• Both NVIDIA and Qualcomm are working to exploit autonomous driving and the in-vehicle experience.
• NVIDIA has announced a new SoC – Drive Thor – which will replace Drive Orin.
• Qualcomm has introduced the Snapdragon Ride Flex SoC portfolio, the “industry’s first super-compute class SoC portfolio".

The automotive industry is moving towards software-defined vehicles. We expect a centralized approach to the powerful processors that will be needed to manage the data flowing from the various sensors, and running machine learning (ML) models for safe automated driving.

NVIDIA and Qualcomm have pivoted their business to address the automotive opportunity as the traditional car is becoming an extension of computers and smartphones. Both companies are working to exploit autonomous driving and the in-vehicle experience. As the car becomes more intelligent, these two areas have the highest growth opportunities.

NVIDIA recently held its GTC event and Qualcomm an investor day focusing on its automotive business. In this blog, we will look at key announcements from both players in relation to autonomous driving and the digital cockpit.

NVIDIA GTC 2022

From September 19 to September 22, 2022, NVIDIA held its bi-annual event called GTC, where it announced a new SoC – Drive Thor – which will replace Drive Orin. Earlier, at GTC 2021, NVIDIA unveiled its plan for Drive Atlan SoC to target 2025 models and replace Drive Orin. But Atlan appears to have been scrapped. So, Thor will be ready for vehicles arriving in 2025. The Thor processor is packed with high-power computing to support the idea of having a single chip to handle autonomous driving and cockpit functions. To put this into perspective, Thor can handle 2,000 trillion operations per second (TOPS), which is twice the speed of Atlan and eight times the speed of the current Orin processor.

                                                                                           NVIDIA DRIVE SoC Roadmap

NVIDIA has already announced its first partnership for Drive Thor with Chinese car company Zeekr. Thor will be first used in an EV model from Zeekr in 2025. Currently, the Drive Orin chip is used in models including the Li Auto L9, Nio ET7 and the recently launched Xpeng G9.

Qualcomm Automotive Investor Day

Qualcomm held its first-ever Automotive Investor Day in New York immediately after the conclusion of NVIDIA’s GTC. Qualcomm, whose automotive business is flourishing thanks to its connectivity and digital cockpit solutions, has seen the majority of the design wins since the launch of its Snapdragon Digital Chassis. During the event, the company introduced the Snapdragon Ride Flex SoC portfolio, which Nakul Duggal, Qualcomm's Senior Vice President and General Manager of Automotive, called the “industry’s first super-compute class SoC portfolio”. This will be a family of chips that helps Qualcomm to scale down from high tier to entry tier. The software architecture of the chips will be Service Oriented Architecture (SOA) compliant. The company has not revealed much information on the Flex SoC, but the graphics displayed during the event show that the Flex SoC will have around 2,000 TOPS of computing power to handle operations like the cockpit, driver assistance, automated driving, and connectivity functions.

                                                                                                                Snapdragon Ride Flex Soc Family

                                                                                                           Central Compute Scalable Architecture

Another announcement that underscores Qualcomm’s strong position in digital cockpit solutions is that starting in 2023, Mercedes-Benz’s in-vehicle infotainment (MBUX) will be powered by a Snapdragon chip.  Currently, Mercedes’ S-class MBUX is powered by NVIDIA chips. In the last 10 months, 20 models have been launched with Qualcomm’s digital cockpit solutions. Qualcomm sees a total automotive market of $100 billion by 2030 with ADAS/AD being the major contributor with around 60%, followed by digital cockpit with a TAM of $25 billion.

                                                                                                              Qualcomm Automotive TAM

Now, let us look at both companies to compare where they stand and which of them has the upper hand in the automotive industry:

Portfolio

Qualcomm has a little bigger portfolio as it has offerings in connectivity, in-vehicle infotainment and ADAS/AD domains. Qualcomm integrates all these services under one umbrella term called Snapdragon Digital Chassis. NVIDIA, on the other hand, offers solutions for the cockpit and autonomous driving but does not have a portfolio in connectivity. Hence, the battleground for the two semiconductor companies is autonomous driving and smart cabin features. NVIDIA calls its offerings for the automotive industry DRIVE.

Financials

In the second quarter of the fiscal year 2023 (FY23), NVIDIA’s automotive segment had the best quarter with revenue of $220 million, an increase of 59% QoQ and 45% YoY. FY20 saw the highest ever automotive revenue of $700 million, achieved due to the success of the Tegra processor, which once powered Tesla’s infotainment (media control unit) and instrument cluster. The automotive business has already reached the $358-million mark in the first half of FY23. With more models, like the Polestar 3 and Volvo EX90, driven by Orin chip ready to be launched this year, it is likely to cross $700 million by the end of FY23. During the Q2 earnings call, the company described Q2 FY23 as the inflection point for its automotive revenue.

                                                                                                     NVIDIA Automotive Quarterly Revenue

Qualcomm also recorded its highest ever automotive revenue in the third quarter of FY22 by reaching the $350-million mark. The automotive revenue is mainly driven by connectivity and an increase in demand for its digital cockpit products. This year, Qualcomm launched 20 new models with a digital cockpit. The future growth in the automotive business will be mainly driven by the digital cockpit. From FY26, the company is expecting its ADAS/AD services to scale up and experience a ramp-up starting in 2026.

                                                                                               Qualcomm Automotive Quarterly Revenue

Qualcomm is expecting to finish FY22 with $1.3 billion in automotive revenue, up from $975 million in FY21. At present, the cumulative automotive revenue for FY22 stands at $945 million. Qualcomm is realizing the segment’s growth potential with the current and future product roadmap. As a result, the company has increased its targets for FY26 from $3.5 billion to $4 billion. For FY31, it has revised its forecast revenue from $8 billion to more than $9 billion.

                                                                                                                      Revenue Forecast

Design-win pipeline

NVIDIA announced that its design-win pipeline had reached over $11 billion (in FY23) for the next six years from over $8 billion in FY22 due to the traction earned by the Drive Orin SoC. Currently, Orin has over 35 customers, which include automobile, truck and robotaxi companies.

                                                                                                                    NVIDIA  Design-win Logos

                                                                                                                  NVIDIA Automotive Partners

Qualcomm’s design-win pipeline has reached $30 billion, up from $19 billion two months ago, and has more than doubled in the last 10 months from $13 billion. As the automobile industry is moving towards more advanced driving systems and highly automated driving, most of the design wins are coming from ADAS/AD, followed by the digital cockpit. With over 50+ customers for the digital cockpit, ADAS/AD has not been able to mark its presence or gain traction from car companies or robotaxis like NVIDIA, barring a few customers like BMW, Stellantis, VW, Renault and Great Wall Motor. Qualcomm will announce the design win for the new Flex SoC at CES 2023.

                                                                                                                      Snapdragon Ride Design-Win

                                                                                                              Qualcomm Automotive Partnerships

Conclusion

NVIDIA has enjoyed early success with its chips like Parker (1 TOPS) and Xavier (30 TOPS), required to perform Level 2 and 2+ functions. The Xavier SoC can be seen on models like the Xpeng P7 and IM L7. NVIDIA, known for its high-power computing chips for gaming and data centers, leverages its technology to introduce high-power computing chips like Orin and now Thor. Due to the legacy of its GPU, AI chips and the availability of a high-power computing platform to support autonomous driving, it has generated interest from car manufacturers, robotaxi companies and self-driving truck companies like Kodiak.

Compared to its competitor, Qualcomm entered the ADAS/AD market late. During CES 2020, the company entered the self-driving market by unveiling its Snapdragon Ride Platform. The platform ranges from 30 TOPS for Level 1/2 functions to 700 TOPS for Level 4/5 applications. GM is the first company to use Snapdragon Ride for its hands-free assisted driving. The Ultra Cruise will be powered by two Snapdragon SA8540P SoCs and one SA9000 AI. The platform will be making its debut in the Chinese market in October as Great Wall Motor will start selling its Mocca DHT-PHEV LiDAR variant fitted with Snapdragon Ride. So far, Qualcomm has managed wins with few OEMs and has not been able to win any robotaxi or self-driving truck customers. With its latest announcement of Flex SoC, which will directly compete with Drive Thor, we expect Qualcomm to remedy this gap.

Both companies are adopting a similar approach of taking all their technologies and integrating them into a single chip to make it a central computing hub for the vehicle. However, the major difference is that NVIDIA is at the top of its game when it comes to advanced computing, while Qualcomm brings in a more integrated solution, from connectivity to cockpit to ADAS/AD, to get closer to NVIDIA in terms of compute capabilities but at relatively very low power. We believe that extreme low-power high computations with a very scalable architecture for battery-powered devices is Qualcomm’s biggest advantage over NVIDIA which has helped the company attract a design win pipeline almost triple of NVIDIA, as power efficiency for future EV-based vehicles is going to be the key.

It would be great to hear more about the power-versus-performance metrics which Qualcomm can achieve against NVIDIA in the months running up to CES 2023.