BETA
This is a BETA experience. You may opt-out by clicking here

More From Forbes

Edit Story

Chips Don’t Float: Reviving Semiconductor Manufacturing Alone Won’t Assure America’s Security

Following
This article is more than 2 years old.

The global semiconductor shortage went high profile in 2021 with presidential and industry calls for the revival of American chip manufacturing. But just bringing semiconductor manufacturing back to U.S. shores won’t assure national security or prosperity. The entire microelectronics ecosystem needs to be rebuilt, a task in which America has made little progress.

Microchips provide vital processing, control and memory functions for electronics systems in consumer sectors from phones and automotive to social media. But they’re only part of the ubiquitous silicon infrastructure, the tiny hardware that supports advanced technologies including the radars, electronic warfare, communications, signals intelligence, targeting, navigation, virtual and augmented reality (VR/AR), and autonomous systems that the U.S. defense apparatus relies on.

Microelectronic systems require more; printed circuit boards (PCBs), transistors, capacitors, inductors, resistors, diodes and other components. Much of the broader microelectronics infrastructure is designed in America but almost none of it is made here.

The parts and pieces - and the vital intellectual property they’re imbued with - are made in five countries in Asia. America accounts for just 12 percent of global microelectronics manufacturing. And as any supply chain expert can tell you, what you don’t produce, you can’t control.

China recognizes that reality. “Self-reliance is the foundation for the Chinese nation,” President Xi Jinping said in a March 2021 speech. The Chinese Communist Party is investing over $200 billion, pushing its largest technology firms including Alibaba Group and Tencent into designing as well as producing their own semiconductors and other components.

Among such vital components are integrated circuit substrates (IC substrates), the silicon, carbon or ceramic medium by which microchips are attached to printed circuit boards (PCBs). The U.S. used to lead in IC substrate manufacturing. Those days are gone.

But in the still remote Upper Peninsula of Michigan, a small $40 million-a-year company is one of the few American businesses trying to bring those days back.

Chips Don’t Float

Calumet Electronics is a PCB manufacturer of the sort that used to dot the American landscape, a “build-to-print shop” that can make circuit boards to order for industrial controls, medical equipment and defense applications. These days, it’s seeking to step up — not to build semiconductors but rather advanced IC substrates crucial to the latest defense and commercial systems.

“You can have all the microchips you want,” says Calumet COO Todd Brassard, “but if you can’t attach them to something, how can you use them? Even if we could revive foundries and build microchips in the U.S., we would still have to send them offshore to be assembled into advanced packaging.”

The dwindling number of American microelectronics makers have recently started encapsulating the problem with catch phrases like Silicon Is a System or Chips Don’t Float. Brassard admits he really doesn’t think either slogan hits the nail on the head but they are illustrative of the need to rebuild the entire domestic microelectronics ecosystem, not just semiconductor capacity.

An IC substrate is essentially layers of thin material (forming a baseboard) onto which semiconductor chips are placed. It fixes the chips in a pattern and connects them electrically to each other and to a PCB beneath. The substrate not only supports and connects the chips, it provides channels to dissipate heat.

Advanced IC substrates are critical to the latest generation of defense technologies. They allow for higher densities of chips on a PCB, even allowing chips to be stacked. The more semiconductor chips a PCB has for the same given area, the shorter the distance electrical signals have to travel. That makes for high speed processing of data, referred to as a High Bit Rate.

Very tight spacing between semiconductor chips also reduces electrical power consumption and decreases susceptibility to signal integrity/interference problems, a key quality in defense systems. Asian manufacturers have specialized in “ultra high-density” substrates.

Companies like Japan’s Kyocera and Ajinomoto Group and Taiwan’s Unimicron Technology MU and ZD Tech lead the globe in the production of new advanced microelectronic packaging hardware, including substrate-like PCBs (SLP) and high-density interconnect (HDI) PCBs.

These companies use additive manufacturing processes to turn out substrates for PCBs that can space chips and circuits as close together as 30 microns in hardware like mobile phones from Apple AAPL , Samsung or Huawei. The most advanced can reach 10 microns or less. Most of the remaining American PCB makers etch their circuit boards with spacing rarely below 75 microns.

The resulting “commodity” PCBs aren’t fit for the state-of-the-art defense systems that DoD buys. So the PCBs and substrates inherent in those systems come from overseas.

“Whoever can build the highest density, highest speed systems is going to win,” Brassard asserts.

David Hagerstrom agrees. Hagerstrom is program director of national security and space at Draper (formerly Draper Laboratory), a non-profit advanced engineering firm with a broad technology design and biotechnology portfolio.

The New England-based company takes on custom projects for DoD and other clients that typically require bespoke software and hardware. Draper struggles to find microelectronics hardware in America for its designs.

“I’ve run programs where we needed to find packaging for custom circuits comparable to commercial handheld products,” Hagerstrom attests. “We come up with the design then scour the country for advanced substrates but quite often we’ve had to go offshore for the substrate components for our custom design.”

When it receives foreign-s0urced hardware, Draper is careful to vet it, Hagerstrom says. The company takes time to use analytical tools like high resolution X-ray systems to detect unwanted, nefarious layers in substrates. The problem eased slightly when Draper found R&D Altanova, an advanced PCB, substrate and test equipment maker in South Plainfield, New Jersey.

But in early December R&D Altanova was acquired by Advantest, a leading Japanese semiconductor equipment company. That could be a problem for Draper. Japanese microelectronics firms often shy away from defense-related U.S. customers whose volume demand they regard as too small. Thanks to Japan’s pacifist leanings and constitution they also frequently refuse to sell microelectronics destined for defense applications.

The problem is front and center in Silicon Valley where a new generation of defense tech companies like Anduril are forced to go overseas for the microelectronics they need for the advanced sensing and control systems they design. Anduril founder Palmer Luckey recently discussed the issue in a Reagan National Defense Forum interview.

Calumet has a budding supplier relationship with Northrop-Grumman NOC which cannot integrate its systems without the latest high density substrates and PCBs. Defense firms aren’t alone, Brassard says. “We know from first-hand experience that companies like Microsoft MSFT , Facebook and SpaceX are looking for domestic [substrate supplies] for AR and VR systems. They’re not finding them.”

The fact that American commercial players can’t source microelectronics from our own shores is both a loss to national security and an opportunity. Defense cannot drive the microelectronics market. Its share is under 2 percent of U.S. demand.

And yet, the United States still has the largest market share in sales of broader finished electronics products, at about 47 percent of the global total, according to the Semiconductor Industry Association. If America could couple its lead in design and finished product sales with domestic manufacturing of microelectronic hardware, it could be both less critically dependent on one area of the globe and more prosperous.

That potential reality, along with a dose of self-interest and patriotism, has propelled Calumet to invest its own resources in advanced IC substrate design and manufacturing. It’s a tall order thanks to the capital-intensive nature of microelectronics manufacturing and the loss of American expertise.

“Our trade association thinks it would take someone six years to start a circuit board company from scratch with the expertise currently available,” Brassard observes. “Circuit boards are hard to make. There’s a lot of chemistry, lithography, CNC machining.”

But as the industry has shrunk, as Calumet’s customers have retired or lost experience, it is feeling pressure to offer more services. The company believes advanced substrate production is close enough to what it already does that it can vertically integrate.

“We can put chips on circuit boards. We can start doing some layouts,” Brassard says. We have opportunities on either side of our business as a PCB manufacturer to push into assembly and design.”

Calumet’s management thinks it can do so quickly. In fact, the company says it will have its first substrate prototypes ready in about a month.

“The know-how we have, the prototypes are coming off the line,” Calumet’s CTO, Dr. Meredith Labeau, says. “The struggle is how quick can we get capitalization to scale? Scaling requires money.”

To get this far, Calumet has stood up its own R&D team, employed a full-time designer, bought used tooling equipment and adapted existing equipment to turn out its first prototypes.

“It’s hard to scale on the back of your manufacturing engine which has small margins because your product is considered a commodity,” Labeau adds.

The company is at a crossroads. Calumet estimates it will need $3 million to $5 million to scale to low volume advanced substrate production in nine months. In DoD or large tech company terms, that sum doesn’t even approach pocket change. But for Calumet and other small firms that form the vital backbone of the domestic microelectronics supply chain it’s a real obstacle.

“We have the space,” Labeau says with frustration. “We need the capital.”

Easy Money Across The Water

Finding capital is no problem in Asia, and hasn’t been for four decades, says Chris Toffales, chairman and managing director of CTC Aero LLC, an aerospace and defense consultancy firm. He is also the chairman of a three-year, 60-person microelectronics study commissioned by former Undersecretary of Defense Ellen Lord.

The study not only seeks to make recommendations to address the loss of the U.S.-based microelectronics supply chain but traces its decline. Toffales, who spent a portion of his early career with the now defunct California-based Fairchild Semiconductor ON , says American electronics manufacturing began to change around the late 1970s.

At the time, the U.S. military, particularly the Air Force, was the biggest global customer for semiconductors and microelectronics. The consumer electronics industry had yet to explode but within a year or two, it did.

“At that point, countries like Japan recognized that semiconductors and microelectronics were a highly complex, high margin business with a highly valued, growing global market,” Toffales explains. “They said, ‘We the government of Japan will subsidize electronics businesses like Sony, Matsushita, Hitachi and more.’ The government subsidy model started in Japan in the early 1980s.”

In Asia, it extends to the present. South Korea followed Japan’s lead in the 1990s and Taiwan embraced the direct subsidy model by the early 2000s. China has been doing the same for the last decade.

Back in the 1980s, the Japanese government essentially agreed to finance 30 percent to 40 percent of the cost of constructing foundries and acquiring capital equipment, which cost several billion dollars to get up and running. Non-Japanese companies including American electronics manufacturers and integrators took note. They could build new dedicated facilities in Japan for two-thirds to half the cost of building them here.

“It was a home run,” Toffales says.

Today, China is offering a similar deal. American OEMs like Apple can build subsidized, tax-incentivized manufacturing facilities in China for far less than in the U.S. (typically $15 billion to $20 billion for a foundry). They can count on lower labor costs and Toffales says China dangles its market of 1.5 billion citizens. If that’s not enough, China will deny market access to OEMs who don’t manufacture within its borders.

The strategy has paid off and been paid up. Late last year, media reports claimed that in 2016 Apple’s Tim Cook made a five-year $275 billion agreement with Chinese officials to invest in facilities in China. As part of the deal, Apple reportedly promised to use more components from Chinese suppliers in its devices, collaborate on technology with Chinese universities and directly invest in Chinese tech companies.

“If I were a microelectronics designer, do I really want to build a foundry in the United States when I know I can get one for 50% less in another country?” Toffales asks. “Commercial enterprise will always go where they make the best bucks.”

The lack of competitive incentives in the U.S. saw its electronics manufacturing industry move across the Pacific. Forty years later, Asia controls 75 percent of the microelectronics hardware market. With Asian investment estimated at $500 billion to $750 billion over the next few years, the study concludes that share could be 90% by 2030.

“This is 40 years in the making,” Toffales concludes. “This is a lot of neglect on our part and on our Allies. We need to have another supply chain. We cannot be locked into one source and one region of the world. It’s too risky for both national and economic security.”

What To Do and Who To Do It?

For years, there has been an understanding of the supply chain problem by those in manufacturing and a few in government. In 2016, the Department of Commerce Office of Technology Evaluation, in coordination with the U.S. Navy’s Naval Surface Warfare Center, Crane Division, began an assessment of the industrial base for manufacturing bare printed circuit board products.

The Secretary of the Navy was assigned as the DoD Executive Agent for Printed Circuit Board and Interconnect Technology. The Pentagon appeared to be recognizing the broad microelectronics supply chain issue. But the recognition has not so far translated to action.

While the recently hyped “CHIPS Act” — included in the FY 2021 National Defense Authorization Act — focused chiefly on throwing $50 billion at establishing new semiconductor foundries in the U.S., it centered less on other holes in our microelectronics hardware sector.

“The government keeps talking about ‘chip, chip, chip’ but we need it to look at the entire ecosystem,” says Chris Peters, executive director of the U.S. Partnership for Assured Electronics (USPAE) industry association.

In passing the U.S. Innovation and Competition Act (USICA), which includes federal investments for the domestic manufacturing provisions in the CHIPS Act, the Senate somewhat broadened the potential legislative view of microelectronics but the Bill’s continued stall in the House of Representatives has rendered Congressional initiative meaningless for the time being.

There certainly appears to be an emerging alphabet soup of Pentagon bureaucratic organizations purportedly dedicated to the supply chain problem. At a recent USPAE meeting, DoD representatives included folks from the Office of the Undersecretary of Defense, Research & Engineering; the Defense Microelectronics Cross-Functional Team and the DoD Industrial Base Analysis and Sustainment (IBAS) office among others.

OSD and IBAS did not respond to requests to discuss substrates and the wider microelectronics ecosystem.

What could they and Congress be doing? The microelectronics study suggests some ideas.

“We’re recommending that we offer companies value and the ability to make as much money here in this market as elsewhere,” Chris Toffales says.

America can do that the study says by offering competitive incentives (tax and otherwise) and enlarging the scope of the domestic market by thinking beyond the defense sector.

The U.S. needs to conceive of national security as more than defense activity says a study co-leader, former deputy undersecretary of OSD acquisition and sustainment, Alan R. Shaffer. America’s utilities, medical, financial, transportation, technology and other sectors should be considered part of an enlarged commercial sphere incentivized to support and demand local manufacture of microelectronic components

“We need to factor these into [policy making] to build up a large enough domestic market for which we could attract domestic producers to compete at some level,” Shaffer asserts. “We don’t have that right now. It’s not a good situation.”

Toffales suggests that American businesses like Ford, GM and Stellantis that rely on foreign sourced electronics hardware - now more than ever since the EVs they are determined to build require 2.3 times more microcontrollers than combustion engine cars - would jump at the opportunity (and possibly requirements) to buy American made substrates, PCBs and chips.

“If you don’t think that a lot of these companies aren’t terrified of what’s going on in Asia right now, don’t kid yourself. They are.”

The idea of an emphasis on “pedigreed parts” for the sectors above could kickstart investment in U.S. electronics manufacturing the study contends. Some sort of ISO-type certification of components like substrates could be attractive to OEMs looking to signal their commitment to U.S. safety, security and prosperity. Products with certified onshore sourcing would theoretically be preferred and hence, more highly valued.

Pairing defense with consumer electronics and technology demand could ultimately expand the U.S. share of microelectronics manufacturing to something like 25% the study authors reckon.

Direct public investment will be needed as well, but Toffales and Shaffer are wary of overtly national trade policy or buy-American requirements for fear of a trade war.

“We’re trying to work through some of the steps we take to escalate slowly to [building] a competitive market rather than coming in and mandating a market,” Shaffer says. “I think the free market will still work with big enough incentives.”

Others I spoke with contend the time for such concern has passed. We are effectively in a publicly financed international trade competition now, they argue.

The microelectronics study recommends that the U.S. not try to do it all, instead emphasizing a preference for rebuilding hard-to-replicate capabilities. It estimates that restoring memory chip/semiconductor capacity will take three-plus years. Rebuilding logic-related semiconductor, substrate and advanced packaging capacity could take five to seven years.

Picking investment targets should be a forward looking exercise, Toffales argues. Microelectronics processing technology has slowed down he points out with two dimensional processors about as refined as they’re going to be. The market including the defense primes and companies like Meta (a.k.a Facebook) are moving to 3D processors, a jump ahead.

“Under that condition, there is no world leader today,” Toffales affirms. “Our study says the U.S. should take a leadership role in 3D and advanced packaging.”

The investment envisioned in USICA (which sets aside $10 billion for advanced microelectronics technology development) and other legislation shouldn’t just be tossed at the market. There should performance goals and deliverables required in return Toffales stresses.

“We have to carefully think through where we invest the $10 billion and how do we bring industry to not only match the $10 billion but exceed [that investment] so we have a massive pool of next generation technology innovation here.”

All of the above would be nice to have. But there are no concrete signs such policy or investment will materialize, keeping American companies on the hunt for a few U.S. suppliers or going overseas.

Can You Hear Me Now?

Chris Peters points out that it’s not just specific microelectronic bits and bobs which are difficult to source in the U.S. - it’s also the materials they’re made of. Rare earths, resins, e-glass (a material consisting of numerous extremely fine fibers of glass) and copper are hard to find here.

Ironically, Calumet Electronics was founded after the last copper mines in the Calumet, Michigan area closed in 1968, partly to replace the jobs and careers lost with the closures.

“We’re willing to take some of our profits and do the R&D for what our OEM customers are asking for that they can’t get in the U.S.,” Brassard reiterates. “We are struggling to get anybody in the U.S. government to realize that this is a clear and present danger and provide some funding for it.”

Brassard and Meredith Labeau attended the 2021 Defense Manufacturing Conference in Colorado, a gathering at which most all of the DoD organizations mentioned above were present. The pair met with representatives from the Air Force Research Laboratory which has been supportive of Calumet’s efforts to branch into advanced substrate manufacturing.

Calumet has been selected for Air Force sponsored Small Business Innovation Research (SBIR) contracts Brassard says but ultimately not funded. USPAE’s Chris Peters was at the conference as well, introducing Partnership members like Calumet to the various service and Pentagon housed organizations. Their interest he says is most often related to solving specific electronics problems, not to the entire ecosystem.

That’s clearly a source of frustration for Todd Brassard who noted that the conference featured a raft of speech-making on the need to invest in the supply chain and the value of working with small defense electronics suppliers.

“We’re the kind of small company you’re talking about,” he says. “We’re right here in the room - now. We have the solution to a problem you don’t know you have yet.”

The schizophrenia of the business can be seen in the skies above the Pacific, Brassard says.

“How does sourcing chips from some Intel plant inside the U.S. help a DoD prime contractor who then has to ship them back to Taiwan or China to put those chips onto a substrate and into a package?”

Meredith LaBeau finishes his thought with the addendum, “Then they have to source their PCBs somewhere else and they have to ship those to yet another overseas location to get the [chips/substrate] attached to the circuit board. They call it the ‘FedEx Supply Chain’. It’s crazy.”

As they say, Chips Don’t Float.

Follow me on Twitter