Chip shortage costing billions

by Jacob Stoller

Auto plants shutting down have created some of the biggest disruptions to Canada’s economy during the Covid-19 pandemic. Unlike incidents at the outset of the pandemic, the issue was not a shortfall of demand but of supply – in this case, of the microelectronic components that comprise the brains of today’s vehicles.

The disruption is expected to cost automakers US$110 billion this year in lost revenue, according to consulting firm Alix Partners.

Most experts expect the problem to persist into 2023. “While the industry has taken steps to address near term constraints it could still take a couple of years for the ecosystem to address shortages of foundry capacity, substrates and components,” said Intel CEO Pat Gelsinger in an online presentation at the Computrex trade show in Taipei.

The shortage of a specific type of chip is affecting automotive manufacturing. The onboard computers are designed and prototyped by the automotive manufacturers and contracted out to foundries that can produce complex chips at high volumes.

Foundry capacity

What’s in short supply is foundry capacity. Very few companies have these capabilities – nearly all global production comes from Korea-based Samsung and Taiwan-based Taiwan Semiconductor (TSMC).

Securing production capacity at these plants, therefore, is a critical point in the automotive production cycle. Manufacturers enter into negotiations for allocation of wafer starts, or the number of silicon wafers per week the foundry will commit to in production.

The more complex chips, such as the brains of an Automatic Driver Assist System (ADAS), could involve twelve or more layers of masking and etching and weeks of production time. Producing these at high volume requires extensive setup and planning.

Foundries schedule production slots so as to minimize setup times and maximize utilization. “There are hundreds of process elements that can be different from one component to the next,” says Ken Morris, director of partnerships at Next Generation Manufacturing Canada (NGen), an organization promoting advanced manufacturing in Canada. “The foundry tries to line them up so there is as little change as possible between the wafer starts.”

Cancelled orders

At the outset of the pandemic, automotive manufacturers, seeing a downturn in vehicle sales, sat down with chip manufacturers to cancel orders. Foreseeing a boom in consumer electronics as people started working from home, the foundries moved on.

“Initially, they will accept your cancellation because they have other customers”, says Morris. “So they ramp up their foundries for laptops. But when the automotive customers return to negotiate capacity, they are told that the lead times are five to seven months.”

The auto industry found itself in the unusual position of being a relatively small customer – it provided only three percent of TSMC’s 2020 sales compared with 48 percent for smartphones. Consequently, the OEMs had little leverage.

Becoming more resilient

Not all automotive tier-one companies were affected by the chip shortage. In a surprising announcement in April, Toyota revealed that it had a four-month stockpile of chips and no immediate plans to cut production. The company credits this move to improved supply chain management practices developed in response to the Fukushima earthquake.

“We are now able to make assessments of alternative products in a speedy matter,” said Toyota’s CFO Kenta Kon in a media briefing. “That is one of the factors [enabling] us…to mitigate the impact of semiconductor supply shortages.”

Toyota, interestingly, is widely known for pioneering the just-in-time approach, which uses low inventory as a means for reducing lead times and improving efficiency. Less known is that the company puts considerable emphasis on maintaining close ties with all of its suppliers. This, combined with improved information systems, gives the company early warning of risks such as the potential impact of a chip shortage.

Visibility

Visibility is clearly becoming more important as global circumstances become less certain. “End-to-end supply chain and real-time inventory visibility were always critical, even in a just-in-time scenario,” says Andrea Richards, supply chain strategist at FAHM Technology Partners in Richmond Hill, Ontario, which provides AI and blockchain-enabled supply chain technology solutions, such as order management systems, for the automotive sector.

“But I think everybody has gotten too comfortable with the predictable supply chain. It just seems like there are more and more disruptions – the question is, what’s next and how can you best mitigate it?”

While the chip shortage has had the largest impact, it is by no means the only supply disruption impacting automotive. Many of FAHM’s customers operate in the automotive aftermarket, which is experiencing unprecedented shortages of everything from sensors and window switches to tires and brake drums.

Reducing dependency

In order to secure their supply chains, nations around the world have recently announced plans to build foundries on their own soil. U.S. president Joe Biden has made domestic chip production a key pillar of his post-Covid recovery strategy, and in March, Intel announced its intentions to build its first large-scale foundry in Arizona. Several large foundries are planned in Europe, and China is planning to make itself self-sufficient in chip production in the near term.

Foundries don’t come cheap. According to data from the Semiconductor Industry Association (SIA), building a state-of-the-art facility would require an investment of between US$10 and 40 billion. Securing the sales volume to amortize such a plant in North America will be challenging because the customers for these chips – manufacturers of laptops, phones, automotive control modules – are now mostly located in Asia.

“When we build all these chip factories, we’re going to have a higher cost chip, because our costs are higher here,” says Harry Moser, president of Florida-based Reshoring Initiative. “And we’re going to have to sell a substantial amount of the output into China, which at any point in time can say ‘we don’t want to buy your chips.’ So we need to get overall U.S. costs down so that chip-using industries also come back.

Building a Canadian facility would require significant government funding, which seems unlikely. “I think the talks would stall pretty quickly if we talked about getting a state-of-the-art commodity semiconductor factory here like the ones you hear about in Taiwan and Korea,” says Gordon Harling, CEO of CMC Microsystems, a not-for-profit that promotes technology development and adoption in Canada.

“These plants only create a couple of hundred jobs because they’re highly automated, and all the money to build the plant would leave the country because there are almost no equipment providers in Canada. Canadian taxpayers would never stand for it.”

Canadian advantage

Canada does, however, have significant strengths. “What we’re trying to do is identify areas where Canada has a real advantage – where we have a highly trained people, clever researchers, and an installed base of technology. We want to take those areas and accelerate them to lead to more commercial products based on the technologies,” Harling says.

To that end, CMC is proposing that Innovation, Science and Economic Development Canada (ISED) funds investments in three strategic areas: photonics, which uses light for sensing and high-speed data communications; micro-electromechanical systems (MEMS) sensors, which have a growing use in automotive and mobile devices; and, quantum technologies for sensing, communications, and computing.

Regardless of whether Canada enters the ring to produce semiconductor chips at high volumes, the global market appears poised to change dramatically. Market intelligence company Interact Analysis predicts that as investments in new capacity are realized, and automotive OEMs accumulate large emergency stocks, there will be a crash in the semiconductor market in late 2023 or early 2024.

Chips, therefore, may no longer be an issue when the next disruption occurs.