Qualcomm Officially Enters Self-Driving Market With Snapdragon Ride Platform And Extends Partnership With GM To Include ADAS

By Patrick Moorhead - January 24, 2020
Qualcomm Snapdragon Ride platform 

Known traditionally for its ubiquitous mobile semiconductors powering most of the planet’s smartphones and tablets, Qualcomm has leveraged its expertise and IP to elbow its way into other emergent sectors—namely, automotive, IoT, and datacenter edge ML. I’ve covered these automotive moves fairly extensively, from its CES announcements last year, to last week when I shared some high-level insights into Qualcomm’s automotive strategy from the company’s recent financial analyst day.

Today, in Las Vegas, at CES 2020, Qualcomm made what I think is the company’s biggest announcement yet in the automotive market: the unveiling of its new Snapdragon Autonomous driving platform for self-driving cars, called the “Qualcomm Snapdragon Ride”. In addition, Qualcomm announced it is extending its partnership with General Motors that will now include ADAS solutions. Let’s take a closer look.

Qualcomm Snapdragon Ride platform

Qualcomm describes the new platform as “the automotive industry’s most advanced, scalable and open autonomous driving solution.” That’s a big statement and a big way to start off, right? The platforms overall purpose is to enable automakers to better address the increasing complexity of autonomous driving. These complexities are big, ranging from the variance of needs from L1 to L5, internetworking across heterogeneous modules, connectivity to the dashboard, and to the telematics and now a connected cloud system.

Qualcomm’s CES 2020 L2+ demo platform using current gen 

The platform includes heterogeneous capabilities ranging from image signal processors for camera sensors, dedicated DSPs (digital signal processors), GPU technology leveraged for visualization and high-end user experiences, and dedicated safety and security systems. Additionally, and very new, the platform includes an autonomous driving accelerator, an ML ASIC, to provide the highest level of performance at the lowest energy use. Low power and low heat matters a lot in the future of automotive as many of these cars will be electric and any car can save on active cooling systems.

Along with Qualcomm’s new Snapdragon Autonomous Ride software stack (also announced today, more on this later), the platform seeks support three segments of autonomous systems:

  1. Active Safety ADAS (autonomous braking, traffic sign recognition and lane assist). I see this as L1 and L2. This is addressed with one passively cooled ADAS chip delivering 30 Tera Operations Per Second (TOPS).
  2. Convenience ADAS (Automated Highway Driving Self-Parking and Urban Driving with Stop-and-Go apps). I see this as L2+. This is addressed with the combination of SoCs and is expected to be anywhere in the range of 60 TOPS to 125 TOPS.
  3. Fully Autonomous Driving (autonomous urban driving, Mobility-as-a-Service).  I see this as L4-L5. This is addressed with two ADAS chips and one to two ML accelerators, all actively cooled, but “no need for liquid cooling”, delivering up to 700 TOPS at 130W. That 700 TOPS number includes two ADAS and two ML accelerators.

The TOPS per watt (5.4) Qualcomm says it can deliver at the system level (higher at the raw silicon blocks) at its peak is mind-boggling. While the peak 700 TOPs isn’t the highest announced for a system, I believe there’s a potential to connect subsystems near-linearly. Therefore, I could envision something like 2,800 TOPS at 520W if Qualcomm wanted to do a drag race.  The other thing that we don’t know for certain is efficiency. I’m assuming Qualcomm’s design is more ASIC-like than GPU-like and therefore my expectation would be higher efficiency for those TOPS. But we’ll have to wait for more numbers given there isn’t some industry-standard self-driving benchmark we can run.  

If Qualcomm can scale and support all of these levels, from safety to comfort to fully autonomous, and do it with the highest levels of safety, the platform should enable automakers to quickly and efficiently deploy autonomous driving systems spanning all mass-market vehicle tiers.  And that’s a very big deal for the industry.

Snapdragon Ride sensor support for L2+ 

Software stack

A comprehensive autonomous driving stack also comes integrated with the platform, which includes software and applications designed to assist in complex autonomous driving issues such as self-navigating human-like highway driving. Qualcomm says the software stack is modular and scalable and can be co-hosted on the platform with customer-specific stack components. Qualcomm developed the stack to allow OEMs and Tier 1 suppliers to pick and choose the elements of the stack that best suit their specific needs.  When Qualcomm uses the term “open” this is what it means and is in contrast to other solutions that are more “black box” where the vendor owns the entire stack with minimal OEM value-add.

When I talk with Qualcomm, it is well aware that software is more than half the battle, a good sign, as many players aren’t there yet. People tend to forget that Qualcomm was the trailblazer with heterogeneous compute across CPU, GPU, and DSP and almost a decade ago released SDKs for it. From an ML-perspective, Qualcomm already optimizes its mobile stacks for Tensorflow, ONNX, etc., and I don’t think this is a weakness for the company. Given how “fresh” Qualcomm’s Ride stack is, I believe this will be an area the competitors will pick at but could be quickly forgotten about given one or two big announcements (like GM) with customers and stated safety certifications.

Qualcomm Ride Autonomous software stack 


Obviously, when we’re talking about autonomous driving, security concerns are paramount, and safety, like software, is an area I expect Qualcomm’s self-driving competitors to pick at. Qualcomm says the Snapdragon Ride SoCs and accelerator are designed for functional safety ASIL-D systems.

One thing to consider here is that this isn’t Qualcomm’s first automotive rodeo. As I wrote about here, Qualcomm has been engaged and its chips shipped inside millions of cars and in areas of safety like telematics. Remember that GM OnStar button and the system that automatically notifies if the car was in an accident? Well that call and data exchange was based on Qualcomm silicon.

Qualcomm automotive customers. 

According to Qualcomm, the platform includes an integrated safety board support package as well as safety frameworks like adaptive AUTOSAR. Some of the core benefits of the platform, as cited by Qualcomm, include AI tools to improve model efficiencies and runtimes, optimized foundational function libraries for the likes of computer vision, signal processing, and standard arithmetic, Qualcomm Vision Enhanced Precise Positioning for cost-efficient localization, and data management tools.

I was struck at how many Qualcomm partners came out of the woodwork discussing safety, including QNX, Arm, Synopsys, and Infineon. That’s a good sign when another company sticks their neck out like that. It’s a vote of confidence for Qualcomm.

I’m looking forward to researching more on Qualcomm’s self-driving safety features and processes.

GM expands automotive relationship with Qualcomm ADAS

Qualcomm isn’t just throwing out ADAS technology and hoping to secure a customer. Qualcomm announced today that it was extending its partnership with GM to now include ADAS, in addition to connectivity and infotainment.  Details of the partnership were not included, but it seems to me that if you read between the lines, by launching the platform and extending the partnership could result in us seeing GM cars with Qualcomm ADAS in 2H 2023, based on generic Qualcomm-provided dates.  

Wrapping up

This announcement is huge as it now pits semiconductor giants Intel, NVIDIA, and now Qualcomm, against each other in the self-driving space, each with certain advantages. And, we can’t forget legacy automotive vendors like NXP, STMicro, Texas Instruments and even full-stack providers like Tesla, either.

As I wrote earlier this week, Qualcomm’s extensive hardware and software IP, originally developed for mobile, gives it many advantages that the legacy players in the industry will struggle to match, and there’s a reason the company currently has an order pipeline of over $7B (increased from $6.5B two months ago) as it looks to be dominating the modern, premium dashboard and telematics.

Investors should like Qualcomm’s leveraged automotive R&D model as well, as I discussed here.

I’m looking forward to more disclosures with more performance and power details and safety information over the coming months and years.

Qualcomm expects the new platform to be in production cars on the road in 2H 2023 with pre-dev units available to automakers and tier-1 suppliers in the first half of this year.

+ posts

Patrick founded the firm based on his real-world world technology experiences with the understanding of what he wasn’t getting from analysts and consultants. Ten years later, Patrick is ranked #1 among technology industry analysts in terms of “power” (ARInsights)  in “press citations” (Apollo Research). Moorhead is a contributor at Forbes and frequently appears on CNBC. He is a broad-based analyst covering a wide variety of topics including the cloud, enterprise SaaS, collaboration, client computing, and semiconductors. He has 30 years of experience including 15 years of executive experience at high tech companies (NCR, AT&T, Compaq, now HP, and AMD) leading strategy, product management, product marketing, and corporate marketing, including three industry board appointments.