The cancellation of 20A is particularly notable as Intel CEO Pat Gelsinger had positioned it as the start of the company’s “Angstrom era” when first announced in 2021. Now, the shift to external manufacturing for Arrow Lake is a significant departure from those plans, but it aligns with the company’s recent cost-cutting measures announced following concerning financial results last quarter. The company is undergoing major restructuring, including laying off 15,000 workers, amid financial challenges.
Intel first demonstrated the Arrow Lake chips on 20A wafers at its Innovation 2023 event. At the time, Intel targeted a 2024 release for Arrow Lake.
Intel CFO Dave Zinsner said the company would “skip over productizing” the 20A node to reduce capital expenditures. Resources previously allocated to 20A development will now be directed toward the more advanced 18A node, slated for 2025 launch. The company projects the change will save $500 million by avoiding costs associated with ramping up 20A production.
Intel’s 18A node development continues
While Intel claims the shift was motivated by strong early yields on 18A, the move raises questions about potential issues with 20A development. The node was set to introduce key technologies like RibbonFet transistors and PowerVia backside power delivery.
While Reuters had reported that test silicon wafers built on the 18A had suffered setbacks during the tests conducted by Broadcom, Intel maintains its 18A node remains on track, citing a defect density below 0.40 – a key metric indicating production readiness.
Sell stated, “Intel 18A is powered on, healthy, and yielding well. We remain fully on track to begin high-volume manufacturing next year.” The company says learnings from 20A contributed to 18A’s progress.
Despite manufacturing setbacks, Intel has secured some high-profile foundry customers, including Microsoft and the US Department of Defense, both of whom have committed to using the 18A node.