A photo taken on November 8, 2022 shows the logo of German chip manufacturer Elmos Semiconductor at the company's headquarters in Dortmund, western Germany. INA FASSBENDER/GettyImages.com

South Korean pure-play foundry SK keyfoundry and German automotive chipmaker Elmos Semiconductor SE signed long-term wafer supply agreement Tuesday, locking in 8-inch (200mm) capacity on a 130nm process through 2037 — a deal that upgrades one of the semiconductor industry's most durable fabless-foundry partnerships and directly targets the smart power chips that software-defined vehicle architectures now require. The contract's first product, the E550.01 smart 4-channel electronic fuse (eFuse) controller, is designed to replace the passive blade fuses in next-generation zonal vehicle power networks with a device that can be monitored, adjusted, and reset entirely through software.

Eighteen Years on One Node — Then a Generational Jump

The two companies have rarely needed to discuss their relationship in public: it has simply worked. Since approximately 2008, Elmos has sourced wafers exclusively from SK keyfoundry on its 350nm analog mixed-signal platform, using them for the body electronics chips that move through virtually every vehicle built in the past decade and a half. Tuesday's agreement ends that long stability — deliberately — by migrating the entire relationship to 130nm.

The shift matters to anyone who buys or makes automotive chips. A 350nm process offers excellent analog device performance, but its digital logic density is limiting: each logic gate occupies roughly seven times the silicon area it would at 130nm. As software-defined vehicle platforms demand that a single IC simultaneously manage multiple current paths, report fault conditions through a digital interface, adjust protection thresholds in firmware, and eventually store that firmware in embedded Flash memory on the same die, the economics of 350nm break down. The feature set simply does not fit on a die anyone will pay for.

The technical anchor of the new platform is deep-trench isolation, or DTI — a process in which narrow, deep trenches are etched into the silicon substrate and then filled with dielectric material. The purpose is to create hard walls between circuit elements that would otherwise interfere with each other.

What Deep-Trench Isolation Actually Does

For automotive mixed-signal designs, DTI matters specifically because high-voltage circuits and low-voltage digital logic need to coexist on the same die. A 4-channel eFuse must contain the DMOS transistors that actually switch high current — operating at the 12V or 48V (13.8V to 60V range) levels common in automotive electrical systems — alongside precision analog circuitry for current sensing and temperature detection, alongside the digital logic that decodes SPI commands and drives a status register. Without strong isolation between those domains, substrate leakage currents from the high-voltage side corrupt the low-voltage analog signals. DTI eliminates that problem by creating near-perfect electrical separation between regions, enabling highly integrated designs with low substrate current — without sacrificing voltage headroom or consuming excessive die area for guard rings.

GlobalFoundries, X-FAB, and STMicroelectronics all offer comparable automotive BCD processes with DTI; SK keyfoundry's offering brings the same capability into the automotive segment with AEC-Q100 Grade 1 qualification, meaning the process has been validated across the temperature range from −40°C (−40°F) to +125°C (+257°F) that automotive electronics must survive.

First Product: E550.01 Smart eFuse for Zonal Power Distribution

Traditional vehicles distribute power from a central fuse box containing passive blade fuses scattered throughout the chassis. A blade fuse trips once and must be physically replaced. It cannot report its state, cannot be remotely reset, cannot adjust its trip threshold for different loads, and cannot tell a central computer that a window motor on the rear left door is pulling abnormal current.

Zonal architectures change that by organizing vehicle electronics into physical regions — front, rear, door, roof — each managed by a dedicated zone controller connected to a central domain computer. The zone controller needs to manage and protect power to all the actuators and sensors in its zone without routing every wire back to a central fuse box, which in a premium vehicle might contain 200 fuses and the wiring harness to match. The semiconductor solution replacing that fuse box is the smart eFuse.

The E550.01 is a smart 4-channel eFuse controller: it manages and protects four independent current paths simultaneously, can cut power to any path when it detects overcurrent or other faults, reports status digitally to the zone controller, and can be reset by the zone controller without anyone physically touching the car. Elmos demonstrated an eFuse technology predecessor at CES 2026 in January; the E550.01 on SK keyfoundry's 130nm platform is that concept arriving at full production readiness. Mass production is scheduled to begin in 2027, with samples already available.

The two companies also plan to jointly develop an embedded Flash (eFlash) memory cell on the same 130nm platform — a complementary component that would let automotive microcontrollers store their own firmware on-chip rather than relying on an external Flash device.

Why Long-Term Foundry Deals Have Become the New Normal

The 8-inch wafer segment occupies a structural squeeze point in the global semiconductor industry. Leading-edge foundries — TSMC, Samsung, GlobalFoundries at their most advanced — have directed capital almost exclusively toward 12-inch (300mm) facilities where nodes below 10nm require larger substrates for yield and cost reasons. That migration has systematically starved 8-inch capacity even as automotive, industrial, and IoT demand for mature-node chips continues to grow.

SEMI projected global 200mm fab capacity growing 14% from 2023 through 2026 to a record 7.7 million wafers per month, with automotive and power semiconductors as the primary investment drivers — a category projected to grow 34% over the same period. Even so, supply has tightened rather than loosened, because demand from automotive electrification has grown in parallel. Research firm TrendForce documented 8-inch fab utilization nearing 90% in 2026, up from roughly 80% in 2025. Meanwhile, Samsung is winding down its Giheung S7 8-inch facility in the second half of 2026, cutting monthly capacity by approximately 50,000 wafers; TSMC is closing its Hsinchu Fab 2 by 2027; and NXP has announced four 8-inch fab closures as it transitions production to 300mm.

The Elmos deal is a direct response to this structural reality. For Elmos, a commitment that runs through 2037 locks in guaranteed 130nm manufacturing access through the bulk of this decade and the next, insulating the company from the capacity competition that drove automotive chip lead times above a year during the 2021-2023 shortage.

One important caveat from The Elec's reporting on the deal: the agreement includes a provision allowing wafer supply prices to be adjusted over its term. Elmos has secured supply certainty, not price certainty — a rational tradeoff given that supply disruption is what stranded automotive production lines in 2022, not price spikes. Over 11 years, however, SK keyfoundry will have at least some ability to recapture value as 8-inch capacity tightens further. Procurement teams reading this deal as a complete cost hedge should note the distinction.

For SK keyfoundry, a long-term customer of Elmos's caliber provides utilization visibility that justifies sustaining and potentially expanding 130nm automotive capacity. An SK keyfoundry official was direct about the logic: "Securing a long-term supply agreement is meaningful not only in terms of revenue but also for our technological growth."

What Made This Partnership Possible

The Frankfurt signing ceremony Tuesday formalized what both companies describe as a mutual upgrade of an 18-year relationship. Elmos Chief Executive Officer Dr. Arne Schneider framed the supply dimension explicitly: "SK keyfoundry has been a reliable and innovative partner for many years. This new contract lays an excellent foundation for our future collaboration. The additional wafer capacity ensures planning certainty through 2037, thereby strongly supporting our supply reliability and continued growth." Derek D. Lee, CEO of SK keyfoundry, pointed to the automotive revenue growth argument: "Building on SK keyfoundry's strong technology platform, we will pursue our shared goal of developing innovative semiconductor solutions for automotive electronics. At the same time, we will continue to expand our supply of automotive semiconductors and strengthen our competitiveness as a global foundry." Both executives' quotes are drawn from the joint press release issued Tuesday.

The Dortmund-based company has spent more than four decades building depth in automotive analog mixed-signal ICs, including ultrasonic sensing for parking systems, LIN-bus driver ICs, motor management, and lighting control. The software-defined vehicle transition is expanding the addressable market across all of those domains — but most acutely in power management, where the shift from passive fuse protection to intelligent zonal power distribution represents a near-complete product category replacement that Elmos is positioning the E550.01 to capture.

Is This a Model Other Automotive Chipmakers Will Follow?

The painful lesson of the 2021-2023 automotive chip shortage — that treating mature-node foundry capacity as a commodity available on short notice creates catastrophic supply risk — has already pushed multiple fabless automotive semiconductor companies toward long-term anchor agreements. For specialty Korean foundries like SK keyfoundry, automotive customers are a particularly attractive segment: they command premium pricing relative to consumer electronics, require rigorous qualification processes that create sticky long-term relationships, and generate stable, predictable volumes.

Whether the E550.01 becomes the commercial anchor of a broader 130nm product family at Elmos will depend partly on how quickly OEMs standardize on true zonal power distribution at volume — a transition that is moving across the industry but unevenly across vehicle segments and price points. The 2037 horizon of the supply agreement leaves Elmos enough runway to find out.

What Is a Deep-Trench Isolation Process, and Why Does It Matter for Auto Chips?

Deep-trench isolation is a semiconductor fabrication technique that creates physical barriers — trenches etched several microns deep into silicon, then filled with insulating material — between different types of circuit elements on the same chip. For automotive power management ICs, it solves a specific problem: high-voltage transistors (operating at 48V or higher) generate substrate currents that can corrupt nearby precision analog circuits operating at millivolt signal levels. DTI blocks that interference, allowing both circuit types to coexist on the same die without the large, area-inefficient guard-ring structures that older isolation techniques require. The result is a smaller chip with more capability per unit of silicon area — a prerequisite for the feature density that zonal vehicle power management demands.

Frequently Asked Questions

Why did Elmos switch from 350nm to 130nm after 18 years on the same node?

The 350nm process gives excellent analog device performance, but its digital logic density is inadequate for next-generation automotive power management chips. A 4-channel smart eFuse like the E550.01 needs analog current sensing, digital interface logic (SPI), and high-voltage power switching on the same die. At 350nm, the die area required for that combination makes the chip economically uncompetitive. Moving to 130nm with DTI reduces digital gate area by roughly 7x while adding the isolation capability to put high-voltage and low-voltage circuits next to each other — enabling feature integration that simply was not practical on the older node.

What is a zonal vehicle architecture, and why does it need smart eFuses?

A zonal architecture organizes vehicle electronics into physical regions — front, rear, door zones — each managed by a local controller connected to a central computer, rather than routing every function through domain-specific boxes spread across the car. This approach dramatically reduces wiring harness length and weight, and enables software updates to reach power management systems that were previously hardwired. Smart eFuses are the enabling component at each zone: they replace passive blade fuses with devices that can monitor current, protect against faults, report status digitally, and be reset remotely — turning what was a purely passive safety component into a software-managed element of the vehicle's power network.

Does the 11-year supply agreement guarantee Elmos fixed wafer prices through 2037?

No. The Elec's reporting on the deal notes that wafer supply prices may be adjusted under the agreement over its term. Elmos has secured supply capacity certainty — the right to a defined volume of 130nm wafers from SK keyfoundry through 2037 — but not a fixed price for those wafers across the full 11 years. This is a significant and deliberately chosen tradeoff: supply disruption, not price inflation, was what halted automotive production lines during the 2021-2023 chip shortage. The price-adjustment provision gives SK keyfoundry room to reflect market conditions as 8-inch capacity continues to tighten, while Elmos retains the manufacturing access it needs to serve customers regardless of those market conditions.

When will the E550.01 and the jointly developed eFlash cell reach automotive buyers?

The E550.01 smart 4-channel eFuse controller is in sample availability now and is scheduled for full mass production in 2027. The jointly planned embedded Flash (eFlash) memory cell on the same 130nm platform is also targeted for 2027. Both timelines are subject to automotive qualification completion, which for AEC-Q100-certified automotive ICs typically involves 12 to 18 months of qualification testing before volume production can begin.