SoftBank, Arm and lessons in IoT

So Nvidia abandoned its plans to buy Arm. It was all over the news last week, as reported by myself for EE Times here in Nvidia Abandons Arm Deal, Segars Steps Aside for IPO, and by almost every mainstream newspaper. The implications of this combined with other news, such as Intel Foundry Services broader goal of embracing multiple instruction set architectures beyond x86, including Arm and RISC–V means it was also a big week for RISC-V.

There are probably some lessons to learn from SoftBank's strategy of using a big-ticket purchase of Arm to pursue an IoT strategy.

On the surface of it, back in 2016, some of the strategists who may not have been so embedded in the semiconductor industry probably thought that it seemed a natural progression for Arm, given its dominance in mobile phones: low cost and low power seemed to be what would be needed for the massive IoT market that all the analysts have been forecasting.

It seemed a no-brainer. Just add the connectivity (Stream Technologies), and data management (Treasure Data), and you have the full stack for an IoT services platform, as Rajeev Misra, then CEO of SoftBank, said at a conference in London in 2018. However, it became apparent that wasn't working well and Arm had said it would divest its IoT services business to the SoftBank Group two years ago, to focus on semiconductor IP and particularly compute and data.

But by then, while Arm's technology for data centers is certainly right for the task, it maybe that it lost valuable time in tackling the computing market while SoftBank focused Arm's resources on targeting the trillion connected devices vision. The Financial Times quoted Arm co-founder Tudor Brown today saying he found Arm's heavy investment in IoT "strange" given there was never going to be any money in that market.

The challenge for Arm may have been the business model: the license fees may just not be suited to the fragmented IoT market: in order to get a return on investment, device manufacturers need volume to generate the royalties needed to cover the upfront investment, but there are just too many variables involved in targeting IoT to make that volume requirement a certainty. This is where developers adapting RISC-V architecture may have been able to make quicker inroads into the IoT market, just because of lower upfront costs.

There's certainly a lesson in all of this. It's all very well going for huge markets. Many a startup has quoted billion dollar markets in their business plans with the hope of attracting investment money.

That's not the case necessarily for Arm, as the company was most likely thrown into the situation as a result of SoftBank's vision. But targeting the trillion connected device market was simply not a case of throwing money at it to create the full platform, as SoftBank has no doubt now learned.

For other news over the last week, check out the weekly round up on embedded.com, where you can read about the latest from Microsoft, Infineon Technologies, congatec, Quantum Motion, Communicate2Integrate, Semtech, Tanvas, Sensata Technologies, Basemark, Percepio, MORAI, CogniFiber, Xilinx, and Flex Logix Technologies.