Arm, Intel make it easier to churn out Arm SoCs from Intel fabs
Energy sipping designs for mobile, ultimately coming to DC, too
Intel and Brit chip design outfit Arm have put aside their differences and penned an agreement to make it easier for Arm licensees to have their products manufactured at an Intel fab using an upcoming advanced production node.
Labeled as a "multigeneration agreement," the move will see Arm and Intel Foundry Services (IFS) work to optimize their respective technologies to help chip designers have low-power system-on-chip (SoC) devices made on the Intel 18A process node.
The pair said that the focus is on mobile SoC designs first, meaning chips for smartphones, but also leaves the door open for future optimization for a wider range of applications, including automotive, Internet of Things (IoT), datacenter, aerospace and government applications (which is likely code for defense).
Arm CEO Rene Haas welcomed the move, and emphasized Arm's credentials in building more energy efficient compute infrastructure.
"As the demands for compute and efficiency become increasingly complex, our industry must innovate on many new levels. Arm's collaboration with Intel enables IFS as a critical foundry partner for our customers as we deliver the next generation of world-changing products built on Arm," he said.
Intel CEO Pat Gelsinger instead focused on the move offering more options for chip design companies seeking advanced fabrication processes for their products.
"Intel's collaboration with Arm will expand the market opportunity for IFS and open up new options and approaches for any fabless company that wants to access best-in-class CPU IP and the power of an open system foundry with leading-edge process technology," Gelsinger noted.
The Santa Clara chipmaker last year started an overhaul of its IFS operations, hoping to make them run more along the lines of silicon contract manufacturing companies such as Taiwan's TSMC. It dubbed the process the "IDM 2.0" strategy, where IDM stands for Integrated Device Manufacturing.
As The Register reported at the time, this should not only provide reassurance to foundry customers that there is separation between that side of the operation and Intel's own chip designers, but also puts the relationship between the two Intel operations on a more formal footing.
Intel said that its open system foundry model goes further than the traditional wafer fabrication services offered by other companies to include packaging, software and access to its chiplet technology.
It's all about the money
It won't have escaped the attention of Reg readers that Intel and Arm are to a certain extent rivals, with Arm trying to muscle in on the datacenter processor market that Intel still enjoys the lion's share of with its Xeon chips.
However, a partnership such as this one makes commercial sense if Intel really wants to make more money with its foundry operations as Arm licensees typically ship billions of chips every quarter.
"Intel is in the process of reinventing itself (or at least part of itself) as a foundry provider competing with the likes of TSMC and Samsung, in which case it is essential to have an agreement in place to produce chips including Arm IP," said Richard Gordon, Vice President of Semiconductors & Electronics at industry analyst Gartner.
"This is really a case of nothing to see here, other than the observation that Intel's x86 competes with Arm-based processors in some applications," he added.
Intel has good reason to look for additional sources of revenue: the chip giant recently reported a $700 million net loss for the fourth quarter of its 2022 financial year, while revenues dropped 32 percent, partly because of falling PC and laptop sales following the boom in demand during the pandemic.
Power and performance
The latest agreement will see Intel and Arm undertake design technology co-optimization, meaning the two companies will work together to ensure that chip designs and Intel's process technologies are tweaked so that chips produced using them are optimized for power, performance, area and cost.
This will cover Arm cores targeting the 18A process node, which is slated by the company as a 1.8nm technology and which Intel itself plans to use first for an upcoming version of its Xeon processors codenamed Clearwater Forest, due to arrive in 2025.
- Intel successfully ships an updated datacenter roadmap
- Nexperia claims Newport Wafer may close if sale goes ahead
- How Arm aims to squeeze device makers for cash rather than pocket pennies for cores
- Why Microsoft is really abandoning evaporative coolers at its Phoenix DCs
Intel claims that 18A brings with it two enhancements: PowerVia, which separates out the power wires from the transistor layer to reduce interference, and the RibbonFET gate-all-around (GAA) transistor architecture, which succeeds the FinFET design.
Will it float? Yep, apparently on NASDAQ
In other news, the Financial Times reports that Arm's parent company SoftBank is expected to sign an agreement with the Nasdaq stock exchange in New York to list the chip design company there following its initial public offering.
The move is said to represent the first formal step in the IPO process. Arm confirmed last month that its shares will be listed only in New York, dashing the hopes of the UK government and others that it might be listed in London. ®