Americas

  • United States

Asia

lucas_mearian
Senior Reporter

About that ‘alarming’ chip shortage — don’t count on new fab plants to help quickly

news analysis
Jan 26, 20229 mins
Computers and PeripheralsCPUs and ProcessorsIntel

A US Commerce Department report says the computer chip shortage is even worse than thought; at one point in 2021 there was just a five-day supply worldwide, and there's no sign the situation will improve anytime soon.

AMD 3rd Gen Ryzen & Navi chips
Credit: Adam Patrick Murray/IDG

The global semiconductor shortage made worse by the coronavirus pandemic and supply chain issues shows no signs of improving as industries worldwide have had to halt production while waiting for processors.

A report from the US Department of Commerce released this week revealed an “alarming” shortage of computer chips at a time when the median demand for them was as much as 17% higher in 2021 than 2019.

“This is a major supply and demand mismatch,” the report stated.

Commerce Secretary Gina Raimondo said the lack of chip inventory leaves auto manufacturers and other chip users with “no room for error.”

“It’s alarming, really, the situation we’re in as a country, and how urgently we need to move to increase our domestic capacity,” Raimondo said while presenting her agency’s findings.

A recent surge in coronavirus cases in Malaysia, Vietnam, and the Philippines has further exacerbated problems in an already hobbled computer chip manufacturing and supply chain industry.

Last week, the White House held a summit to bring together semiconductor manufacturers and buyers to hash out solutions for the problem. A major sticking point is the demand today for older chips, not the state-of-the-art semiconductors being made by the world’s largest chip makers — TSMC, Intel, and Samsung.

The shortage began pre-pandemic

The problem actually began before the global pandemic hit in 2020. Older semiconductor fabrication plants were already at maximum capacity, according to Alan Priestley, a vice president analyst at Gartner Research. “Covid exacerbated the problem because all the demand forecasting for the industry was thrown into the air,” he said.

Last year, the dearth of computer chips forced auto manufacturers to close down manufacturing and cut production by as much as 7.7 million cars; that, in turn, led to a massive vehicle shortage just as nations began lifting widespread quarantines, allowing people to travel again.

Other industries stopped production, too even as the consumer electronics market took off during the pandemic. Companies and their employees increased their purchases of desktops and laptops for home-office use, and consumers leaned into purchases of a wide varierty of devices such as televisions, gaming systems, headphones, and other electronics equipment.

“All market forecasts became skewed and so manufacturers had to shift their manufacturing capacity to meet the new demand,” Priestley said. “Then in 2021, when things began to recover, industries that weren’t buying products came back on line, and manufacturers had to tell them to wait – they weren’t making those products at the time.

“The car industry was a prime one. They said give us [semiconductors], and the manufacturers said we’re not making them at this time. You’ll have to wait,” Priestley said. “The chip manufacturing isn’t instant. It takes three or four months to begin to manufacture a chip, let alone to have it ready to ship.”

Those chips then have to be embedded into the various systems that make up any given product, such as an automobile or its various subsystems — and that means shipping chips to Asia, where most product manufacturing occurs.

In September, the Department of Commerce sent a survey to semiconductor manufacturers and buyers of computer chips, getting in return 150 responses from nearly every major chip producer and many consumers of the technology.

RFI respondents graphic US Department of Commerce

Survey respondents broken down by role in the supply chain and, for semiconductor consumers, by industry.

The survey showed that at its worst point in 2021, there was less than five days of semiconductor inventory, down sharply from an average 40 days’ supply in 2019. “These inventories are even smaller in key industries,” the report said.

Since the semiconductor shortage began, manufacturers increased plant production capacity dramatically, operating at a rate of more than 90% utilization of their facilities. That high production rate is unusual because semiconductor machinery requires regular maintenance and uses massive amounts of power.

Quarterly Fab Utilization bar graph Semiconductor Industry Association

Semiconductor manufacturers increased plant production, often operating at a more than 90% utilization rate.

The bottom line: there is not enough manufacturing capacity to meet the spike in demand that preceded the COVID-19 outbreak.

Increased demand for older processors to support new technologies, such as electric and self-driving vehicles — and smart technology that embeds sensors in manufacturing equipment, buildings and products to gather data and automate systems — were already placing a strain on the semiconductor industry. Those systems don’t require state-of-the-art chips, meaning the current shortage doesn’t primarily involve the latest GPUs and CPUs built on 7-nanometer (nm) technology, according to Priestley.

Instead, the dearth in chip capacity is mainly for older semiconductors built on legacy logic chips, including, for example, 40nm, 90nm, 150nm, 180nm, and 250nm nodes. Those older transistors are ancillary to primary computer processers, but are necessary for things such as computer displays, mobile phone RF components, analog operations, and system power management. All the latter functions are necessary in the production of computers, medical devices, broadband systems — and automobiles.

“Unfortunately, these are the things you need to build products. You can talk about the latest generation microprocessor or application processor in a smartphone, but a lot of other chips go around it,” Priestley said. “Delivery of a new laptop today is delayed because of display controllers. They don’t need the smallest-size transistors.”

And, because existing semiconductor factories are at capacity, producing more of the older chips isn’t likely to happen any time soon. More to the point, many factories have retooled their machinery to produce newer chips that were in demand prior to the pandemic; they’re not likely to retool to current demand for older processors.

The global automotive industry has, in particular, felt the reverberations, according to Priestley. In the US, where automotive manufacturing is a major component of the national economy, the federal government responded the last year with legislation and new programs aimed at increasing domestic semiconductor manufacturing. Such programs attempt to curb future disruptions to the supply chain by reducing reliance on semiconductor fabrication plants in Asia.

Most semiconductors are made outside the US

Over the last 30 years, the US share of global semiconductor production has fallen from 37% to just 12%, according to figures published by the White House last week. Today, three-quarters of semiconductor production occurs in East Asia. And, 90% of the state-of-the-art semiconductors, such as 7nm-process wafers, are manufactured in Taiwan.

In June, the Senate passed the US Innovation and Competition Act (USICA), which the House and Senate are still working to finalize. It includes full funding for the CHIPS for America Act; that would provide $52 billion to spur private-sector investments.

“The reality is Congress must act,” Raimondo said. “Every day we wait, we fall further behind.”

But for a semiconductor industry that spends as much as $20 billion to bring a new manufacturing facility online, $52 billion is drop in the bucket, Priestley said.

The lack of semiconductor supply didn’t hurt chip manufacturers. The global market was particularly strong in 2021, with semiconductor revenue increasing 25% year-over-year to $583.5 billion — the first time it passed the $500 billion mark, according to Gartner. The growing 5G smartphone market was a significant driver of demand, as was the continued strength of the worldwide PC market.

It’s important to note that the growth is not due to high demand alone. The chip shortage led to logistics and raw material price increases, which drove up semiconductors’ average selling price (ASP). All 10 of the top semiconductor vendors saw revenue growth from 2020-2021, with some hitting growth rates upward of 50% for the year, according to Gartner.

Manufacturers look to boost US production

Semiconductor companies have poured more money than ever before into expanding capacity. In its 2021 report, the Semiconductor Industry Association forecast semiconductor industry capital expenditure (capex) would reach close to $150 billion in 2021 and beyond $150 billion in 2022. In comparison, before last year, the industry never spent more than $115 billion on annual capex.

intel ocotillo fab 42 1 Intel

Intel’s Ocotillo campus, which covers approximately 700 acres of land in Chandler, Ariz. Last year, the company broke ground on two new manufacturing facilities on the campus — named Fab 52 and Fab 62.

In October, 2020, Intel opened its newest semiconductor fabrication plant, Fab 42, in Chandler, Arizona. In March  2021, Intel announced it would invest another $20 billion to build two more plants there as part of its Integrated Device Manufacturing plan (IDM 2.0), a long-term strategy to boost production and meet demand for state-of-the-art semiconductors.

Some new fabrication plants are expected to come online as early as the second half of this year, but most are still two to three years away from opening.

Intel CEO Pat Gelsinger said Intel plans to boost its production in order to become “a major provider of foundry capacity in the US and Europe to serve customers globally.”

“Intel is the only company with the depth and breadth of software, silicon and platforms, packaging, and process with at-scale manufacturing [that] customers can depend on for their next-generation innovations,” Gelsinger said in a statement.

Last week, Intel also announced plans to invest $20 billion to build two new semiconductor manufacturing factories in Ohio. To support development of the new site, Intel pledged an additional $100 million toward partnerships with local universities to build a pipeline of worker talent and bolster research programs in the region.

In November, Samsung announced it is spending $17 billion to build a new semiconductor factory in Taylor, Texas. That plant will focus on processors to support mobile, 5G, high-performance computing (HPC) and artificial intelligence (AI). That same month, Texas Instruments announced it would spend $30 billion for a new plant in Sherman, Texas. And, Cree plans to spend $1 billion to expand a current plant in North Carolina.

Gartner’s Priestley said he does expect semiconductor supply to improve throughout the year ahead, partly because demand for many electronics is beginning to show signs of slowing. But as 5G communication ecosystems begin rolling out and the computer cloud industry continues to grow, demand is expected to surge again.

In the meantime, just as with the pandemic, businesses and consumers will simply have to learn to live with a shortage of some electronics-based goods.

“We’ll get used to it,” Priestley said.