US Secretary of Commerce Gina Raimondo has made her recommendations for the revitalization of America’s semiconductor industry, a plan that is pivotal to retaining US economic leadership but one that also raises hard questions about government interference in the private sector and potential damage caused to US allies in Europe and East Asia.
In a speech entitled “The CHIPS Act and a Long-term Vision for America’s Technological Leadership,” delivered on February 23, Raimondo spoke about “the incredible opportunity we have as a nation to unleash the next generation of American innovation, protect our national security, and preserve our global economic competitiveness as we implement the historic CHIPS and Science Act.”
Signed into law by President Joe Biden in August 2022, the CHIPS and Science Act allocates US$52.7 billion to semiconductor production and R&D, and authorizes another $24 billion in tax credits for semiconductor production.
In her speech, Raimondo also outlined the Biden administration’s desire for the US to dominate the global tech industry:
I want the United States to be the only country in the world where every company capable of producing leading-edge chips will have a significant R&D and high-volume manufacturing presence.
We will be the premier destination in the world where new leading-edge chip architectures can be invented in our research labs, designed for every end-use application, manufactured at scale and packaged with the most advanced technologies.
This combination of technological leadership, supplier diversity, and resiliency does not exist anywhere else in the world today.
Raimondo’s statements seem to suggest she would like to reverse the globalization of the semiconductor industry and establish a complete supply chain for every product on US territory, regardless of comparative advantage.
Is this a good idea? Morris Chang, founder and former CEO and chairman of TSMC, the world’s leading semiconductor foundry (contract chip manufacturer) and a 25-year veteran of Texas Instruments, wouldn’t think so. Speaking in Taipei in 2021, he said,
If you want to reestablish a complete semiconductor supply chain in the US, you will not find it to be a possible task. Even after you spend hundreds of billions of dollars, you will still find the supply chain to be incomplete, and you will find that it will be very high cost, much higher cost than what you currently have.
It should be noted that most new leading-edge IC designs always have been and are still created in the US.
In 2022, according to market research firms and media reports, seven out of TSMC’s top 11 customers were American (Apple, AMD, Qualcomm, Broadcom, Nvidia, Marvell, Analog Devices and, number 11, Intel), one was Taiwanese (MediaTek), one Japanese (Sony) and one European (STMicro).
And, we might ask, must Sony, the world’s top maker of image sensors, establish R&D and manufacturing operations in the US? Sony has just formed a joint venture with TSMC in Japan.
Outside Japan, Sony Semiconductor has design and development bases in Belgium, France, Finland, Spain, Switzerland, Israel and Taiwan; manufacturing bases in Thailand (image sensor assembly) and China (optical pickups); and sales bases in mainland China, Hong Kong and Taiwan, South Korea, Singapore, the UK and the US (San Jose). Presumably, management has a pretty good idea of what should be done where.
To be sure, Raimondo is not a complete techno-nationalist:
Very importantly, we are not aiming for self-sufficiency or looking to close ourselves off from global markets or competition. And of course, we are eager to continue working with our partners and allies to create diverse, resilient, and sustainable supply chains, write tech standards that align with our values and invest in our shared digital future.
After reviewing the decline of semiconductor manufacturing in the US – its share of global chip manufacturing is down from 37% to 12% since 1990; none of the world’s most advanced semiconductors are now made in the US – Raimondo went on to say:
This manufacturing atrophy has real consequences. For starters, it’s a threat to our national security. So many of our defense capabilities – like hypersonic weapons, drones, and satellites – depend on a supply of chips that aren’t currently produced in America.
That could indeed turn out to be a problem, Raimondo said:
But our dependence on foreign semiconductor supply chains also hurts our economy. In 2001, the US had more than 300,000 semiconductor manufacturing workers. In the past 20 years, we lost a third of those jobs while the global semiconductor industry has more than tripled in size. We sacrificed our manufacturing capacity and workforce in the mistaken belief that we could somehow maintain our technological leadership without them.
Raimondo correctly identifies the link between manufacturing and technological advances and the consequences of rampant outsourcing driven by an obsession with financial returns. Pointing out that “funding in technology hardware makes up only 3% of America’s venture capital, down from 20% in 2005,” she stated:
The brutal truth is that, without manufacturing strength in the US, and the innovation that flows from it, we are at a clear disadvantage in the race to invent and commercialize future generations of technology.
This is not always the case, as the overwhelming dominance of American semiconductor design and software companies demonstrates. However, as a general proposition, it is a correct diagnosis of what has been going on in the semiconductor, consumer electronics, industrial robot and other tech-focused industries for the last three decades.
As US production has been outsourced overseas, the development of some key semiconductor production equipment, flat panel TVs, industrial robots and other products has gone with it.
China, of course, has taken the opposite tack, to its great benefit and to the alarm of the US government. Raimondo noted that “over the last two years, China has produced more than 80% of new global capacity for certain mature chips, and their market share is growing.”
Is her concern justified? Data from Gartner show that among the top 10 semiconductor vendors in 2022 ranked by revenue, seven were American (Intel, Qualcomm, Micron, Broadcom, AMD, Texas Instruments and Apple), two South Korean (Samsung and SK hynix) and one Taiwanese (MediaTek).
Samsung and SK hynix ranked first and third, respectively, while Intel ranked second. Five were design companies that outsource to TSMC.
On the other hand, Raimondo stated that “Taiwan alone produces 92% of the world’s leading-edge chips.”
Defined as ICs made at the 7nm process node and below, that means TSMC. The rest are made by Samsung. Although not the subject of her speech, the key semiconductor industry-related risk to US national security is dependence on TSMC’s factories in Taiwan.
To hedge this risk and rebuild the US semiconductor industry, Raimondo said,
We need the private sector to invest with us, using our $50 billion of public investment to crowd in at least $500 billion in additional funding for manufacturing and R&D.
Specifically, the US will have at least two new large-scale clusters of leading-edge logic fabs, that will have been built by highly-skilled union labor.
Each cluster will include a robust supplier ecosystem, R&D facilities to continuously innovate new process technologies, and specialized infrastructure. Each of those clusters will employ thousands of workers in well-paying jobs.
Additionally, the US will develop multiple high-volume advanced packaging facilities, and become a global leader in packaging technologies.
That’s a tall order, history shows. Judging from the US auto industry, the application of the US-style adversarial relationship between union labor and management could disrupt and increase the cost of the project, and undermine the frontier spirit and entrepreneurial mentality that created companies like Intel, Microsoft, TSMC and Apple.
Already there is a “culture clash” at TSMC’s new factory in Arizona. Last year, EE Times quoted an American engineer working there as saying, “The work culture in Taiwan is really different than in the US. TSMC will have to change to an eight-hour workday five days a week.”
But TSMC engineers often work longer hours and are always on call, on weekends and at night, in case of an emergency.
There has also been an effort to organize workers at Intel’s factories in Oregon in order to “address long hours, pay disparities and job security.” This comes even as Oregon Live reports that “Workers in Oregon’s chip industry earn an average of more than $150,000 annually, according to state wage data.”
On top of that, $550 billion is a lot of money. The Taiwanese, South Koreans, Japanese and Europeans, who are also subsidizing and otherwise promoting their semiconductor industries, might see it as a threat – perhaps as more dangerous than doing business with China.
And there is a risk of oversupply if investment decisions are made by ambitious politicians rather than by experienced, market-sensitive executives.
Raimondo also aims to shore up semiconductor R&D in the US:
The $39 billion in incentives [under the CHIPS Act] will bring semiconductor manufacturing back to the US, but a robust R&D ecosystem will keep it here. That is why we will invest $11B to build a strong semiconductor R&D ecosystem to generate the ideas and the talent we need to support these efforts.
The heart of these investments will be the creation of the National Semiconductor Technology Center. The NSTC will be an ambitious public-private partnership where government, industry, customers, suppliers, educational institutions, entrepreneurs, and investors converge to innovate, connect, and solve problems.
We envision a network of several centers around the country, solving the most impactful, relevant and universal R&D challenges in the industry. Their work – fueled by industry support – will generate new devices, processes, tools, and materials for our manufacturing ecosystem.
So does this mark the advent of an expensive, unfocused, bureaucratic boondoggle or is Raimondo’s plan ultimately sound industrial policy?
The American Semiconductor Innovation Coalition (ASIC) thinks the NSCT can work as long as it can develop and implement a “practical, technical agenda focused on the transition from innovation to commercialization” and “be held accountable for reaching clear and measurable goals.” To this end, ASIC has published detailed recommendations.
Raimondo phrased her bottom line this way: “Here’s the truth: If we don’t invest in America’s manufacturing workforce, it doesn’t matter how much we spend. We will not succeed.”
Tens of thousands of engineers, technicians and scientists must be trained, she acknowledged.
“In the 10 years after Kennedy announced his mission to put a man on the moon, the number of physical science PhDs tripled and engineering PhDs quadrupled,” she said.
That level of commitment and inspiration is needed again. It already exists in China, after all.
Follow this writer on Twitter: @ScottFo83517667