5G and artificial intelligence (AI) startup EdgeQ today announced that its upcoming modems will be built on the RISC-V architecture.

This approach allows machine learning inference capabilities to be integrated directly into the 5G modem, thereby requiring a chip that consumes less power and providing for greater deployment densities, according to EdgeQ CEO Yinay Ravuri.

As its name suggests, RISC-V is based on a reduced instruction set computer (RISC) architecture. However, unlike IBM Power, Arm, or even Sun SPARC, which also use a reduced instruction set, RISC-V is open source and freely available to anyone.

The technology, introduced in 2010, is also a relatively new instruction set architecture (ISA) that got its start at the University of California, Berkeley, and was eventually open sourced to promote its adoption and development in academic environments. And with just 47 instructions, it is far less complex compared to architectures like Arm, which has more than 200 instructions, or x86, which can have well in excess of 1,500 instructions.

RISC-V forms the basis on which more complex chips can be developed through the addition of custom instructions for specific applications, Ravuri explained.

“We’ve basically inserted our own custom instructions,” he said. “By doing this I get to invent new applications, but yet, use all the ecosystem that comes for free with RISC-V.”

That ecosystem includes free and open source tools, which EdgeQ can take and use to write custom extensions for its modems.

“When I was at Qualcomm, more than 50% of our development was on tools, because you need that basic foundation,” he said, adding that the use of RISC-V frees EdgeQ to focus on writing more efficient instructions for use in modems.

This also means that customers that want to differentiate their implementation of EdgeQ’s modems will be able to reprogram the chips to use their own custom algorithms using common programming languages like C and C++.

Because EdgeQ doesn’t have to implement functionality that isn’t necessary for a specific application it can instead develop modems with efficiency similar to that of an ASIC or FPGA, Ravuri explained.

EdgeQ is targeting its AI-infused modems at industrial applications where the company claims the low-latency and high-availability promised by 5G technology will open the door to new use cases such as factory automation.

For example, in an automobile factory, today’s infrastructure might involve the limited use of robotic welders that fuse pieces of metal as the vehicle passes from one part of the factory to the next. EdgeQ envisions a future where 5G connectivity will enable the robots to go to the cars.

In addition to its adoption of RISC-V architecture, EdgeQ added former Qualcomm CEO Paul Jacobs and CTO Matt Grob to its advisory board.

“The advent of 5G comes with changes in the architecture of wireless networks and an exciting expansion in the types of applications and services,” Jacobs said in a statement. “This inflection point creates the opportunity to build and leverage an open ecosystem based on new platforms and a greater diversity of participants."