The chip shortage for carmakers and some other industries has alerted the world to the complexity of the chip supply chain.
A Qualcomm executive recently called it “supply spaghetti,” given the many intertwined piece parts involved in making complex chips that run seemingly every machine on the planet, from smartphones to supercomputers.
Some background
There are many billions of transistors in a single modern semiconductor. Its design can take five to 10 years. Engineers work with machines and software to chart the pathway for commands that must weave their way at the speed of light through circuit gates and along the rails.
In addition, once a chip design gets to a fabrication facility—most are based in Asia—it can take four to six months to produce it, depending on how customized it is. For cars, there can be customization for each car model’s trim specification, which adds to the delays. Car companies also face the added complexity that they may have thousands of components inside their cars, and many rely on legacy chips built on an older 28nm node (10nm or less is the new wave) with older chipmaking machines that are short in supply.
When viewed as a global whole, chip production is indeed a massive technology mess in every sense, something like spilling huge quantities of spaghetti on the heads of car executives across the globe, and even on the heads of their children riding in the back seat.
The issue is complicated by the interdependence that the chip industry has on different regions in the world. Nearly 80% of semiconductor foundries and assembly and test operations are concentrated in Asia, even though the U.S. still has about half of overall chip sales. And for chips smaller than 10nm, South Korea and Taiwan control all production.
RELATED: Could more U.S. chip production lead to price increases?
In the past year, major U.S.-based chipmakers such as Intel have worked with trade groups to push for more domestic chip production. President Biden included $50 billion to bolster chip manufacturing in his $2.2 trillion infrastructure plan, while Intel recently proposed two new chip fabs in Arizona at a cost of $20 billion.
The focus on U.S. chip manufacturing is complicated by an uneasy U.S. relationship with China, partly based on major trade differences over many goods, including chips and agricultural products, but also concerns over charges made by Biden and others about human rights violations at the behest of China’s Communist Party. Biden has also held up China’s growth in technology prowess as a motivation for the U.S. to invest more in technology R&D and development, including chip manufacturing in particular.
SEMI equipment report from 2020
A recent report on 2020 global semiconductor manufacturing equipment sales shows the distinction between China and North America in that sector, in which the U.S. dominates the chip production in North America.
RELATED: Chip equipment sales soar in China, while North America tanks
Last year, sales of chip equipment tanked 20% in North America, while China saw 39% growth. The North American sales were $6.5 billion while the Chinese sales were $18.72 billion, according to the SEMI trade group.
According to SEMI’s market research team based in Asia, China’s growth was driven by both its indigenous chipmakers and international companies with facilities in China, including Samsung, Intel, TSMC and SK Hynix. Out of that massive amount of spending on equipment in China, international companies contributed 40% to 50% over the past three years, SEMI told Fierce Electronics.
Chinese indigenous fabs have also been aggressive in building up their fab capacity since 2018, SEMI said. China’s central and local governments use investment funds and policies to support China’s ambition to be self-sufficient in chipmaking, SEMI researchers noted.
Jack Gold’s take
“China is on a tear as the government there seeks to create a viable and world-class semi manufacturing and design capability that isn’t reliant on foreign companies,” noted Jack Gold, an analyst at J. Gold Associates in comments to Fierce Electronics. “It’s logical they spend on manufacturing equipment as they open up new fabs and expand the ones they have. They want to be less reliant on the rest of the world…and I expect the amount of spend will continue for at least a few more years.”
Gold added that the decline in the U.S. in 2020 after three years of growth “is not all that surprising…There hasn’t been that much expansion in U.S. chip manufacturing.” He noted that Samsung, based in South Korea, and TSMC, based in Taiwan have expanded to the U.S.
Despite the big decline in the U.S. last year, Gold said he expects a significant increase in 2021 with Intel’s promise to invest $20 billion and as the U.S. government pushes to increase U.S. chipmaking, which could cause more investment by U.S. and foreign firms operating in the U.S.
Still, it can take two to three years to build a massive chip production fab. “The results of investment are not immediate new capacity,” Gold said.
Gold joins others in forecasting continued international investment in chip production as well, especially in Asia. U.S. companies often outsource chip manufacturing to China and Taiwan, and Gold predicted there will not be a major shift to bringing that production back to the U.S. anytime soon. “U.S. expansion will be in addition to the current Asia capacity, not really a replacement,” Gold said.
Various studies have forecast that the world will need 50% more chip manufacturing capacity in a decade that all regions currently provide. Boston Consulting has forecast demand to grow 5% annually until then, and up to 10% annually in the next few years.
Leonard Lee’s take
Leonard Lee, managing director at analyst firm NeXt Curve, noted that the 2020 semiconductor equipment numbers confirm that 83% of the spend happened in Asia, giving the region “significant leverage on global equipment providers.” The geopolitical implications are apparent, he said.
“The question for the U.S. administration is how much spending and foundry build out will be required to materially shift the global tech supply chain?” Lee asked. “Will the economics and trade dynamics support the geopolitical objectives?”
China’s lead globally is not a surprise given firms like SMIC based in Shanghai have been investing in fabs and capacity “with continued encouragement and investment from the central Chinese and local governments,” Lee said.
“In light of the increasingly souring relations between the U.S. and China even under a fledgling Biden administration, the Chinese tech economy continues to look domestically for foundry capacity for supply chain security and control.”
Lee said restrictive U.S. policies toward China have not helped matters. He mentioned the Department of Commerce effort to prevent SMIC’s acquisition of ASML EUV machines in late 2019 “accelerated China’s push toward semiconductor self-determination. This push was sustained over 2020 during the pandemic as the U.S. State Department instituted the Clean Network initiative and furthered sanctions against China’s tech industry.”
How the Biden administration handles trade with China and helps bolster the chip industry domestically will be the focus of trade groups and chip executives in the next year. On Monday, Biden said a bi-partisan group of members of Congress agreed with him on $50 billion in federal chip manufacturing investments, but the question remains whether that elected group represents a clear majority in each chamber.
RELATED: Chip shortage: How soon is relief possible?
Parting thoughts
There is some optimism afoot for increased U.S. chip production, even if the short term may be difficult.
“I think we are in good shape with all the political tail winds behind the increasing U.S. semi output,” Gold said. “Of course, it will take some time for the U.S. to catch up, if ever, to what’s taken place offshore from an investment perspective over the last decade or more. The consensus is growing that having local semi making facilities is critical to the future of the U.S. and we simply can’t rely on foreign plants.”
One reason that chip designers in the U.S. have traditionally used Asian fabs was its lower cost, primarily due to lower labor costs, Gold added. “I don’t think the cost differential is anywhere near as great now, given how automated and robotic-powered everything is now,” he said. “Labor costs have become a much smaller component of the cost, so that decision is easier now.”