industry. Part 1 focused on the scramble for dollars. Part 2 is a look at legislative proposals in Congress. And Part 3 will explore the Pentagon’s struggle with chip security.