You are currently viewing SemiWiki as a guest which gives you limited access to the site. To view blog comments and experience other SemiWiki features you must be a registered member. Registration is fast, simple, and absolutely free so please, join our community today!

  • ARM acquisition by SoftBank – is it good or bad for the UK and the electronics industry?

    Article: The Need for OASIS in Post-layout IC Databases-softbank-chip-min.jpgAs a member of the UK semiconductor community and as an ARM competitor Monday’s announcement of Softbank acquiring ARM came out of the blue. So what does this mean for UK tech and for the electronics industry more generally?

    Firstly, I must say that ARM has done very well emerging from the ashes of the 1980s UK semiconductor industry. In the 1980s home grown giants such as GEC, Plessey and STC were in decline and very dependent on UK Government contracts; which were rapidly disappearing under the Thatcher regime. Start-ups were few and far between and, when I was at GEC Hirst Research, I was interested to hear that my colleagues Harry Oldham and Peter Harrod were recruited to join Acorn and work on processor chips. When ARM was spun out of Acorn they did very well to get investment from Apple associated with the Newton product and even better to comprehensively win business in the mobile phone space.

    While in the 1990s the MCU market was dominated by semiconductor companies with their own (often good) architectures such as Microchip, ST, Hitachi, Motorola, TI, Fujitsu, NEC, Infineon, Atmel & Toshiba. Amazingly ARM has achieved a nearly clean sweep in replacing in-house and good processor architectures at almost everywhere but Microchip (excluding the Atmel part). This has been a very impressive achievement indeed but is it good for the electronics industry? Can an MCU really stick out from the competition just based on its memories and peripherals? I doubt it.

    Perhaps an easy question to ask is why ARM accepted Softbank’s offer of £24.3B ($32B). The offer was in cash and at a 43% premium to the market price which was high despite turbulence due to the Brexit referendum. However as a member of the UK tech community I can join Hermann Hauser and David Manners in being sad. ARM now joins CSR as one of the UK’s big start-up successes that has become foreign owned. The Financial Times BTW has a fascinating account of how the acquisition offer took place in Marmaris.

    On the political front our new Prime Minister, Theresa May, spoke out against British companies being sold to transient shareholders and called for “a proper industrial strategy” and furthermore she said “it should be capable of stepping in to defend a sector that is as important as pharmaceuticals is to Britain”. Yet on the ARM acquisition our new Chancellor of the Exchequer (Finance Minister) Philip Hammond said the deal was a “big vote of confidence” in British business. Is this yet another example of UK political opportunism or is it Government policy that pharmaceuticals are strategic to the UK while electronics is not?

    One of the most frustrating things about competing against ARM with perfectly good processor cores is that ‘nobody gets fired for buying ARM’. I am old enough to remember that in the 1980s ‘nobody got fired for buying IBM’ mainframes and yet a decade later IBM was being replaced by a new generation of PC manufacturers. Many people say that they choose ARM because of its ecosystem and yet say the embedded market the industry standard is the high level language – C or C++ - not the architecture to which it is compiled. The main hardware dependency is the peripherals and one MCU vendor will need different drivers to their competitor even if the same ARM core is used.

    The big question is what SoftBank wants to do with ARM. Are they more interested in developing IoT chips than licensing processor IP? If that is the case they will compete with the ARM customer base. Even the risk of this should make semiconductor companies think of using an alternative to ARM. Or is the focus of ARM shifting to the system level? In EETimes Junko Yoshida makes some interesting comments about SoftBank needing to get into hardware.

    So is this good for the electronics industry as a whole? Who knows? It is too early to say how this will turn out. However one point ought to be self-evident which is that the near monopoly of ARM in processor IP is neither healthy nor sustainable. Although Xilinx pioneered FPGAs, the industry has benefited greatly from strong competition from Altera. There needs to be a viable alternative to ARM for the good of the market whether somebody like my company Cortus or an initiative like RISC-V.