Arm has formed a new “solution engineering” team to develop prototype chips for smartphones, laptops and other consumer electronics and to demonstrate the technology’s capabilities. financial timesThere are concerns in the industry that Arm plans to either sell the chips themselves or license such designs in competition with licensees, but sources close to Arm say they are not planning to sell products. Or deny any licensing plans and claim they are only working on prototypes.
Traditionally, Arm licenses its instruction set architectures, logical designs for CPUs or GPUs, silicon-proven physical designs for CPUs or GPUs, and various other IP blocks to its clients. However, the chip, which is being developed by a solution engineering team led by industry veteran Kevork Kechichian, is said to be more advanced than ever, according to FT industry insiders. Given the complexity of the project, some in the industry believe Arm may build his own branded SoC, or at least license a reference design rather than license his IP. . Arm declined to comment for this information.
According to FT sources close to Arm, this is not the case. The company will reportedly only develop a prototype chip or chips to show what its IP can do in terms of capabilities and performance. On the other hand, the development of complex systems on his chip is very expensive.there is Estimate A fairly complex 5nm SoC design can cost $540M (including software), while a complex 3nm SoC development cost can be as high as $1.5B including software. I have.
“Working on intellectual property is one thing, but actually designing and working with production partners to turn those efforts into physical chips is a whole other area. “At some point in the future,” a former Arm executive familiar with the effort told the Financial Times. [Arm] You will definitely need a return to justify that large investment.
Arm itself has not commented on the issue, so we can only speculate as to what their solution engineering team is talking about. Considering the increasing cost of chip design, there may be good reason to invest in Arm’s chip design.
For example, a company may have developed a customizable, silicon-proven reference design containing IP that is guaranteed to work perfectly when implemented in a specific process technology. Few companies can put $500 million to $1.5 billion into a chip design, but he might want to license something that’s guaranteed to work.
Another reason Arm develops physical implementations of IP is that in the next few years many customers may license chiplets or chiplet designs instead of IP due to cost. is.
But in either case, Arm could be competing with its own customers, such as Qualcomm, MediaTek and NXP, who sell their chips to device makers. This will undoubtedly increase the trend to adopt other instruction set architectures such as RISC-V. This is of course a strategic threat to Arm. On the other hand, if small businesses can’t license the latest technology to stay competitive with larger companies, they will either go out of business or adopt his open-source RISC-V design, which also poses a strategic threat. increase.
For obvious reasons, Arm will have to address a variety of strategic challenges before going public with its IPO later this year, so building a “solutions engineering” team is one move towards that end. It may not be too much.
