Technology Leader of the Year: The U.S. Semiconductor Industry
Factory closings, labor issues and shipping bottlenecks, all sparked by the COVID pandemic, continue unabated, and semiconductors continue to top the list of critical components in short supply for many manufacturers.
If there was a recurring narrative in 2022 it was the ongoing attempt to account for this shortage and reshore supply back to the United States. Passage of the CHIPS (Creating Helpful Incentives to Produce Semiconductors) and Science Act, signed into law by President Biden on August 9, was supposed to accomplish that.
So, much as we recognized COVID-19’s role in the forced adoption of automation in 2021, IndustryWeek’s 2022 Manufacturing Technology Leader of the Year goes to the industry’s biggest headache—the semiconductor industry that so dominated manufacturing technology headlines this year.
The semiconductor supply chain tendrils out to hundreds of industries for whom chips are critical manufacturing components—automotive, consumer electronics, IIoT sensors, PCs and servers just to name a few. Factories go idle when chips aren’t available. The United States military increasingly depends on semiconductors for its most advanced weapons systems, so chip shortages also become national security issues, especially when the chips are sourced from abroad.
To make matters worse, solving the semiconductor shortage is a solution measured in decades. It may take 10 years before the United States is only positioned to take a global leadership position in semiconductor manufacturing, once the infrastructure exists. That makes aggressive, immediate steps to reshore chip manufacturing even more important.
That said, the industry’s massive spending and decisive moves only provide a partial solution to solve today’s supply chain challenges. The investments tackle one-half of the semiconductor market—expensive, complicated, high-end chips for computing and storage. Simpler processors needed by automakers and other industries to handle audio systems or manage flows in pumps—those will remain scarce, despite the high-profile groundbreakings. There are many industries, like automotive, that may have little to celebrate from this year’s new plant announcements.
It’s a start, and at least Washington and industry leaders are on the same page, but these high-profile steps aren’t likely to provide enough supply chain relief and stem the continuing economic damage.
Stitches or Band-Aids?
Almost two-thirds of the world’s semiconductor supply come from Taiwan, an island nation slightly larger than Maryland that sits less than 100 miles east of mainland China. A single company, Taiwan Semiconductor Manufacturing Co. (TSMC), alone manufacturers 55% of the world’s chips. Around 90% of the world’s most advanced chips specifically hail from Taiwan. Maritime shipping problems during the pandemic hit the semiconductor industry particularly hard.
Goldman Sachs in April 2021 listed 169 industries affected by the shortages. The highest-profile examples hail from automotive, with Ford, GM, Honda, Toyota and Volkswagen, among others, all reducing or halting production for lack of chips. The semiconductor industry responded with a series of massive investments in new chip manufacturing plants.
Texas Instruments and Samsung announced plans in 2021 to build new fabs in the U.S., to the tunes of $30 billion and $17 billion respectively. This year, Intel announced a $20 billion investment for a new fab site in Ohio and Micron announced a $100 billion investment for a new fab in New York State.
Click here to see IndustryWeek's regularly updated list of major semiconductor plant and support facility announcements.
This year’s announcements generated additional fanfare, owing to their connection to the CHIPS and Science Act. The legislation created a $52.7B fund for grants, loans and other incentives, new tax credits for fabs investment in the U.S. and $170B for R&D programs.
Between five new fabs and federal funding, will we see the beginning of the end of the semiconductor shortage? It depends on which industries you’re talking about.
High-End Gains and Low-End Losses
The chip market roughly breaks down into two categories of chips. “High-end” or advanced chips largely for computation (read: PCs) and data storage (read: servers), and “low-end” chips for less-demanding applications like automotive hardware, consumer electronics and IIoT sensors; i.e., “mature technologies.”
Micron primarily manufacturers memory, a chip component but not a chip itself, so its new fab could hypothetically serve both sides of the semiconductor market. Other recently announced fabs, however, are squarely aimed at the high-end market, not the mature technologies.
“These fabs won’t really address demand for lower-end microcontrollers, logic chips and other smaller, less expensive semiconductors,” says Glenn O’Donnell, VP, research director at Forrester. “This ‘low end’ of the market remains in seriously short supply.”
ABI Research Vice President Malik Saadi says backlogs for some parts are stretching into 2024.
“The immediate need is really to create more fab capacity to address the long-tail chipset shortage (sensors, MCUs, RF, etc.) and cater for the needs of a multitude of end-products that rely heavily on this type of chip,” Saadi continues.
A core problem, says Bharat Kapoor, partner and global lead of Kearney’s industrial redesign practice, PERLab, is that chip makers focus on where they have competitive advantages and high profit margins. It’s a classic manufacturing issue, companies want to add value to products by being critical suppliers of unique components, not commodity players selling large volumes of things that almost anyone can make.
“An iPhone 14 will only run on the latest 5nm chip which only TSMC can make, no one else can. So, TSMC is incentivized to make more capacity [for high-end chips] and can demand high prices.” Kapoor says.
If It Doesn't End Bottlenecks, What Does the CHIPS Act Do?
“If you look at the CHIPS Act of America Fund, the report states clearly that the department of commerce (DOC) funding incentives for chipset manufacturing is at $39 billion, of which only $2 billion (or 5% of total incentives) dedicated to mature semiconductor technologies,” says Saadi. “The CHIPS Act is mainly designed to promote the competitiveness of the USA in the cutting-edge semiconductor business, not for addressing the current supply shortage addressing mature semiconductor technologies mainly.”
The CHIPS Act passed Congress with bipartisan support that suggests why the legislation focuses so heavily on the cutting-edge: The American military increasingly depends on high-tech solutions like unmanned vehicles and smart weapons. Taiwan controls 90% of advanced chip production. China claims Taiwan as its own territory. It's a national security supply chain nightmare waiting to happen.
Even if the CHIPS Act doesn’t offer any short-term relief, it deals with a long-term problem that needs to be addressed. The legislation also bolsters R&D in the United States. Our defense technology will only get more sophisticated and innovating here also serves national security interests.
But how much help does the CHIPS Act actually represent?
“Building new fabs is insanely expensive — upward of $20 billion for an advanced fab and $2 billion minimum for a less advanced fab. The big three fab operators collectively committed to hundreds of billions of U.S. expansion. CHIPS Act funding won’t make much of a dent here,” wrote O’Donnell in August. “As this money gets divided across its many recipients and spread across five years, disbursements will be in much smaller slices.”
During a presentation at IndustryWeek’s 2022 Manufacturing and Technology Show in October, Intel Principal Engineer Paul Schneider explained why chip making is so costly. The process of layering microscopic circuits on top of each other is more like making a pizza than carving a tablet, he explained. Silicon wafers move from station to station in fabs, gaining a layer of pathways each time, with dozens of layers added to each chip. Each station is a machine costing at least $100 million and about the size of a school bus.
For the companies that will benefit from the construction of these new fabs, relief is still years away. Not only are fabs expensive to build, they take a long time to get into production.
“Expect it will take new semiconductor fabrication plants at least 4 years from date of groundbreaking to be in full production, and that’s assuming everything goes according to plan,” says Kapoor. “In general, it takes two-plus years to build a fab, assuming fab equipment is available on a timely basis, and another one-year-plus to get it producing at scale.”
“This process could take even longer [than 4 to 5 years] in the case of building wafers used for high-performance computing applications,” says Saadi.
“We won’t see the fruits of these new ‘fabs’ until 2024 through 2026,” says O’Donnell. “It takes a LONG time to build one of these facilities.”
Removing Taiwan from the Supply Chain
For the immediate future, then, we can expect the supply of chips for mature technologies to continue hailing from Taiwan. Prioritizing reshoring high-end chips makes sense if national security concerns are driving this effort, as China continues to view Taiwan with an eye toward annexation. This bipartisan concern arguably ended the logjam that stalled the CHIPS Act in Congress.
Will we ever see a day when all semiconductor production, not just high-end chips, shifts out of Taiwan to the Americas and Europe?
“TSMC is going to remain at the leading-edge of semiconductor manufacturing for the foreseeable future,” says Kapoor. “Their existing dominance in producing the most advanced chips, partnered with a new geographic diversification strategy, and supported by recent U.S. chip export guidelines, will keep them at the front of the race while Samsung, Intel and others continue to chase them.”
There may be more hope for the future, however.
“In 15 years, the next gen fabs will be highly automated and could be operated from anywhere in the world where skills are available,” says Saadi. “Automation will enable Europe and North America to lower the cost of operating these fabs and become extremely competitive against low labor cost regions like Taiwan and China. All indications point to the fact that Europe and the USA are positioned to lead the 2030s era.”
As O’Donnell sees it, “We won’t stop sourcing chips from Asia, but we will get more from other places as those sites go live. Globalization remains a good thing. Too much of a good thing is a bad thing, which is precisely what happened with semiconductor supply.”
The steps taken by the semiconductor industry in 2002 won’t address the shortages for some time and even then not for all the industries suffering from chip shortages. But the collective, decisive action by multiple companies to get the process started with high-end chips and R&D demonstrates enough foresight and resolve to celebrate the semiconductor industry as the IndustryWeek Manufacturing Technology Leader of the Year.
