Intel has repeatedly tried to get into such areas as foundry supply and failed and meanwhile the manufacturing leadership that gave it control of the PC industry and other sectors, has evaporated. It was current chip-manufacturing golden child Taiwan Semiconductor Manufacturing Co. Ltd., that has taken Intel's crown as manufacturing leader. But it too has made mistakes. TSMC's six-year flirtation with solar and LED manufacturing as diversifying alternatives to IC foundry work came to an end in 2015.
ARM has never been one for the big merger or acquisition preferring to pick up small dynamic companies as "tuck-in" additions to its portfolio of IP.
ARM has achieved gradual and almost inexorable success. Its processor core design wins have spread from mobile phones, to embedded applications, to servers and most recently to laptop computers (see Apple confirms switch to ARM-based processors for Mac computers). At the same time it has also moved up to offer the mbed operating system and IoT and data services. The result is it has become a bit of sprawling behemoth.
It is notable that the CEO and CTO are now both based in Silicon Valley (see Dipesh Patel takes over as CTO of ARM). It is perhaps also notable that Patel is a former president of the IoT Services Group.
So perhaps a refocus is appropriate? Well yes and no. It is important that ARM does not throw the baby out with the bath water. Not least because it now has a strong competitor on its home turf in the form of the ecosystem developing around RISC-V (see SiFive creates global network of RISC-V startups). RISC-V is a more recently developed, modular RISC instruction set architecture for processors, that carries none of baggage of having to meet legacy requirements from another era of computing.
These are the same advantages that ARM originally had when it went in to battle against x86 and 68000 processors from Intel, Motorola and others in the 1990s. But now it is ARM carrying the baggage. The one advantage ARM does have over RISC-V is the breadth and strength of its ecosystem and the ability to put firmware and early layers of software next to its chips. So to focus too much on cores and ignore the importance of the platform sale would be a mistake. It would give RISC-V a chance to compete head-to-head. It is not a mistake a expect ARM to make.
ARM has the resources to do great things in the area of artificial intelligence and machine learning, which could at least hold RISC-V at bay. ARM will need to focus to do that and most likely the competition will come from several of the tens of startups aiming for that opportunity. So ARM has its work cut out before it gets distracted by ride-sharing services and subscription models for data sharing.
The IoT Services Group, the part which ARM is passing along to SoftBank, was built up in part through acquisitions of IoT startups Treasure Data and Stream Technologies in 2018. The core offering is the company's Pelion IoT Platform. That was after SoftBank's acquisition of ARM for $32 billion in 2016, when SoftBank's Son said the deal was done because of ARM's foundational technology and its growth potential in IoT.
With this latest move coming after that announcement the conclusion must be that ARM no longer has growth potential in IoT services. Why would that be?
One possibility is that a potential buyer of ARM has gone through the company structure telling Segars and Son what is it prepared to buy and what it does not require. "Dump this, this and this and we will take what we really care about, the semiconductor IP generating capability." That allows the acquirer to pay the minimum price and have a minimum of post-acquisition housekeeping to do.
Next: Apple as potential buyer