Accelerating SDV Development through Virtualization and AI

In this episode, host Suresh Sivavarman, Customer Chief Engineer at ETAS, sits down with Raj Paul, Mobility Industry Leader at Microsoft. Together, they explore the engineering shifts and technological transformations that are redefining the future of mobility.

The discussion focuses on how "cloudification" and the integration of AI are moving from industry buzzwords to essential drivers of competitiveness and development velocity. Raj shares insights into the power of data-driven decision making, the transition from on-premise tools to cloud-native engineering, and how AI-powered "co-pilots" are enhancing developer productivity.

Key topics in this episode include:

• The Shift-Left Mindset: How virtualization and simulation in the cloud allow for hardware and software development to run in parallel, drastically reducing development timelines.
• Breaking Data Silos: Leveraging cloud compute and storage to create end-to-end visibility across engineering, manufacturing, and service.
• The SDV Backend Transformation: Building the feedback loops and data pipelines necessary for over-the-air (OTA) updates and proactive vehicle monitoring.
• The Road to 2030: A look at the forcing functions, such as competitive threats from new-age OEMs, that will accelerate the industry’s digital evolution over the next decade.

Thanks for listening!

• Email us at: contact.us@etas.com
• Learn more about ETAS on our website 
• Follow us on LinkedIn: @ETAS

Transcript

00:00:02 Voiceover

Welcome to the Empowering Tomorrow's Automotive Software Podcast, brought to you by ETAS, a single source of cutting-edge software and hardware solutions that make automotive embedded systems safe, smart, secure, and sustainable.

00:00:15 Voiceover

Each episode, we'll be joined by ETAS and industry experts to discuss how electrification, automation, and connectivity are impacting the automotive industry.

00:00:25 Voiceover

Now, sit back and enjoy the discussion.

00:00:32 Suresh Sivavarman

Hello all, and welcome to a new episode of Empowering Tomorrow's Automotive Software Podcast by ETAS.

00:00:38 Suresh Sivavarman

We're exploring the technologies, transformations, and engineering shifts that are shaping the future of mobility.

00:00:44 Suresh Sivavarman

Today, I'm your host.

00:00:45 Suresh Sivavarman

My name is Suresh Sivaraman, and I'm a Customer Chief Engineer at ETAS.

00:00:49 Suresh Sivavarman

And I'm glad to be your host, and I'm looking to talk about different trends in the automotive industry with respect to cloudification and how the play of AI in that world.

00:00:58 Suresh Sivavarman

To set a context for today's episode, we'll be talking a lot about how engineering teams across the world, including leadership teams, are focused on four themes-- data activation, AI and engineering workflows, cloud-enabled shift left, and SDV backend transformation.

00:01:17 Suresh Sivavarman

What we know is these are not just buzzwords, but they're becoming core drivers of transformation and competitiveness.

00:01:23 Suresh Sivavarman

And it's also increasing the velocity of development.

00:01:26 Suresh Sivavarman

So today I have a guest with me.

00:01:28 Suresh Sivavarman

His name is Raj Paul.

00:01:31 Suresh Sivavarman

He's the mobility industry leader for the Americas and global industry leader in automotive at Microsoft.

00:01:37 Suresh Sivavarman

He's got over 20 years of experience in the automotive industry with in-vehicle innovations, connected services, and digital infrastructure, as well as transforming companies into the cloud world.

00:01:47 Suresh Sivavarman

Raj brings a powerful combination of technical depth and industry foresight in the modernization journey.

00:01:53 Suresh Sivavarman

We at ETAS have been working closely with Raj to modernize our portfolio.

00:01:57 Suresh Sivavarman

And Raj is here to bring us a little bit of insights into how the SDV platform can transfer from regular on-prem tools to cloud native engineering applications.

00:02:08 Suresh Sivavarman

So Raj, thanks for being here.

00:02:09 Suresh Sivavarman

We're glad to have you on the show.

00:02:11 Raj Paul

Thank you, Suresh.

00:02:11 Raj Paul

Appreciate the opportunity.

00:02:12 Raj Paul

Looking forward to this good discussion with you.

00:02:15 Raj Paul

I know we'll had a lot of conversations, a lot of them philosophical as well, around these topics.

00:02:21 Raj Paul

I'm really looking forward to this conversation.

00:02:23 Suresh Sivavarman

Perfect.

00:02:24 Suresh Sivavarman

So let's get right into it.

00:02:25 Suresh Sivavarman

So I think one of the first areas that I wanted to focus on was something that's of deep interest to me, which is data-driven decision-making.

00:02:35 Suresh Sivavarman

Historically, data-driven decision-making was always limited by the capability of the tools that were at hand.

00:02:42 Suresh Sivavarman

We always had data, but we didn't know how to properly utilize it, or how to effectively utilize it to make decisions that can impact how close we can get to a real-world model.

00:02:52 Suresh Sivavarman

And we hear that data is kind of like the new fuel, right?

00:02:56 Suresh Sivavarman

It's driving North American OEMs and activating them to unlock new capabilities.

00:03:02 Suresh Sivavarman

What do you see as kind of your, in your world, as data-driven decision-making?

00:03:07 Suresh Sivavarman

Where do you see OEMs leveraging it?

00:03:09 Raj Paul

So

00:03:11 Raj Paul

Data being the new oil, data being the new fuel, I mean, that's been talked about for quite some time in different connotation.

00:03:19 Raj Paul

It's lot more now for the last, I would say, the couple of years.

00:03:23 Raj Paul

What I would say is since the ChatGPT movement, when Gen.

00:03:27 Raj Paul

AI has become the latest thing where everybody's focusing on, you cannot have a good AI strategy if you don't have a good data strategy to back.

00:03:39 Raj Paul

So

00:03:40 Raj Paul

You touched on data-driven decision-making.

00:03:43 Raj Paul

So if you look at organizations today, in my opinion, there's no shortage of data.

00:03:48 Raj Paul

We'll have the capability to collect data telemetry from vehicles for quite some time now.

00:03:54 Raj Paul

And every OEM's maturity on the granularity of the data they can collect probably varies.

00:04:00 Raj Paul

But the capability to collect data has been there.

00:04:03 Raj Paul

I mean, I'll have the privilege to work in this industry from a connected vehicle standpoint, probably.

00:04:08 Raj Paul

since very early days when OnStar started this.

00:04:10 Raj Paul

But then over the years, initially, what was Holy Grail, connecting to vehicles and all that, those are all now becoming table stake.

00:04:20 Raj Paul

But on the flip side, how to put the data to use is still a challenge even today.

00:04:25 Raj Paul

So for me, when you talk about data-driven decision-making, one of the most talked about topic today is how can we bring down the overall development time.

00:04:34 Raj Paul

While the focus has been on how to bring the development time, I think every OEM is taking a step back now with tier ones like you to figure, hey, we have this data.

00:04:43 Raj Paul

Why can't the decisions we can drive from the data can actually play a bigger role to bring the overall development time?

00:04:49 Raj Paul

And it's not about cutting corners and then trying to essentially optimize the overall timeline.

00:04:54 Raj Paul

So we see everyone basically looking at bringing down that development timeline using the metrics and insights they can get out of the data they have.

00:05:02 Raj Paul

The data could be vehicle data,

00:05:04 Raj Paul

and trying to understand what is happening in the vehicle and reacting to a much better way.

00:05:08 Raj Paul

But it could also be telemetry you can collect out of your own CI/CD pipelines so that you can actually try to do better in redefining your pipeline if it makes sense, whether it's collecting quality metrics or development metrics and whatnot.

00:05:23 Raj Paul

I think there is all kinds of data which can play a role in the decision-making process.

00:05:28 Suresh Sivavarman

I think the decision-making process is very clear, but

00:05:32 Suresh Sivavarman

What are some gaps to actually make those decisions, right?

00:05:36 Suresh Sivavarman

I think what we've learned in the industry today is there's data.

00:05:39 Suresh Sivavarman

There's a vision of how we can make those decisions happen.

00:05:43 Suresh Sivavarman

But I see cloud as kind of the bridge to enable that, right?

00:05:48 Suresh Sivavarman

So can you explain a little bit about where you see the compute power of the cloud actually being able to enable this data-driven decision-making compared to where in the past we were limited by PCs?

00:06:00 Raj Paul

In the past, there were two scenarios which were pretty common.

00:06:03 Raj Paul

One is how a developer was more PC-centric, right?

00:06:08 Raj Paul

I mean, I have a tool, I have my development tools, I have my test tools, and I'm working on my PC and I'm testing.

00:06:14 Raj Paul

Then, of course, the whole concept of the whole data center started becoming prevalent, I mean, what we call on-prem today, where obviously there were collaboration patterns where it's beyond your PC, but now you're dealing with the back end of service to essentially help you collaborate.

00:06:30 Raj Paul

But the paradigm shift we have seen now for quite some time now is as and when these on-prem services or servers move to the cloud, that collaboration becomes a lot more easier for developers.

00:06:43 Raj Paul

Because now you're dealing with a common plane in the cloud for development and for collaboration.

00:06:51 Raj Paul

And I think that removes a lot of the challenges and bottlenecks

00:06:55 Raj Paul

between organizations, between developers, when they are actually working on a common code base.

00:07:01 Raj Paul

And this not just works within one region, it actually plays a much bigger role when teams are distributed across the globe, working on the same code base as well.

00:07:10 Raj Paul

So the cloud is now becoming the common ground to bring things together and helping the collaboration and then...

00:07:17 Raj Paul

helping with that decision-making process from the data, which can get used.

00:07:21 Suresh Sivavarman

So I think what you're saying is the old classical siloed mentality can be kind of closed off here a little bit, right?

00:07:27 Suresh Sivavarman

Because one of the things that I always see is that in engineering, you have a very siloed environment of how they handle their work.

00:07:34 Suresh Sivavarman

Then you went into manufacturing, then you go into service, then you go into the end customer, right?

00:07:40 Suresh Sivavarman

But in this world, where cloud becomes a bigger source of data storage and data decision-making,

00:07:47 Suresh Sivavarman

You could work in a world where data coming out of manufacturing and service can impact your decisions in engineering.

00:07:55 Suresh Sivavarman

And this kind of closes the entire loop of how you can have an end-to-end viewpoint of how you make decisions based on not just the data you have in front of you today, but data you may have years from now.

00:08:10 Suresh Sivavarman

Do you see this happening now in the real world that people are starting to see this as a proactive approach to using data that's further down the life cycle to make decisions on product information today?

00:08:23 Raj Paul

So a classic example to the scenario which you're talking about, Suresh, is quality data.

00:08:31 Raj Paul

OEMs have had quality databases for a long time, and the quality data gets fed from different sources.

00:08:38 Raj Paul

including what happens at the service bay.

00:08:42 Raj Paul

This quality data obviously is within the manufacturing organization, a manufacturing and quality organization, I would say.

00:08:49 Raj Paul

But then how you can tie things together from an overall end-to-end tooling.

00:08:55 Raj Paul

That's been a challenge because a quality database, there are certain organizations that have access to that quality data.

00:09:00 Raj Paul

Then if you look at the connected vehicle data is probably another part of the organization.

00:09:05 Raj Paul

Your product development aspect of things, but like for example, where you guys fit in with a lot of the tools you have, all these datas in a typical enterprise reside in silos.

00:09:17 Raj Paul

But breaking that silos and then bringing the data together so that you see a continuum of what happens to a vehicle, not just

00:09:25 Raj Paul

currently, but the past historical data, if that problem was seen, probably, let's say, four, five years back, and how can you relate to the learnings and what kind of recalls were done and what was done to the fix?

00:09:36 Raj Paul

That whole timeline of events pertaining to a part or a vehicle or a particular feature is a lot more possible now with some of the compute and the data storage capability which the cloud can bring.

00:09:48 Raj Paul

The problem of silos is not a problem unique to this industry.

00:09:52 Raj Paul

Data silos, I would think,

00:09:53 Raj Paul

A lot of the industries have a problem.

00:09:55 Raj Paul

This is more a classic enterprise data governance problem.

00:09:59 Raj Paul

But more relevant in our industry being an auto manufacturer, you have a lot of players in our world, right?

00:10:06 Raj Paul

Whether it's an OEM, a tier one, or a part supplier.

00:10:09 Raj Paul

So I think consciously building that pipeline so that you can learn from the past, not just look at the current data and build your data pipelines, I think will help in the decision-making process.

00:10:20 Suresh Sivavarman

So I think to add to that,

00:10:23 Suresh Sivavarman

A decision-making process and cloud, the key word here is efficiency, right?

00:10:28 Suresh Sivavarman

We want to get things done faster.

00:10:31 Suresh Sivavarman

But another area that the industry is really focused on is how we can blend not just the cloud compute power, but also AI into our workflow.

00:10:42 Suresh Sivavarman

How can we use AI and cloud to modernize engineering workflow, engineering efficiency?

00:10:48 Suresh Sivavarman

And then using the data that we've collected and having AI analyze that data, means that we can even get more insight quicker and make decisions faster.

00:10:58 Suresh Sivavarman

So I want to talk a little bit about, everybody says they're using AI, right?

00:11:04 Suresh Sivavarman

The industry says everyone needs to use AI, they are using AI.

00:11:08 Suresh Sivavarman

But as Microsoft, you guys are heavily involved into the integration of AI solutions into your day-to-day work, not just in automotive, but in your regular business practices.

00:11:18 Suresh Sivavarman

What do you see for the automotive industry to actually ensure that AI becomes a meaningful adoption, right?

00:11:26 Suresh Sivavarman

It's being used in the right context, not just for the sake of using AI, but there's some value coming out of it.

00:11:32 Suresh Sivavarman

Where do you see AI driving that for auto?

00:11:35 Raj Paul

So let me, it's a pretty hot topic, and I think this question will probably warrant me answering it from different vantage points.

00:11:43 Raj Paul

So let's start with the first vantage point.

00:11:46 Raj Paul

The first one is developer productivity.

00:11:48 Raj Paul

An engineer's developer productivity, right?

00:11:51 Raj Paul

So the tools which a developer or an engineer can use and then try to gain productivity, that's one of the lowest hanging fruits, I would say.

00:12:02 Raj Paul

When I say the lowest hanging fruits, whether AI is infused on the day-to-day development from an engineer standpoint, like for example, in our world, for example, GitHub Copilot, right?

00:12:14 Raj Paul

So we have coding assistance, as I would say, to help the development aspect of things.

00:12:19 Raj Paul

We are partnering with you to infuse AI into your tools, which is another set of tools that engineers are going to use.

00:12:25 Raj Paul

And if AI can essentially help streamline the process, get them to be more productive, great.

00:12:31 Raj Paul

And then now let's look at the testing aspect of things from an engineering standpoint.

00:12:35 Raj Paul

If I can do test case generation a lot better, if I can automate some of my tests, some of the tests which I am used to doing manually,

00:12:42 Raj Paul

or even if I basically have to do integration testing, and there's a lot of things which me as a developer can actually leverage AI today.

00:12:50 Raj Paul

We say Copilot, Microsoft Copilot is the UI for AI.

00:12:55 Raj Paul

It is not just for your general knowledge questions, but it's also your UI for all those agents which can actually power my day-to-day development.

00:13:04 Raj Paul

task and I can make a developer more productive.

00:13:07 Raj Paul

So that is 1 aspect of things I would say is an easy thing for a developer to basically look at and leverage AI.

00:13:12 Raj Paul

As I said, both using tools from us, tools from you, and in general, how AI can play.

00:13:18 Raj Paul

That is 1 aspect of things.

00:13:20 Raj Paul

Now let's get into a little more complex scenarios.

00:13:24 Raj Paul

We talked about in the previous conversation about the amount of data which sits in an enterprise.

00:13:28 Raj Paul

Specifically, let's take vehicle telemetry.

00:13:31 Raj Paul

There is historical information,

00:13:33 Raj Paul

There is runtime, real-time information being gathered by the vehicle.

00:13:37 Raj Paul

And we also talk in our industry about predictive analytics, proactive monitoring, and how can we essentially avoid an expensive recall, which means the capability, one, to store massive amounts of data, the capability to get insights from the data leveraging AI so that you can predict a failure even before the failure becomes a lot more common.

00:14:00 Raj Paul

And not just stop there, look for a fix or probably even kick off a development tool chain to basically find a fix and then try to essentially update the vehicle so that the recall is not expensive or probably doesn't even happen.

00:14:15 Raj Paul

So this capability is not easy, but I think with the power of AI, power of the cloud, I think those kind of tooling and

00:14:26 Raj Paul

tool chaining and developer tool chain is possible today with the way technology is moving.

00:14:32 Raj Paul

And I see that's a pretty powerful use of AI coming in as well.

00:14:37 Raj Paul

Outside of this, let me use my last example.

00:14:40 Raj Paul

We are classic V model, which you know is something which engineering and product development adopt quite a bit with most OEMs or all OEMs.

00:14:52 Raj Paul

The streamlined process, I'm struggling to find the word here.

00:14:56 Raj Paul

I'm not going to say waterfall exactly, but on the flip side, there is waterfall-ness in the V model.

00:15:02 Raj Paul

But now just imagine I can use agents to actually help me with functions within the V model so that one, I can optimize that process so that I can actually condense my development time.

00:15:16 Raj Paul

There is a lot of value in leveraging agents around the V model.

00:15:20 Raj Paul

What we are seeing out there is we see players like you come out with their own agents, which can actually help within the V-model, either for testing or even trying to help with requirements gathering, for example.

00:15:32 Raj Paul

I mean, we have tools for that in leveraging AI.

00:15:34 Raj Paul

So there are different aspects of AI which can play a role within that VT-model to bring the development time.

00:15:40 Raj Paul

One of the common things which every OEM is looking at is how to condense the development time so that they can compete better with new age OEMs and with China OEMs.

00:15:50 Raj Paul

Which means leveraging AI to bring the development time around the V model.

00:15:54 Raj Paul

I think there's a lot of benefits around that, in my opinion.

00:15:57 Raj Paul

I think I touch three vantage points that are more vantage points.

00:16:00 Raj Paul

More than happy to talk, but I think these are some of the main pivots I can think of.

00:16:03 Suresh Sivavarman

So I'm a huge AI early adopter.

00:16:06 Suresh Sivavarman

I use it for everything day-to-day.

00:16:08 Suresh Sivavarman

But one of the things that I also face is engineers don't necessarily embrace it.

00:16:13 Suresh Sivavarman

A lot of people are sometimes nervous about it or don't fully trust it.

00:16:19 Suresh Sivavarman

or there's a general resistance or this fear that it's going to take someone's job, right?

00:16:24 Suresh Sivavarman

Whereas Microsoft, do you guys see as a means to have more users embrace AI and have users believe that AI is a way to speed up their day-to-day work, but without taking the value away from what they provide, right?

00:16:41 Suresh Sivavarman

Where does AI deliver its value first?

00:16:44 Suresh Sivavarman

How do you sell this?

00:16:46 Suresh Sivavarman

to engineers so there could be a more mass adoptation of AI.

00:16:50 Suresh Sivavarman

Because as you said, AI is a big part of the final solution, but it goes hand in hand with cloud.

00:16:57 Suresh Sivavarman

Having the two together, you have the mass benefit of shifting left or moving towards a faster development cycle, or getting to market faster.

00:17:06 Suresh Sivavarman

So where's the resistance?

00:17:07 Suresh Sivavarman

Where do you see the resistance?

00:17:09 Raj Paul

I'll take a step back and then I'll get to your point.

00:17:12 Raj Paul

China OEM was not the focus until recently.

00:17:16 Raj Paul

It used to be new age OEMs.

00:17:18 Raj Paul

The general belief was when a traditional OEM is looking at a four to five years for a product to get launched from development inception to launch, but a new age OEM in general is believed to be around three years.

00:17:33 Raj Paul

But now the comparison with China OEMs are around two years.

00:17:37 Raj Paul

So when you're thinking about bringing five years to two years,

00:17:41 Raj Paul

I'm not going to debate about how valid the two years here is, but just for discussion, argument's sake, the expectation is to cut 50% development time.

00:17:51 Raj Paul

So if you put that in perspective, are we really trying to get rid of an engineer here?

00:17:55 Raj Paul

No.

00:17:56 Raj Paul

What we're trying to do here is literally if you take the English meaning of copilot, I am looking at co-pilots helping engineers to basically pilot whatever they're piloting much faster.

00:18:08 Raj Paul

So that is the way I would look at the world where it's a copilot trying to help you and get you more productive because the goal here is to cut down the overall time of development to half, if not more.

00:18:20 Raj Paul

That's #1.

00:18:21 Raj Paul

Number 2, as with any technology, there is always going to be an adoption curve where there is going to be skepticism, cynicism, paranoia, you can talk about all the English words.

00:18:35 Raj Paul

And to add to that, when AI started off, there's also, we also had the whole problem of hallucination, right?

00:18:41 Raj Paul

Are the answers which in OE, which AI is going to give us really, really 100% foolproof.

00:18:48 Raj Paul

So I'm gonna use a case theory which is very published very widely from by us and Toyota.

00:18:54 Raj Paul

Toyota, the way they use agents today, I mean, there is program which they launch called Obeya.

00:18:59 Raj Paul

Obeya, I think in Japanese means large conference rooms.

00:19:02 Raj Paul

They experimented with AI in their powertrain group, and now they're rolling it out, where they tried to solve multiple problems, but I'll just try to hone in on two problems which drew my attention.

00:19:15 Raj Paul

The first one is there's so much of institutional knowledge within a Tier 1

00:19:20 Raj Paul

community like yours or an OEM, which for all the documentation you can do, sometimes there's a fear of that institution knowledge walking away.

00:19:31 Raj Paul

But now imagine an agent is able to essentially preserve that institution knowledge.

00:19:38 Raj Paul

When a new developer comes, he has access to this agent to basically tap into that institution knowledge, which we talked about quality data.

00:19:45 Raj Paul

I mean, it could be anything in the past.

00:19:48 Raj Paul

That's super valuable in that regard.

00:19:50 Raj Paul

I would think anybody who has access to that knowledge with a very simple conversational interface, I'm sure they would love it.

00:19:57 Raj Paul

So that's one way of looking at that Obeya story.

00:20:01 Raj Paul

Another way of looking at that Obeya story is from a productivity standpoint today, when you basically bring engineers together to basically whiteboard and have a conversation, the amount of prep which happens or amount of smaller meetings which happens before you get into decision-making mode,

00:20:18 Raj Paul

is human intensive today.

00:20:21 Raj Paul

And if a fraction of that can be eliminated, where the developer can deal with an agent who's an expert of a particular function within that department, in this case, the powertrain department, then I think your productivity obviously increases because I'm not waiting for a human's availability to have a meeting because now I'm dealing with an agent to give me that knowledge so that I am well prepped.

00:20:41 Raj Paul

And here again, we're not eliminating the human.

00:20:44 Raj Paul

The amount of time it gets to a decision-making process, bringing the right set of humans, we all know within organizations we spend a lot of time.

00:20:53 Raj Paul

So if I can take a fraction of the time, I think that adds a lot of value.

00:20:56 Raj Paul

So going back to your question, why is there adoption hindrance?

00:21:01 Raj Paul

I think as with any technology, we have some challenges.

00:21:04 Raj Paul

But then I think as people start using and then they see value, I think some of the fear of being displaced, I think is going away.

00:21:13 Raj Paul

then I think the adoption starts increasing as well, which we are seeing by the way out there.

00:21:18 Suresh Sivavarman

So I think you're dead on, right?

00:21:19 Suresh Sivavarman

The fear of displacement, you're changing it to embracing the change, right?

00:21:24 Suresh Sivavarman

But how do you maybe embrace change is by showing measurable gains in engineering efficiency.

00:21:30 Suresh Sivavarman

And I think this is one way that as more people use it, they'll start to feel it and they'll start to see it frees up some capacity for them to do more value-added work.

00:21:39 Suresh Sivavarman

When we talk about value-added work,

00:21:42 Suresh Sivavarman

We say AI is a piece of it, cloud is a piece of it, but then we always hear this mindset shift-left validation, right?

00:21:50 Suresh Sivavarman

We talk a lot about the industry about SDV transformation and all this kind of stuff, and we'll get to that shortly.

00:21:54 Suresh Sivavarman

But we also talked earlier about the vModel, the traditional vModel.

00:21:58 Suresh Sivavarman

The vModel is you do all of your coding, your requirements on one side and your validation on the other side.

00:22:04 Suresh Sivavarman

But this was time-consuming and kind of slowed down your development cycle in comparison to China, right, and the way they work.

00:22:11 Suresh Sivavarman

So there's a lot of stuff going on right now about how do we shift left?

00:22:15 Suresh Sivavarman

How do we embrace the fact that we have AI, we have cloud, but at the same time, this enables us to work in a virtual environment with virtual ECUs.

00:22:24 Suresh Sivavarman

And this means that we can detect issues earlier without even having physical hardware available.

00:22:31 Suresh Sivavarman

Where does Microsoft see?

00:22:34 Suresh Sivavarman

this shift left mindset really driving?

00:22:37 Suresh Sivavarman

And where do you think that it's going to bring its biggest presence in the automotive industry over the next, let's say, three to five years?

00:22:44 Suresh Sivavarman

Because there's a lot of activities going on around virtualization right now.

00:22:49 Raj Paul

So there are two aspects to this, Suresh, simulation and virtualization.

00:22:55 Raj Paul

I know we'll talk about software-defined vehicles soon, but I'm going to touch a little bit of software-defined vehicle to help answer this question.

00:23:01 Raj Paul

The decoupling of software development from hardware development has been talked about for quite some time in our industry.

00:23:09 Raj Paul

But the reality is, how much progress have we made in that regard?

00:23:12 Raj Paul

Initially, when software-defined vehicle was talked about, the initial days, the comparison obviously was the mobile phone.

00:23:20 Raj Paul

And the comparison was about, okay, if the mobile industry can monetize services and in-app purchases, for example, as an example, why can't the car industry do the same?

00:23:31 Raj Paul

right?

00:23:31 Raj Paul

Comparison of it is a mobile phone on wheels, data center on wheels.

00:23:35 Raj Paul

I mean, you have a lot of interesting comparisons to the vehicle.

00:23:38 Raj Paul

End of the day, the vehicle is a vehicle.

00:23:40 Raj Paul

It has its own sets of challenges, which the mobile phone will never have, particularly when it comes to regulation, for example.

00:23:46 Raj Paul

So what do we do?

00:23:47 Raj Paul

A waterfall model, hardware then software, you know is not going to scale, which means it's got to be hardware and software development running in parallel, which means the dependency I come from, I come from days in my earlier days,

00:24:01 Raj Paul

where my first customer when we built software used to be engineering so that they can validate another supplier, hardware supplier's deliverable before they can do anything.

00:24:12 Raj Paul

My first customer was not taking to production this connected vehicle software, which we did, but more engineering.

00:24:19 Raj Paul

The reason is

00:24:20 Raj Paul

They need the piece of software we developed to test their hardware so that hardware can be tested against somebody else.

00:24:26 Raj Paul

All these interdependencies is the world we lived with.

00:24:29 Raj Paul

I mean, I'm talking about not 20 years back.

00:24:32 Raj Paul

Now, if you look at it, and of course, even in those days, we used to build emulators and simulators so that we can test without that hardware dependency from engineering so that engineering can in turn use our software.

00:24:42 Raj Paul

There's a nice nested dependency there.

00:24:46 Raj Paul

With software-defined vehicle, right?

00:24:47 Raj Paul

Now, of course, there is ECU consolidation happening inside the vehicle, right?

00:24:52 Raj Paul

I mean, I'm not going to get into zonal model domain.

00:24:54 Raj Paul

I mean, I'm not going into the vehicle architecture here.

00:24:56 Raj Paul

End of the day, we are seeing consolidation happening inside the vehicle.

00:25:01 Raj Paul

We are also seeing a lot of compute now getting into the vehicle to support some advanced workloads like ADAS, for example.

00:25:08 Raj Paul

And not just ADAS, even infotainment workloads, so the amount of compute getting in.

00:25:12 Raj Paul

So now,

00:25:13 Raj Paul

The immediate thinking is, hey, with so much of compute and so much of development happening inside the vehicle from a software standpoint, why can't I virtualize everything so that I don't have to wait for the hardware?

00:25:25 Raj Paul

That interdependency has to happen if we have to cut down this overall development time I'm talking about.

00:25:33 Raj Paul

The simulation needs a lot of what gets done is compute intensive, which means you can scale up and down at ease

00:25:42 Raj Paul

in the cloud better than anything else.

00:25:45 Raj Paul

So I don't think there is any argument when it comes to simulation.

00:25:48 Raj Paul

Now let's get into virtualization.

00:25:51 Raj Paul

So the goal for us is to, again, ours is a very partner-led approach here.

00:25:56 Raj Paul

I mean, you are a great example.

00:25:57 Raj Paul

We partner with you.

00:25:59 Raj Paul

So that if we can help virtualize that ECU in the cloud, then the developer who is working on that ECU doesn't need to work for the hardware.

00:26:10 Raj Paul

so that they can continue development while the hardware is getting big, so that testing can happen when the hardware is available, rather than waiting for the hardware to be available.

00:26:19 Raj Paul

Because now I'm actually building and working on a virtual environment.

00:26:22 Raj Paul

So our goal is to essentially create that environment for engineering product development, so that development can happen in parallel, both in terms of simulation and in terms of virtualization.

00:26:35 Raj Paul

And our approach is to provide the platform for partners like you

00:26:39 Raj Paul

to provide those virtual tools and those simulation tools so that the developer community at large gets a benefit.

00:26:46 Suresh Sivavarman

And I think there's also another aspect of this, right?

00:26:49 Suresh Sivavarman

As the industry moves towards virtualization, we're also seeing a reduction in the amount of money that OEMs and customers are spending on physical artifacts.

00:26:59 Suresh Sivavarman

So not just hardware modules, but physical vehicles, right?

00:27:03 Suresh Sivavarman

And the more you can shift your development to the left, if we use that acronym here, right?

00:27:09 Suresh Sivavarman

The more you're testing on a virtual environment, the less money you have to spend to build actual vehicles for validation.

00:27:16 Suresh Sivavarman

And this means that you're kind of transforming the business as well, right?

00:27:20 Suresh Sivavarman

You're going from a traditional CapEx, capital expenditure

00:27:24 Suresh Sivavarman

business model to an operating expenditure business model.

00:27:27 Suresh Sivavarman

So your investment upfront is also reduced, and you're actually getting quicker feedback at maybe a fraction of the cost you'd have to invest.

00:27:36 Suresh Sivavarman

So I think there's a lot of driving factors into this virtualization world.

00:27:41 Suresh Sivavarman

But one of the things I want to talk about here is virtualization is something that's come up over the last few years very aggressively.

00:27:49 Suresh Sivavarman

But as

00:27:50 Suresh Sivavarman

OEMs are making that transition to SDV, software-defined vehicle.

00:27:56 Suresh Sivavarman

People have, I think, customers have realized that it's not something that you're going to just do overnight.

00:28:00 Suresh Sivavarman

You can't just say, okay, I'm going to become a software-defined vehicle company.

00:28:03 Suresh Sivavarman

There's a transformation that has to happen.

00:28:06 Suresh Sivavarman

And that transformation has to happen from everything we just talked about, right?

00:28:10 Suresh Sivavarman

There is a transformation towards cloud computing.

00:28:13 Suresh Sivavarman

There's a transformation towards AI.

00:28:15 Suresh Sivavarman

There's a transformation towards virtualization.

00:28:17 Suresh Sivavarman

All of this coming together is going to develop your back-end system to enable the vehicles to go into a software-defined vehicle environment.

00:28:26 Suresh Sivavarman

So the back-end transformation, the question is,

00:28:30 Suresh Sivavarman

How much is it really talked about?

00:28:32 Suresh Sivavarman

You talk a lot of people, a lot of engineers and a lot of industry specs always talk about SDVs.

00:28:36 Suresh Sivavarman

It's the next big thing.

00:28:37 Suresh Sivavarman

This is what we're going to do.

00:28:39 Suresh Sivavarman

But what does a transformation to an SDV look like from a tooling perspective?

00:28:45 Raj Paul

So the best example I like to use in this regard is how OEMs are building.

00:28:51 Raj Paul

You'd have heard about the word data pipeline, and I like to call it as feedback loops.

00:28:56 Raj Paul

In an ideal world, an OEM

00:29:00 Raj Paul

tries to capture all kinds of telemetry from the vehicle, right, so that you can understand what is happening inside the vehicle.

00:29:09 Raj Paul

And in a lot of the advanced cases, you can also, you also want to know what the car is seeing.

00:29:14 Raj Paul

When I say the car is seeing, you also basically try to tap into some of the camera feeds which come, what the car is detecting, so that you can constantly learn from your fleet of cars.

00:29:23 Raj Paul

Not test fleet alone, I'm talking about production cars as well.

00:29:26 Raj Paul

But now, if I have my feedback loops

00:29:29 Raj Paul

built in such a way that I can detect issues much earlier on, because now why am I looking at an aggregate view of what my cars are seeing from a telemetry standpoint and everything else.

00:29:40 Raj Paul

And now if I can essentially now identify problem a lot earlier, and my whole tool chain is built in such a way that it could turn around and send a fix to the car even before the person who's driving the car realizes more like the amount of updates we get on our cell phone today.

00:29:58 Raj Paul

That is the world we are marching against.

00:30:00 Raj Paul

And again, every OEM's Uber goal is to be there.

00:30:03 Raj Paul

And the maturity of every OEM could be different as well.

00:30:07 Raj Paul

This is easy said than done.

00:30:08 Raj Paul

And you talked about PLM and a couple of other things integrating as well.

00:30:11 Raj Paul

Now, if I extend that all the way to my manufacturing, that is even more harder, but very doable.

00:30:17 Raj Paul

But that's the world we are trying to go.

00:30:19 Raj Paul

So when you talked about software defined vehicle,

00:30:22 Raj Paul

There are multiple aspects which will, depending on who you talk to and ask them what a software-defined vehicle is, some could say it's a completely feature-driven world.

00:30:30 Raj Paul

I can turn on and off features in the car.

00:30:32 Raj Paul

I can monetize features in the car.

00:30:34 Raj Paul

So that is one nice way of looking at software-defined vehicle.

00:30:38 Raj Paul

In order to do that, another way of looking at it is my car is completely upgradable, just like my phone today, so that my car's hardware value

00:30:48 Raj Paul

elongates because I'm giving it a new breath of fresh air every time with new piece of software coming in, which is the update capability which the car needs to have as well.

00:30:59 Raj Paul

But regardless of how you look at SDV, in both these cases, building those feedback loops, building those development tool chains, building the testing tool chains, all that becomes very pivotal in order to build a successful software-defined baker.

00:31:15 Raj Paul

Today,

00:31:16 Raj Paul

Meaning, I'm going to use an example here again.

00:31:19 Raj Paul

Teams, right?

00:31:19 Raj Paul

I mean, I'm sure you use Teams.

00:31:21 Raj Paul

I mean, right?

00:31:21 Raj Paul

I mean, we all use Teams quite a bit.

00:31:23 Raj Paul

If you look at the amount of updates we send, we're constantly learning from that software so that we can actually take care of issues and send updates proactively.

00:31:32 Raj Paul

And of course, one could argue that's a piece of software.

00:31:35 Raj Paul

We as an organization do quite a bit, even with our Xboxes today.

00:31:38 Raj Paul

So for me, now, of course, you have a car, which is a lot more complex thing on wheels, on the move.

00:31:45 Raj Paul

But the paradigm shift we're seeing pretty much with every OEM, again, as I said, the maturity of this could vary from OEM to OEM, is how can I build a feedback loop?

00:31:53 Raj Paul

How can I build those data loops so that I can constantly learn and recalibrate my development process with an uber goal of cutting down my overall development time, reducing my cost, reading my warranty cost, and all of that?

00:32:06 Suresh Sivavarman

I think you used the example of a cell phone, right?

00:32:09 Suresh Sivavarman

If something goes wrong, you can't update the next day, and most people don't even realize that they just hit.

00:32:14 Suresh Sivavarman

Go ahead and it's just a part of your day-to-day life.

00:32:17 Suresh Sivavarman

I think in the automotive world, the one thing we're going to have to get used to is the fact that over-the-air updates are going to be the future of how things are going to get fixed, right?

00:32:26 Suresh Sivavarman

Absolutely.

00:32:27 Suresh Sivavarman

You're connected to the cloud.

00:32:28 Suresh Sivavarman

The vehicle is connected.

00:32:30 Suresh Sivavarman

The OEM has learned about what's going on with your vehicle or fleet of vehicles.

00:32:35 Suresh Sivavarman

Before you know it, the fix is going to come across.

00:32:38 Suresh Sivavarman

You're just going to accept it like you have on your phone and you go about your merry day, right?

00:32:42 Suresh Sivavarman

So I think we're heading towards that.

00:32:44 Suresh Sivavarman

I think it's an ecosystem that has to slowly develop and build towards that.

00:32:49 Suresh Sivavarman

And I think that's where we are today.

00:32:50 Suresh Sivavarman

We're at kind of that inflection point where not only are we speeding towards, but we're also kind of slowing down to understand exactly what is necessary from a back-end transformation before we can move to the front end of the vehicle to the end user.

00:33:04 Suresh Sivavarman

So I think we're there.

00:33:05 Suresh Sivavarman

We're getting there.

00:33:05 Suresh Sivavarman

But

00:33:06 Suresh Sivavarman

It's not just about feature delivery.

00:33:08 Suresh Sivavarman

It's also about connecting the whole end-to-end lifecycle of a vehicle.

00:33:13 Suresh Sivavarman

And like you said, the closed-loop portion of it is extremely important.

00:33:18 Suresh Sivavarman

But as we talk about all this, Raj, what does 2026 to 2030 look like from a digitalization perspective?

00:33:25 Suresh Sivavarman

What do you think the industry is going to see?

00:33:28 Suresh Sivavarman

What do you think is going to accelerate and what do you think might plateau?

00:33:34 Raj Paul

On one end, I think as part of this industry, I mean, you and me, we all know the competitive threat is only increasing, right?

00:33:44 Raj Paul

I mean, we have one end of our world doing interesting things, and then we have the traditional OEM base, which are doing interesting things, but then the amount of time which takes

00:33:58 Raj Paul

is obviously a lot more than a China OEM.

00:34:01 Raj Paul

So that I think is a very hot topic, will continue being a hot topic, which means that I think is a huge forcing function for every traditional OEM and tier ones to relook at how they develop software today and how they develop hardware as well.

00:34:16 Raj Paul

So that I think we'll see a huge progress when we go into the next decade.

00:34:21 Raj Paul

So that goes without asking.

00:34:23 Raj Paul

Now, a couple of other forcing functions which are happening is

00:34:27 Raj Paul

AI is a board level topic today.

00:34:30 Raj Paul

There is acknowledgement that AI is going to increase productivity.

00:34:34 Raj Paul

Of course, there is, as we talked about, there may be some skepticism, there might be some adoption challenges and all that.

00:34:40 Raj Paul

But on the flip side, we all know how AI could help us in this first pursuit of trying to build things much faster and quicker and probably hopefully cheaper as well.

00:34:52 Raj Paul

So in order for that AI

00:34:56 Raj Paul

mission, vision to be realized, the next forcing function which will happen is to try to get an act on the data, data problems.

00:35:05 Raj Paul

Whether it's data silos, trying to understand the data which you already have, and the capability to probably get even more fine granular data.

00:35:14 Raj Paul

That I think will automatically take care of itself because of the AI thing which you're jumping on.

00:35:20 Raj Paul

Because each one of them are interdependent on the other.

00:35:23 Raj Paul

I think overall as an industry, I think we will evolve in a very nice way going into the end of this decade.

00:35:31 Raj Paul

So when the next decade starts, I think if you and me were going to do a podcast again in 2030, I think we can look back five years and then look at the progress we have.

00:35:40 Raj Paul

I'm a pretty big optimist in this regard.

00:35:42 Raj Paul

I think as an industry, I think we are heading into a time where a lot of the challenges we are talking about today, I'm pretty optimist that it will change quite a bit.

00:35:53 Suresh Sivavarman

So I think at the end of the day, we've got to modernize the backbone, we've got to experiment early with AI, and use data.

00:36:00 Suresh Sivavarman

Right?

00:36:01 Suresh Sivavarman

Simple.

00:36:01 Suresh Sivavarman

Three things.

00:36:02 Suresh Sivavarman

And we'll be heading into 2030 with a much more optimistic idea and modernized workflow.

00:36:10 Suresh Sivavarman

Overall, I think we talked a lot today about how cloud is a fundamental piece of how we're going to move towards this SDB transformation, right?

00:36:18 Suresh Sivavarman

We also talked about how embracing AI will help efficiency and help getting some of that data more appropriate towards what you would like to use it for.

00:36:28 Suresh Sivavarman

And we understood that

00:36:30 Suresh Sivavarman

It's all a part of a larger transformation, right?

00:36:32 Suresh Sivavarman

Shifting left is a clear direction that we want to head to because it means you get to learn faster, you iterate faster, you find out things faster, you fix faster.

00:36:41 Suresh Sivavarman

It's this whole quick information mindset.

00:36:45 Suresh Sivavarman

In order to do all this, we also understand that sometimes we have to slow down a little bit, look at the ecosystem, figure out what the ecosystem needs, and build up from there.

00:36:54 Suresh Sivavarman

And it's going to be a transition over the next few years as everybody migrates towards this world.

00:37:00 Suresh Sivavarman

Raj, I want to thank you for your time today.

00:37:01 Suresh Sivavarman

It was a great discussion.

00:37:02 Suresh Sivavarman

I think we learned a lot about what Microsoft is doing in the automotive industry and how you're helping support some of that activities.

00:37:09 Suresh Sivavarman

Also, some of the things that ETAS and Microsoft are doing together to move forward on the automotive industry.

00:37:15 Suresh Sivavarman

And look forward to having you as a guest in the future again to talk about maybe five years from now how this transformation has actually taken place.

00:37:23 Suresh Sivavarman

But thank you again for joining today.

00:37:25 Raj Paul

So Suresh, appreciate the time and the opportunity to have a nice conversation with you.

00:37:29 Raj Paul

I think you and me have been in this industry for a very long time.

00:37:32 Raj Paul

Looking forward, I think a lot of the work we're doing together between Microsoft and ETAS, I'm sure will help the industry at large and looking for a better collaboration with ETAS and how we can bring tools and productivity to this community.

00:37:46 Suresh Sivavarman

Thanks again, Raj.

00:37:48 Suresh Sivavarman

Looking forward to furthering our relationship with Microsoft and building on what we have already started, right?

00:37:53 Suresh Sivavarman

Looking forward into the next few years.

00:37:55 Suresh Sivavarman

But thanks again for joining.

00:37:57 Suresh Sivavarman

And thank you everyone for tuning in to today's episode of Empowering Tomorrow's Automotive Software Podcast.

00:38:03 Suresh Sivavarman

If you enjoyed today's episode, don't forget to subscribe on Spotify, Apple Music, or wherever you get your podcasts.

00:38:09 Suresh Sivavarman

Feel free to share the episode with your network and leave us a review.

00:38:11 Suresh Sivavarman

We'd love to hear back from you.

00:38:13 Suresh Sivavarman

This concludes our episode.

00:38:14 Suresh Sivavarman

Please check back again for a new one.

00:38:19 Voiceover

Thank you for joining this episode of the Empowering Tomorrow's Automotive Software podcast.

00:38:24 Voiceover

Please leave a comment or review with your feedback or what you'd like to hear in future episodes.

00:38:29 Voiceover

To learn more about Automotive Embedded Systems and ETAS's capabilities, visit our website at ETAS.

00:38:35 Voiceover

That's ETAS.com.