Arm & NVIDIA – RISC Risk

RISC-V gains more momentum.

• One of the big problems of Arm’s perceived loss of independence by being acquired by NVIDIA is that interest in rivals, once thought to be dead, has been greatly revived.
• I don’t think that RISC-V has any chance of challenging Arm at the high end without massive Chinese government intervention, but at the low end, Arm could cede a lot of volume and scale to RISC-V.
• The fact that NVIDIA is acquiring Arm is enough to make some people very nervous regardless of the fact that there are real-world examples (Samsung (see here)) of where complete independence can be maintained.
• Following a strategic review after the UK government intervened to block technology transfer to China, Imagination Technologies is making noises that could be easily interpreted as a move to RISC-V (see here).
• Imagination now sees that the acquisition of Arm by NVIDIA has put a lot of momentum behind RISC-V and it thinks that a number of customers are keen to have an alternative to Arm.
• The company now intends to invest $150m to develop processors using the RISC-V architecture which is more than double the amount that SiFive raised to develop the same thing (see here).
• The biggest hurdle that RISC-V faces in competing against Arm is the huge ecosystem that Arm has built over the last 20+ years.
• This means that everyone develops on Arm, everyone knows it and in order to compete head-to-head, RISC-V would have to recreate all of that almost from scratch.
• Another issue is Arm’s scale meaning that the cost to develop a processor to match Arm’s high-end product would probably cost more than to simply license the product directly from Arm.
• This is because Arm can spread the cost of development over billions of chips shipped allowing it to make very high margins despite charging its clients less in royalties than they would incur to create the product themselves.
• The only way I think that RISC-V could climb this hill is in China where the CCP could decide that the only way to secure its technological independence is to abandon Arm and go with RISC-V.
• It would be in a position to compel companies to shift their hardware and software from Arm to RISC-V.
• I think that the CCP is very far from doing this as, despite its control, this is a very risky move that could jeopardise the viability of a lot of Chinese companies.
• One only has to remember the disastrous TD-SCDMA Chinese proprietary standard for 3G networks to understand that the CCP will think long and hard before making such a profound move.
• However, it will not be lost on anyone that Imagination Technologies is now majority-owned (indirectly) by a series of Chinese government entities raising the possibility that it is using Imagination as a trial balloon for this strategy.
• When one looks at the low end of the market like the Cortex M0, the situation becomes much easier because the amount of work required to make an equivalent RISC-V processor is far less.
• Hence, if RISC-V is going to have any impact at all, this is where I think it is most likely to work.
• This will be a high volume, low margin market but may give RISC-V a beachhead from which to expand over time.
• If the acquisition were to end up in a relisting of Arm on the LSE, then I suspect that the momentum for RISC-V would very quickly die away, leaving Arm to focus on Intel rather than fighting on two fronts.
• I still think that the best place for Arm is as an independent public company but the hurdles for its acquisition by NVIDIA do seem to be lessening somewhat.