Intel CEO Pat Gelsinger said he plans to commit some production at one of its largest non-US fabs for the automotive sector as it moves into a growing market that the global chip shortage has ravaged for months.
The Santa Clara, California-based company is reserving an unspecified amount of output at its Ireland fab to roll out chips for auto makers that urgently need them, the CEO said on Monday at the IAA auto conference. Intel would not say when it plans to start churning out automotive-grade chips from the plant–its largest 300-mm fab outside the US. The factory specializes in chips based on its previous generation 14-nm process.
The move is closely tied to its strategy to break into the contract chip manufacturing business and take on TSMC. It has established an independent foundry business called Intel Foundry Services—IFS, for short—that will take advantage of its vast manufacturing operations to build chips for other firms and even rivals based on their own blueprints. In addition, it has landed Qualcomm and cloud giant Amazon Web Services as early customers.
As part of its push into the contract chip-making world, Intel said that it has also been aggressively courting potential customers in Europe, including many of the auto OEMs and Tier-1 suppliers based there.
The automotive industry is an obvious target as most manufacturers are in dire need of chips. The sector has been buffeted by shortages that have dragged on for most of the year and crippled car production. The likes of Ford and General Motors said last month that they would have to halt some production plants due to a dearth of chips. Nissan, Toyota, and Volkswagen have also slashed global production by as much as 40%.
Over the last year, a rise in demand for personal computers and other consumer electronics has clogged chip manufacturers with a backlog of orders, including for microcontrollers (MCUs), power management ICs, and other chips used in vehicles. The shortage has also strained the supply chains for silicon wafers, substrates, packages, and other key materials for manufacturing chips. As a result, prices are also on the rise.
Intel has previously said that it wanted to start supplying chips to the auto industry by roughly the end of the year to help ease the chip shortage. But it can take a long time to convert fabs over to automotive-grade chips.
These types of chips must be rigorously tested and then qualified against industry standards to guarantee they can tolerate harsh temperatures, vibrations, shocks, and other brutal conditions in vehicles. They also have long lifecycles, with many semiconductor vendors keeping them in production for more than a decade. Because of these factors, it is more challenging for vendors to reequip semiconductor fabs for car chips.
Intel—along with most major semiconductor vendors—sees the automotive sector as a golden opportunity for the future. "Cars are becoming computers with tires,” Gelsinger added. “You need us, and we need you.”
According to market researcher Gartner, demand for automotive chips will double by the end of the decade, growing to $115 billion or 11% of the global market. The silicon content in cars is also growing at an unprecedented rate. As the auto industry enters the era of electric, connected vehicles, Gelsinger said the company believes chips will account for 20% of the material costs of a premium vehicle by 2030, up from 4% in 2019.
Today, most of the microcontrollers and other chips used in cars are based on legacy process nodes such as 40-nm and 28-nm, lagging the more advanced chips at the heart of smartphones such as Apple’s iPhone. While car manufacturers don’t compete directly with consumer electronics firms for the same chip supplies, they are struggling to be placed on the priority line at foundries, which have been fully loaded for months.
But vendors are moving many of these chips to more advanced process nodes so they can run richer automotive-grade operating systems for features such as dashboard displays and advanced safety systems.
To that end, Intel is establishing a program called Intel Foundry Services Accelerator to assist automobile chip designers in moving to more advanced process nodes. At the same time, it is creating a support team to help them adapt their chip designs to be compatible with Intel’s more advanced fabs. It is also offering a range of custom and industry-standard intellectual property to help meet the unique needs of its future automotive clientele.
It is unclear whether any auto OEMs or Tier-1 suppliers have agreed to adopt Intel’s foundry services so far.
Intel is trying to take advantage of the fact that car companies are reimagining their supply chains and overhauling how they purchase chips due to the sourcing challenges that have plagued them for months.
According to KPMG, worldwide chip shortages will cost auto makers $110 billion in lost sales in 2021.
Today, auto manufacturers rarely buy chips—or deal with chip suppliers at all—directly. Instead, they source electronic control units (ECUs) and subassemblies from Tier-1 suppliers. These Tier-1s purchase chips from automotive chip specialists such as NXP, Infineon, TI, and Analog Devices—which work with foundries to get many of their higher-end microcontrollers and other chips made—and then assemble them into subsystems.
That gives them less leverage to persuade foundries to roll out chips for cars. They also have tight profit margins, leaving the semiconductor firms that supply them less flexibility to bid up prices for foundry orders. On the other hand, consumer electronics giants and their vendors rarely retract orders. They also have high-volume, long-term deals in place with foundry partners, giving them guaranteed spots in production plants.
Industry executives have said that they expect the auto industry to start shifting away from “just-in-time” inventory management to keeping larger amounts of crucial parts, including semiconductors, on hand.
Car manufacturers are also trying to forge deeper relationships with chip vendors and the foundries a tier below them to keep the automotive chip shortage from happening again. For example, NXP executives have said they are striking more long-term contracts with auto makers at the top of the supply chain, which is, in turn, sharing more binding forecasts that help them with capacity planning—in-house and outsourced.
While Intel’s decision to commit existing foundry capacity to auto makers may alleviate the bottleneck for some, industry analysts warn there is no short-term fix for the global supply crunch. The problem is that, while many auto makers are adopting more advanced chips, they also rely on a steady supply of chips built on older, cheaper process nodes (produced on 200-mm wafers), where the worst supply constraints remain.
Instead of building more capacity for these more mature nodes, Intel said it plans to work with its foundry customers to port their chip designs over to newer process nodes. But that is still going to take some time.