Intel Foundry Services (IFS) And Arm Ink Processor IP Deal For Intel 18A Process Node

Intel Foundry Services (IFS) and leading semiconductor IP vendor Arm have jointly announced a design technology co-optimization (DTCO) agreement to develop multiple generations of Arm-based processor cores that specifically target the Intel 18A process, which is scheduled to be manufacturing ready in the second half of 2024. (When Intel says, “manufacturing ready,” the company means ready for risk production, the term commonly used by other semiconductor foundries.) The initial focus for this collaboration will be mobile SoC designs that require excellent performance with low power consumption, which means that potential customers include all the usual fabless chip-making suspects supplying the high-volume mobile phone market. However, this agreement also allows for design expansion into other markets including automotive, the Internet of Things (IoT), data centers, aerospace, and other government applications. Arm has not revealed details of the new and yet-to-be-disclosed processor cores covered by this DTCO agreement. However, Arm’s CEO Rene Haas describes IFS as “a critical foundry partner.”

The Intel 18A process is the fifth announced process in the company’s plan to introduce five increasingly advanced semiconductor manufacturing process technologies in an astonishingly short five-year span, as shown in the image below. These are all leading-edge process nodes, but the introduction of Intel 18A is the final step in the company’s plan to reclaim transistor performance-per-watt leadership by 2025. That’s according to Dr. Anne Kelleher, executive vice president and general manager of Technology Development at Intel Corporation, speaking at the 69th annual International Electron Devices Meeting (IEDM) Plenary session in December 2022.

In addition to the timely progression of new process nodes occurring with a one-per-year announcement cadence, as shown above, this DTCO agreement between IFS and Arm is another steppingstone towards building Intel’s reputation and credibility as a semiconductor foundry. Such a reputation requires many critical elements that includes leading-edge process nodes and a broad IP offering. Both are required to develop cutting-edge SoCs.

This agreement with Arm broadens the IFS IP portfolio in a significant way. No credible, competitive semiconductor foundry can realistically afford to omit Arm processors, which IFS will offer to its customers through this agreement, in addition to the company’s existing x86 and RISC-V processor IP.

To date, Arm customers have shipped more than 250 billion ICs with Arm processor IP, and many of those chips went into mobile phones and other mobile devices, the initial target for the Intel 18A process. This DTCO agreement with Arm effectively quashes any question about IFS’s ability to deliver a sufficiently broad IP portfolio, which is required if IFS is to become a top-tier semiconductor foundry. It’s quite unlikely that this will be the last major semiconductor IP announcement that IFS will make on its journey.

IFS is a significant component of Intel’s IDM (Independent Device Manufacturer) 2.0 strategy, and the venture brings multiple benefits to the company. First, it opens Intel’s very expensive fabs – existing, under construction, and planned – to fabless semiconductor client companies. Keeping the fabs filled is job #1 for a semiconductor maker, and IFS is one way for Intel to fill its fabs.

Second, IFS appears to be a central element in Intel CEO Pat Gelsinger’s plan to reclaim Intel’s transistor performance-per-watt leadership by 2025. By driving IFS to become competitive with other leading fabs, Gelsinger is forcing Intel to ramp up its process development efforts. Intel has taken a lot of heat for losing this lead to its chief semiconductor manufacturing rival, TSMC, which in turn has allowed Intel rivals AMD and Nvidia to make inroads into Intel’s market share. Intel 18A is where the company has publicly put its stake in the ground, saying “this is where we regain our top position.”

Third, IFS helps to establish a credible future semiconductor demand that justifies the massive fab expansion that Intel is currently undertaking. It also helps to provide support for Intel’s requests for governmental support to build those new fabs in multiple locations around the globe.

For many, Intel’s jump into the foundry business may appear to be a recent diversion away from making Intel-branded chips. However, Intel’s original microprocessor, the Intel 4004, was designed and fabricated as a custom IC for Japanese calculator maker Busicom. At the time, Intel was still developing its first DRAM, the 1103, and was a way for the young company to fill its fab until the DRAM could be launched. For the 4004 microprocessor project, Intel served as both design house and foundry. More recently, in 2013, Altera announced that Intel would build the next generation of Altera FPGAs in a foundry deal. Intel bought Altera two years later. Consequently, Intel has more than half a century of experience as a semiconductor foundry. It simply has not flexed those muscles for a while. Gelsinger is now remediating that situation and building the company’s foundry capabilities through IFS.

Leave a Reply