Advanced Computing in the Age of AI|Sunday, March 7, 2021
  • Subscribe to EnterpriseAI Weekly Updates:  Subscribe by email

Former Qualcomm CEO Signs On As Advisor to 5G Chipmaker EdgeQ 

EdgeQ, a Silicon Valley startup that’s building a customizable RISC-V system-on-a-chip (SoC) that includes a 5G modem, got a big boost today when it announced that former Qualcomm CEO Paul Jacobs has signed on as an advisor to the company.

EdgeQ was founded three years ago by former Intel and Qualcomm engineers with the goal of developing an efficient and flexible enterprise 5G modem that could be used in industrial settings, backhaul telecommunications infrastructure, and other IoT and edge use cases. It is not looking to place its 5G modem in handsets, a market dominated by Qualcomm and Huawei.

Just the same, EdgeQ competes to an extent with Qualcomm, which has just started to branch out beyond handsets with its 5G modem, with automobiles among the first planned use cases outside of smart phones. But executives with EdgeQ tell Datanami they have a two-year head start on Qualcomm when it comes to developing enterprise 5G modems.

“Qualcomm’s modus operandi is they take a cell phone modem and extend that modem to the enterprise,” says Adil Kidwai, EdgeQ’s vice president and head of product management. “And that modem is not very efficient from a cost and energy perspective.”

EdgeQ says that, by basing its SoC on the open RISC-V platform, it is able to give clients the ability to pick and choose exactly what elements to include in the system. That is a novelty in the world of enterprise modems, says Kidwai, who spent 14 years at Intel, including director of AI for the Nervana Systems group.

“If you buy a modem right now from any company, you get the full stack,” Kidwai says. “We want to make a modem for enterprise which is open, meaning if a customer wants to differentiate by putting their own algorithms in the modem, they can do that with us. Nobody else in the world has an open modem.”

The EdgeQ SoC can run replace multiple standard components for a cell network base station.

To make its chips customizable, EdgeQ extended the 3D vector instructions included in the RISC-V chip, and then exposed those instructions to customers as APIs via a custom library. In addition to a modem–which was by far the toughest engineering challenge for EdgeQ–the SoC can also be used for machine learning workloads. EdgeQ anticipates some customers choosing to use EgdeQ’s stack as-is, while other more sophisticated organizations will want to customize large parts of it with their “secret sauce.”

In addition to being open, EdgeQ’s SoC chips are 50% more energy efficient than competing offerings, the company claims. That gives the company a sizable advantage over Intel when it comes to staying within the energy and thermal envelope for industrial IoT and edge use cases, says EdgeQ CEO Vinay Ravuri.

“It’s hard to compare against Qualcomm. They don’t have anything there yet,” Ravuri says. “But if we’re taking existing base stations that use Intel FPGAs and Xeon processors–nobody ever said Xeon and FPGAs are low power. They’re the definition of power [hogs]. These are several hundred-watt devices that go into these base stations. We’re an order of magnitude lower.”

EdgeQ plans to start shipping its SoC 5G chips later this year. It’s planning to sell to OEMs that will include its chip in the telecommunications gear that enables 5G coverage for AT&T, Verizon, and Sprint customers. EdgeQ says its SoC chips can provide the processing necessary for mid-haul and front-haul workloads in cell site base stations, and do so by consuming 50% less power than current solutions.

EdgeQ envisions its 5G chip being used for multiple use cases and industries.

The company is also targeting more novel workloads, including factory automation, robotics, video security, and other edge use cases where 5G’s reliability and millisecond-level latency can provide a competitive advantage. 5G and AI go hand-in-hand with these outdoor use cases in oil and gas, shipping, and other use cases, Kidwai says.

“We are making a Ferrari when it comes to number crunching. It can crunch a lot of number at a very fast speed,” he says. “And because it’s programmable, we can run AI on that as well.”

There are just a handful of companies developing 5G modems at the moment, and EdgeQ is one of them. That puts it in very exclusive company, and it’s also what drew the attention of Paul Jacobs, who was CEO of Qualcomm from 2005 to 2014 and is now the head of XCOM.

While his family is indelibly linked to the San Diego, California chip giant, Jacobs is no longer officially affiliated with the company after being ousted as a board member and its executive chairman in 2018 following an attempt to buy the company amid the Broadcom hostile takeover saga.

“It’s really because of experience,” Ravuri says of gaining Jacobs’ seal of approval. “We’ve had experience and battle scars of building [modems] for generations, so we have a pretty sizable team. That’s why Paul and company joined us.”

The EdgeQ team learned from past mistakes in developing at the physical layer, the all-important PHY, and figured out how to move the ball further ahead. “We learned what to do here and what not to do, what to avoid,” Ravuri says. “In doing so, we’re able to do something fresh and completely groundbreaking and new with an approach that existing companies can’t do, encumbered by the legacy that they have.”

We’re on the cusp of having new applications and services with 5G technology, according to Jacobs. “This inflection point creates the opportunity to build and leverage an open ecosystem based on new platforms and a greater diversity of participants,” he states in a press release. “EdgeQ’s solution, based on the open design of RISC-V processors, enables innovation deeper in the wireless technology stack. This will allow both general performance improvements as well as the design of wireless systems that are tailored to specific use cases.”

Joining Jacobs as an advisor to EdgeQ is former Qualcomm CTO Matt Grob.

This article first appeared on sister website Datanami. 

About the author: Alex Woodie

Alex Woodie has written about IT as a technology journalist for more than a decade. He brings extensive experience from the IBM midrange marketplace, including topics such as servers, ERP applications, programming, databases, security, high availability, storage, business intelligence, cloud, and mobile enablement. He resides in the San Diego area.

Add a Comment

EnterpriseAI