SDVerse Accelerates ‘SW Sourcing’ for SDV
INTERVIEW: Prashant Gulati, CEO of SDVerse
2024년 11월호 지면기사  / 한상민 기자_han@autoelectronics.co.kr



INTERVIEW Prashant Gulati, CEO of SDVerse

Last year, the automotive industry spent approximately $30 billion on software, but only 10% of that was actually allocated to software that was separated from hardware. Software sourcing is typically done by either developing it in-house, relying on a preferred supplier list, or using Google search and attending conferences. This process can take months. For sellers, even Tier 2 suppliers can take up to eight years to get in front of an OEM. SDVerse aims to break down the opaque and closed practices of the automotive industry related to software, and revolutionize the software procurement and purchasing process for software-defined vehicles. The new sourcing approach offers benefits such as reduced costs, time, and complexity, improved software quality, better resource allocation, and increased vendor revenue. I spoke with Prashant Gulati, CEO of SDVerse, at The Autonomous to learn how the SDVerse marketplace can provide these advantages.

written by Sang Min Han _ han@autoelectronics.co.kr

한글로 보기
SDVerse, SDV 위한 소프트웨어 소싱의 혁신 (autoelectronics.co.kr)
GM-마그나-위프로, 자동차 SW 마켓플레이스 SDVerse 개발 (autoelectronics.co.kr)
TTTech Auto가 The Autonomous의 중심인 까닭 (autoelectronics.co.kr)







Traditional OEMs and Tier 1 suppliers in the automotive industry have already initiated SDV (Software-Defined Vehicle) initiatives based on their strategies. At the same time, they are building ecosystems with stakeholders and recognizing the importance of open-source solutions. Where do you think the automotive industry stands today in terms of SDV deployment?  
Gulati      
 Let me break this into two parts. First, I’ll talk about software, and then I’ll discuss SDV, which is more than just software. Last week, we published a white paper that includes estimates showing that the industry spent about $30 billion last year on software alone. This figure is expected to grow to nearly $60 billion by 2030, essentially doubling. However, only 10% of that $30 billion is actually spent on software that’s decoupled from hardware. So, only about $3 billion of the $30 billion is software that’s sourced separately from hardware.

I mention this because one of the most critical aspects of SDV is the decoupling of software from hardware. Currently, we’ve made only about 10% progress towards SDV from that perspective. Beyond software, SDV also requires advancements in electronics and other areas. As I mentioned in my presentation today, we need new architectures, supply chains, talent, processes, and more. There’s a lot of work the industry still needs to do.

Some companies are a bit further ahead on the architecture front, meaning they’ve already begun planning. Many startups, particularly in China and the U.S., like Rivian and Lucid, have an advantage because they’re starting with a clean slate and aren’t tied to legacy architectures. This makes it easier for them to make rapid progress. For example, Rivian recently redesigned its entire E/E architecture, going from 80 - 130 ECUs to just 17 ECUs.

So, to answer your question, in the current setup, only about 10% of software is procured separately from hardware, meaning there’s a lot of work to be done to fully realize SDV.

Secondly, startups like NIO, which can create clean-sheet architectures, are able to capitalize on SDV capabilities more easily. Established OEMs, on the other hand, need to put in significant effort to develop new architectures. Some companies, like Ford, have attempted to insource everything related to new E/E architectures - they even set up Skunkworks organizations to tackle these challenges, as announced last year. GM, on the other hand, is trying to avoid insourcing everything, while VW originally planned to do so but has now shifted its focus to partnerships.

In short, every company has a different strategy, and many have experimented with both insourcing and partnerships. There’s no doubt that software is an area of investment and a key driver of value in vehicles. Everyone is working on it, just at varying levels of maturity.


I think that’s why you also mentioned how existing OEMs are trying to solve these problems independently, like CARIAD, but are now shifting their focus towards more partnerships. Right? 
Gulati      
 The last thing I will say is that there is recognition among established OEMs and new ones that software presents a tremendous opportunity. To their credit, OEMs are changing quite rapidly, as seen in their partnerships with Google, Android Automotive, and other industry players, including collaborations with TTTech Auto.




The SDVerse marketplace was created because internal sourcing is an inefficient use of resources. There isn't enough time, capital, or talent to handle everything in-house. SDVerse encourages the industry to look for software on the platform before developing it in-house.



I’ve come across a lot of information about SDV initiatives focusing on the standardization of technologies and APIs, such as AUTOSAR, COVESA, Eclipse SdV, and SOAFEE. However, SDVerse is a marketplace. How does SDVerse differentiate itself from other SDV initiatives in the industry? 
Gulati        
Let me summarize how things work today. The consortia mentioned here - COVESA, AUTOSAR, SOAFEE, and Eclipse - have done a great job so far. Without the standards they set, there would be no interoperability. Now, let’s consider what happens in the industry today: a car typically has between 50 and 150 ECUs, each of which has software supplied by different vendors.
As the industry has evolved, particularly with the influence of smartphones and advancements made by Tesla, people have come to appreciate the idea of over-the-air updates and connectivity. However, implementing over-the-air updates across 80 different modules is not feasible due to interdependencies among them. Standard bodies have addressed the interoperability issue to some extent and are focused on thought leadership.

However, established OEMs, as you pointed out, are using the standards set by these organizations. While this is beneficial, many of these modules come from different suppliers, and they don’t always communicate effectively with each other. As a result, OEMs have begun to insource more functions. The issue here is not with the standard-setting bodies but with the existing dynamics of the supply chain.
The SDVerse marketplace was created because insourcing is not a very efficient use of resources; there isn't enough time, capital, or talent to handle everything in-house. We encourage the industry to find software on our platform before attempting to develop it themselves, regardless of whether they are using these standards. We promote the reuse of software.

Unlike other organizations, we are not focused on creating new standards; we are dedicated to helping commercialize software. If I had to summarize the key difference in one word, it would be this: they focus on thought leadership, while we focus on commercialization and improving reuse.

For instance, if an OEM decides to insource everything, AUTOSAR may not be concerned, but we believe this approach leads to an inefficient allocation of capital. We want OEMs to search for software on our platform, enabling them to concentrate on developing differentiated software. All non-differentiated, reusable software can be found elsewhere. This distinction between commercialization and thought leadership is crucial. We welcome any software that complies with these APIs on our platform.


Listening to you, it makes sense why OEMs are focusing on sharing major software. It also seems like they are trying to recover the costs they've already invested in software. 
Gulati      
 Exactly. Yeah, that's a great point. That's actually what SDVerse is focusing on. However, on the other hand, I was thinking that the only software you can find in the SDVerse market might just be ordinary. The nature of software is that it becomes a commodity pretty quickly. So, consider something you might spend tens of millions of dollars developing today.

If OEMs are selling software, for example, you create software and establish your own IP, then release it in the market for one OEM. What is going to stop someone else from doing something similar? You've seen features that show up in one car, like using Bluetooth to turn your phone into a key; it's available in every car. If you don't monetize it, then everyone will be doing this themselves. What’s the point of doing everything in-house?

So, you could sell it for the first six months and also sell it to others on the marketplace. You wouldn't give it away for free, but you could commercialize it by charging some licensing fees.


Next question: Shouldn't technology standardization come before the marketplace? How does SDVerse view the relationship between standardization and commercialization?
Gulati      
 You are absolutely right: ideally, you should standardize first, then build, and then sell. That’s how it should be. However, the nature of software is such that it evolves so quickly that if you wait for everything to be standardized, you will never accomplish anything. This is different from hardware. The same ECU that you developed 20 years ago has evolved a little, but you can only create standards around a select few items. When it comes to software, it's much more difficult. If you keep waiting for software standardization, we will halt all progress.

What we are proposing is to create transparency around software that has already been developed. This means that we are not saying the software listed on our platform is just available for download or plug-and-play like an app on a phone. Instead, we provide information about which compatible products this software works with and what other dependencies exist, etc.

While I will discuss the benefits of SDVerse in a moment, ideally, standardization should come first. However, software is moving fast, and
one of the benefits of SDVerse is that speed is critical. We can’t just sit and wait for standardization to happen. Standardization bodies have been around for over 30 years, and we still face interoperability issues. This is why, when GM, Magna, Wipro, and others were considering this, they evaluated creating new standards or advocating for more of them and realized that this experiment has been ongoing for 30 years. We need something else.


When looking at the founding members, we see companies like Magna, HL Mando from Korea, and Valeo from France. Most of them are chassis suppliers related to safety. However, SDVs are currently also being driven by high-performance computing units, particularly for entertainment purposes. Does SDVerse also cater to this market?
Gulati      
 we have the SDVerse platform, which allows us to list onboard software, offboard software (which includes cloud services), and tool end services, right? We also have list software for battery management, autonomous driving, or ADAS functions. Additionally, you can list software for powertrain management, CHiP design, reference designs, and more.
As for our current customers, NXP is one example; they produce CHiP sets for a wide variety of applications. We have a layer that supports a diverse range of applications. TTTech Auto is another example, as they focus heavily on driving safety. We also have buyers like Cummins and GM. Bosch Engineering is one of our sellers, and while we haven’t announced this publicly yet, I'm sharing it with you. We also signed a partnership with PopcornSAR, a company in Korea. Of course, HL Mando is also one of our partners.

As you can see, we have a broad representation of existing customers on our platform, including companies like FEV and Forvia. Magna, too, is involved in many areas beyond just chassis control, with a strong focus on the software of the future.
The point is that we have a diverse and wide representation of existing customers. On September 30th, we will go live with the platform and demonstrate how the marketplace will operate. We’ll start engaging in conversations with more companies as well.

Currently, we already have over 100 software applications listed on the platform, which we'll showcase on Monday. These cover various areas, including individual components from autonomous driving to full-stack applications. So it’s not just commodity software; it’s not merely non-differentiated software; it provides creative transparency about software for buyers.

For instance, some companies in Europe or Korea may not work with many U.S. customers, but we are trying to get these customers to sign up on the platform. When they search for a specific type of software, these companies will appear in the results.
To answer the question, we do have a diverse range of software available, and it’s not focused on just one thing.


Let’s assume there are already over 100 software applications listed on the platform. If a customer encounters an issue when using it in series production, how would you handle that?     
Gulati      
 So, we are not a software download platform. We provide information about the software, not the actual software. I’ll give you an example that I don’t particularly like, but it tested very well in our focus groups. Do you know Tinder, the matchmaking dating app? People refer to us as the Tinder of automotive software. What that means is that a buyer, like GM or Cummins, is looking for software, and we help them find providers like TTTech Auto or NXP. Then we facilitate the matchmaking. Buyers and sellers can communicate on the platform, ask questions, and issue RFIs.

However, we are not taking over the actual supply chain for GM or Cummins; they still have to communicate with the suppliers to request information. All of that work still has to happen offline. So, yes, it’s a marketplace for matchmaking. It’s not an app store, and it’s not like GitHub. We’re not downloading software here.
The company owns a software component and lists it along with certain dependencies and limitations. When a buyer comes in, they check the information; if it seems accurate or feasible, they contact the source directly.


The procurement method is really about speeding up the process, isn't it?  
Gulati      
 Correct, absolutely right. Now, think about why this is important or helpful. I’ll give you two use cases. One is, how do you write software or obtain it today? A product person will define a feature, and an engineer will be responsible for delivering that software. They will either write it themselves or source it from somewhere else. How do people obtain software from other sources? They typically have a preferred list of suppliers, perform Google searches, or attend conferences - these are the three normal ways, or platforms like GitHub.

However, if you call your supplier, unfortunately, they often provide an ECU that comes bundled with software that is no longer what they want to offer. Google searches can be unreliable, and attending conferences is time-consuming. Consequently, any piece of software that needs to be developed can take months of discovery. Our collaboration with GM and others has shown that first you conduct discovery, then you issue a Request for Quotation (RFQ), and then a Request for Information (RFI). This whole process can take about four months in a 36-month vehicle program.

We are claiming that we can help you reduce your discovery process from months to minutes. Why? Because we have worked with suppliers like TTTech Auto, NXP, and others to standardize comparison criteria. We have a set of 180 attributes that each supplier provides about their software, which we use to compare solutions across suppliers.



CEO Gulati participated as a panelist on the exciting and rapidly evolving topic of risks and opportunities in SDVs at The Autonomous, alongside Michael Fait, Max Lemke, Ethan Sorrelgreen, and Stefan Poledna (CTO, TTTech Auto).  



Does SDVerse have a verification process or procedure to ensure that the software listed in the marketplace is genuinely trustworthy? I'm particularly curious because there are safety-critical products involved.   
Gulati      
 So, we have what we call a KYC (Know Your Customer) check. As I mentioned, we are not taking over any buyer's procurement function; GM, for compliance or other reasons, is not handing over their procurement function to us. In our marketplace, we have established a few barriers to entry to ensure that only quality companies participate. The first barrier is a subscription fee. If a company agrees to that fee, we then conduct a KYC check to verify that they are a valid entity and perform standard checks to ensure there are no export controls on their software.

I don’t know the details of the process that our law firm follows, nor do I know what BLU states. However, the government may impose bans on certain regions as an example, and we are more sensitive to export control software.
In the marketplace, we create transparency. This means we provide space for companies to submit their certification documents. We ask for a lot of information, such as how many vehicles use their software and the level of certification. While these are quality documents, it is a self-certification process. We have established transparency, and the procurement organization will verify this information, or anyone else can review and challenge it.

There is an incentive system in place; for example, if someone lies about a detail and another person raises a complaint, that could lead to issues. While we don't have an official rating system, we prioritize transparency.
Lastly, sometime after our launch next year, we plan to introduce certifications to further ensure quality. However, as of now, we haven’t implemented this, and based on customer feedback, it isn’t their top priority because they will conduct these checks regardless. Their main concern is finding the information they need, which currently takes them months. I understand there is work to be done next year regarding software certifications, but right now, we're focused on solving a different problem.

The self-certification process is very interesting. It’s a standard practice because, regardless of SDVerse, any supplier wanting to get set up in a buyer's system will need to provide a lot of information.


Have new companies like PopcornAR or Bosch Engineering gone through the same process?   
Gulati      
 It’s the same thing - we ran a KYC process for them, but we’re not certifying their software. Yeah, we’re not doing that for anyone. One of the things we’ve been asked by GM, especially, and some others, is that they want new, interesting companies on the marketplace. There is some level of qualifying that we do, but they just want to learn more. They actually want to use the marketplace on a daily or weekly basis to find new solutions and new software. As I said, we have a few barriers to entry, from subscription fees to the KYC process, self-certification, and providing your certificates. Keeping all this information transparent allows procurement departments to verify whether it’s accurate. And sometimes, based on feedback, we have an advisory council and focus groups with our customers. If there’s a need for certification, we do have it on our roadmap, but the timing will depend on customer demand.


Is it possible that autonomous application software, such as Level 3 or Level 4, might not be or could not be listed on SDVerse due to export control regulations? Many countries designate such technologies as part of their national core competencies or national assets, and they strictly control the export or leakage of these technologies abroad.    
Gulati      
 I have not heard that before. It could be possible I'm not a technical expert.We've not come across this problem yet because like I said, I mean, you know, companies sitting in here like you are based out of Europe, you're based out of Korea, I mean TTTech Auto is in Europe that do supply solutions globally. I think mostly where we've kind of run into issues or learned about these issues is where either you're in Russia or in some cases China.


SDVerse claims to be open to everyone. However, it's clear that SDVerse started with founders like GM and Magna, and major launch partners include notable French carmakers and Tier 1 suppliers. There must be a reason for that, right? What are your expectations for SDVerse?    
Gulati      
 That's correct. The entire supply chain is globalized, and we are adapting accordingly, but software has its own set of limitations. As you mentioned, there are specific regulations in Korea. Recently, there was an announcement that the U.S. is considering restrictions on software based in China. We're not entirely sure what that means yet. While a lot of software comes directly from China, some also comes through controlled enterprises in Europe. As I mentioned, we will comply with regulations.

We definitely aim for global expansion. Looking at our current customers, GM is headquartered in the U.S. with a global presence, and Magna is the same. Wipro is mainly based in India but has a global footprint, while HL Mando and PopcornAR are based in Korea, and Valeo is a global player. NXP, TTTech Auto, FEV, Forvia, and Ampere are all in Europe. So, we have a good mix, but this is all before we launch the live platform. Once the live launch begins, we will start expanding globally and showing it to the world. Our advantage is that distribution is global. This is one of the reasons why the software and internet industries have grown exponentially - distribution has become much easier. Of course, the automotive industry is more complex due to regulations, certification compliance, and compatibility issues, but it’s still easier than hardware.


We are here at The Autonomous. What do autonomous driving levels 2, 3, and 4 mean for the SDVerse marketplace? What role does TTTech Auto play in SDVerse, and what does it represent?        
Gulati      
 TTTech Auto is a launch partner. Like all of our launch partners, we have been meeting every week for the last six months. They are part of what we call the advisory council. While TTTech Auto and other launch partners provide us with strategic advice, they don't make decisions for the platform. Secondly, there is the focus group. The advisory council and focus groups consist of technical experts from launch partners who join the discussion. My team and I show them our approach to building the platform, and the launch partners provide feedback. For instance, they may point out that we’re missing a reference design in a chipset or that some terminology is outdated. These are the kinds of discussions that happen. So, TTTech Auto’s contribution is helping shape the platform.
From my personal perspective, TTTech Auto has been very active in promoting and discussing SDVerse as a major collaboration for the industry. We are grateful for TTTech Auto’s recognition of SDVerse as a collaborative platform.





Gov. Whitmer officially declared it Automotive Software Discovery and Innovation Day in honor of SDVerse's groundbreaking product launch event. This special recognition highlights the pivotal role of innovation in the automotive software industry. As a founding member of SDVerse, alongside General Motors and Wipro, Magna is at the forefront of driving the next wave of mobility solutions with cutting-edge software technologies. From enhanced safety features to software-defined vehicles, with over 160 products live on the platform, this is just the beginning of an exciting journey. From left to right: Prashant Gulati, CEO of SDVerse; Jorg Grotendorst, SVP of Magna; Sonal Ramrakhiani of SDVerse; Daniel Nicholson, Vice President of GM; and Jeff Walker, CCO of SDVerse.



Lastly, could you give us a sneak peek of the September 30th launch event?
Gulati        
Lastly, there is something I can’t talk about right now, but I will be able to share it soon.
It relates to the software being listed on the platform for global availability and visibility. One of the major benefits of the marketplace is that, as a seller, I know of Tier 2 suppliers who have taken up to eight years to get in front of an OEM - this is not uncommon. Now, those same sellers can significantly shorten their sales cycles. It's all about awareness. We're creating awareness of products. Unlike hardware, where the environment is highly controlled, software talent and capabilities exist in many different places - Vienna, Korea, India, Pakistan, and so on. The beauty of a software marketplace is that it allows you to be introduced to companies from around the world.

As I mentioned, the benefits for sellers include shorter sales cycles and more targeted access. This is particularly relevant today because organizations - whether OEMs, Tier 1s, or Tier 2s - are undergoing transformations. New people from Silicon Valley are entering these organizations, and some employees are changing roles or being retrained. For example, you may have been speaking to someone earlier, but that person may no longer be in the role, and a new person - someone from Silicon Valley - might have taken over. For instance, someone coming from Apple, Amazon, or Microsoft might not be familiar with the automotive industry and may not know you because they’ve never worked with you before. How will they approach software? They might try to build everything themselves, but is that good for the industry? If everyone builds everything on their own, it will disrupt the supply chain. Essentially, we are working to prevent the supply chain from being disrupted.

For sellers, the benefits are shorter sales cycles and more targeted access. For buyers, the benefit is that they don’t have to do everything on their own - they can focus on their core activities.

Our founding members - GM, Magna, and Wipro - will discuss why they care about SDVerse, their outlook for the platform, why they invested in the company, and what they expect the industry to gain from SDVerse. They will answer some of the same questions you asked, but from their own perspectives - GM from the OEM perspective, Magna from the Tier 1 supplier perspective, and Wipro from the engineering services perspective. Everyone has different views on the opportunities and challenges in the industry and how SDVerse can help solve them. They’ll share their perspectives and will also focus on how they plan to use SDVerse. For example, I mentioned how GM is thinking about using SDVerse in their organization.

We will have a live event where we’ll recognize our partners, including TTTech Auto, Valeo, NXP, and others. There will be a live platform demo, so instead of just talking about it, during the event, you can actually see how the platform will function. I will send you a link for the live event. It might be late at night for you, but you can register and receive the recording later. Additionally, we will be announcing some new customers.



<저작권자 © AEM. 무단전재 및 재배포, AI학습 이용 금지>


  • 100자평 쓰기
  • 로그인


  • 세미나/교육/전시

TOP