Who will survive in processor architecture race?

The microprocessor industry is facing great changes now. It’s been a long time since Intel pioneered the computer architecture industry with its legendary x86. Or take the original ARM CPU with its 3-stage pipeline from the late 1980s performing hand-in-hand with Intel. Now both Intel and ARM dominate in the mobile/server/PC CPU markets for a long time and it’s clear why. Both have mature software stacks and large ecosystems.

Anyways, the emerging market conditions cannot but push them to intention to rethink their business models. Competitors with another approach regarding no license or royalty fees like ARC (Synopsis), Tensilica (Cadence), RISC-V, and MIPS (its future is in doubt though) offer flexibility, customization and comfortable time-to-market conditions. Here is their current performance based on its market capitalization.

Microprocessor vendor

Market Capitalization (USD)

Intel

232.04B −0.26 (0.49%) ↓

Arm Holdings

31.91B +0,33 (+0,44%) ↑

ARC

65.41M +0.0100 (0.71%) ↑

Cadence DS (Tensilica)

18.68B −0.95 (1.40%) ↓

The question is whether the transition to building CPUs based on a non-x86 technology will happen or not.

One of the promises of open-source architectures like MIPS and RISC-V is the possibility of going beyond the instruction set architectures or ISA while designing application-specific SoC. Thanks to this processor architecture becomes customizable without having to pay an upfront fee. Except for paying for tools, test environments, and verification, the cost of entry is lower, as you need no license for ISA. That is what may hurt Intel or Arm in the long term.

RISC-V threatens to displace not only x86 out of high-end PCs and servers but also ARM out of the mobile and IoT market. When you look at how many companies are adopting RISC-V like Nvidia, Alibaba, Qualcomm, Google, Huawei, you must think that RISC-V must be, say 50% faster or efficient. The reason why it’s gaining so much attention is nothing to do with benchmarks or performance metrics. Though, it’s certainly possible to achieve those results if that’s what a solution requires. The real reason why it’s gaining traction with so many large hardware companies has to do with a business model. It gives companies absolute freedom in innovating and advancing their proprietary chips. They can choose where the IP, open-source, or commercial cores, and more importantly, they can extend these cores to fit their needs. That is a huge competitive advantage in the industry that feeds in innovation.