AI and Complexity Drive New Era in EDA
Industrial AI and growing compute demand are contributing to accelerating semiconductor complexity, straining traditional design methods. As transistor counts race toward one trillion per chip, the tools and workflows that engineers rely on must evolve to keep pace.
In a new episode of the Industry Forward Podcast, host Dale Tutt, Vice President of Industry Strategy at Siemens Digital Industries Software, sits down with Ankur Gupta, who oversees Siemens’ integrated circuit design portfolio, to explore how the semiconductor industry is responding to an era of exponential complexity and how AI will help companies navigate growing complexity.
00:12 Dale Tutt
Hello, and welcome to the Industry Forward podcast. I’m Dale Tutt, Global Vice President Industry Strategy at Siemens Digital Industry Software, and I’m your host for today’s episode. And today I am excited to welcome Ankur Gupta, the Executive Vice President of EDA Integrated Circuit Software at Siemens EDA, to our show. Semiconductors are critical to the products that we use every day, from airplanes, automobiles, and smartphones to cutting edge medical devices. They enable the software to find features that we value and accelerate the innovation in the products that we use. So this is a very timely discussion in our quickly changing world. Welcome, Ankur.
00:49 Ankur Gupta
Thank you, Dale. Glad to be here.
00:51 Dale Tutt
All right. Well, I’m really looking forward to today’s episode. Semiconductors are such an important part of all the products that we use everywhere today. Medical devices, our phones, our cars, our airplanes. And Ankur oversees Siemens integrated circuit design portfolio. He’s bringing deep expertise in chip design, verification, and AI-driven EDA solutions. He’s been at the forefront of helping some of the world’s top semiconductor teams navigate the new era of exponentially complex chip development. But before we get into the technology, let’s start on a personal note. So Ankur, what’s your favorite science fiction book or movie and why does it resonate with you?
01:33 Ankur Gupta
Lucy, my all-time favorite.
01:35 Dale Tutt
Lucy.
01:36 Ankur Gupta
Lucy.
01:37 Dale Tutt
All right. For our listeners that aren’t familiar with that, maybe share a little bit about that.
01:43 Ankur Gupta
Yeah, so this is, when was it made? Maybe about 2014, 2015, something like that. Luc Besson, the French director. And Lucy is this character who, however, let’s skip how it happened, but she gains super intelligence by gaining access to a greater portion of her brain. And then she gets powers like telepathy and psychokinetic powers. And that’s just amazing. Morgan Freeman’s in the movie. And I’ve always been fascinated with the brain. So we’re not talking about artificial intelligence, we’re talking about real intelligence. We’re talking about real intelligence and just, it just captures my imagination.
02:30 Dale Tutt
Oh, that’s awesome. That’s fascinating. So actually I’m a Star Wars guy, but I’m an aerospace guy and it really just woke me up in terms of getting me interested in aerospace and space especially. So
So you’re fascinated with the brain and you’ve had a fascinating career. Can you share a little bit about your journey and what’s inspired you to focus on electronic design automation and artificial intelligence in the semiconductor field?
02:59 Ankur Gupta
Yeah, sure. You know, for my graduate program, I wanted to do a PhD in mathematics and control systems because I was fascinated with the brain and space and Mars rovers and stuff like that. And then it hit me first year into the program that I need something more practical. So I pivoted to, at that time it was called VLSI. So I’m talking at the turn of 99, 2000, around that time. And that got me into EDA. So I ended up getting a master’s in EECS. And then EDA was fascinating back then in the early 2000s because you could work with all these chip companies and you’re not really building the chips, but you’re building the software that gets used to design these chips. And you get to see the result of that software.
So I spent about 15, 16 years of my career, early career, in one of these EDA companies. And through that journey, working with these customers, I realized that the application of the software that we’re building to solve real problems is fascinating. So I’ve just kept going deeper and deeper in that direction. And now you fast forward to now, what is AI doing at the end of the day? Many things, yes, but one key thing is the users of our software. They’re becoming more productive because of AI. And in a nice way. Because when you’re sitting in front of a computer and you’re doing your design, you don’t want to spend all your time setting things up or debugging a failure in the software or looking up documentation. AI gives you this ability to naturally, through your English language or natural language, interface with the tool. That’s awesome. And I wish I had it 20 years ago. So that’s really what we are attempting to do with introducing AI into our products.
05:16 Dale Tutt
Yeah. I really like that. And I certainly think as we do that, and when we introduce AI and it’s embedded in the products that the customers are already currently using, it’s going to make it easier for them to adopt it. And they’re going to get the benefits without having to go learn new tools. And so I think it’s a very, I think it’s a great approach. And certainly as a user, that’s what I want to see. I don’t want to have to go learn new stuff. At least on how to use the tools. I want to learn new things. I’m always curious about learning, but I want to be able to not focus on which buttons I have to go push to make it happen. So it is an exciting yet very challenging time in electronics and semiconductors. Chips are getting more powerful and complex, which means designing them is becoming more complex than ever as well. And just think about the capability in the chips and the scale at which they’re now getting down to in terms of the size of the nodes. And so let’s unpack that and explore how AI is going to be changing changing the game in EDA. And so thinking about in simple terms, when we use the term EDA, electronic design automation, not everybody in our audience may be familiar with that. I do get that question a lot when I say EDA, and I just think about it off, it just happens, but a lot of times, EDA, what does that mean? So how would you explain EDA to your family or to your friends when you’re describing what you do?
06:50 Ankur Gupta
There are a couple of analogies that you can draw. One is Microsoft Office. You use a Word document or an Excel program or a PowerPoint program to create something. And the software is that program. So EDA is that software to create that chip, to design that chip. Another analogy may be closer to everybody’s experience is imagine you’re building a skyscraper. So you’ve got design tools. You’ve got to floor plan the building. You’ve got to think about the plumbing. You’ve got to think about the structural strength of the entire building. That’s very similar to the skyscraper chips that we are building these days. And all of the design that goes into figuring out how to floor plan the chip, how to connect layers, how to wire, think about the electrical connection, how to power every single floor, every single room. It’s not much different from, in an abstract sense, right? It’s not much different from getting power into every single transistor and connecting this in a 3D sense. And then verification. It’s not just about design, it’s verification. You don’t just build a building and not test it, right? You test it. So there is testing to manufacturing spec, there is verification of the logic. Certain rooms, you want the power to turn on in a certain sequence, or that’s verification. So all of these things are done by the EDA software. It’s just for things that are the nanometer scale, whereas a building is in meter scale. So now you’re talking 9 orders of magnitude smaller. You can’t see it. Can’t see it?
08:51 Dale Tutt
Yeah. No, that’s awesome. Well, and you know, I love that analogy of the skyscraper. We’re building a skyscraper here and just, and then also I think it’s very helpful the scale of what we’re designing because we have electronics in everything. I mean, we interact with it. It’s in our cars, it’s in our airplanes, but even think about our computers, our laptops, or our tablets, or our smartphones today. I think what a lot of people take for granted is what’s actually going on inside that box. There are certainly people like myself and an engineer that always, we want the most powerful chip, and that number means something to us. But most people, let’s just use the example of a smartphone. They see the smartphone and generation after generation after generation, it’s basically the same size. But that chip is probably orders of magnitude more powerful after four or five, after four or five generations. And it’s changed. And so, let’s talk about what’s changing. We hear about the complexity and we hear about 200 billion transistors on a chip, which is — so that’s like the size of this room, right? No, it’s not. And even having a path towards 1 billion transistors by the end of the decade, which just boggles my mind. I’m sorry, trillion. I’m sorry. Yeah, 1 billion. Yeah, it doesn’t seem like so much. But you know, we have 3D skyscrapers of silicon, as you referred to it. So why are the traditional design methods no longer enough in the face of this explosion of scale, cost and time to market pressures. I mean, there’s so many things going on. So what’s happening with the tools?
10:37 Ankur Gupta
Yeah, so think about maybe two things. One is 200 billion transistors looking at 1 trillion transistors by end of the decade. What’s the form factor? Where do we have to pack these transistors? Now this is where you’re talking about very different form factors when you’re considering a cell phone, a mobile phone, that’s a very fixed form factor, just like you mentioned. The phone itself, the size of the phone itself is not changing from generation to generation, right? These chips tend to be 110, 120 mm square. And imagine packing now double the logic in that — you’ve got to go stack things up because it’s the same physical footprint. On the other hand, when you’re thinking of data centers, there are companies out there that are building dinner plate size chips, like Cerebras recently went IPO. That’s a big, humongous chip. It’s still a chip, but it’s a dinner plate size chip.
And that’s where you can do fancier things with packaging. You can do 2 1/2 D, you can do 3D packaging and all that. So that’s one just image to keep in mind. But then what’s happening inside these chips and why traditional EDA methods are no longer sufficient is that earlier you were operating at higher voltages, you could silo, you could margin the heck out of the design process. You can’t do that anymore. Things are interconnected at a different level. The mobile phones these days, the two nanometer phones, they’re operating somewhere between 500 millivolts to about 750 millivolts. So I mean, how can you margin and how can you keep enough margin for the chip and then you’ve got the package and you’ve got the board and all that, right? It’s just impossible. And they’re so interconnected. The thermal behavior inside the IC, inside the chip is impacting the cooling that you design at the blade level. So think about a data center rack. The chip is the heat generator in this case. So that’s interconnection. The power delivery mechanism that you decide at the package level, it translates into signal integrity issues over the PCB. And even outside in, the cooling of the entire data center and the rack, that ambient temperature, it impacts the timing closure of the chip because the parasitics are dependent on temperature.
So it’s just this massively interconnected world. And you think about traditional EDA. Traditional EDA is used to working in a siloed environment. I’m just going to design the chip. That’s an IC software. I’m just going to design the PCB. That’s a PCB EDA software. It’s no longer possible to draw those lines. Now you’ve got to look at the chip in the context of the system. So you’ve got to evolve. You’ve got to evolve your traditional EDA into something richer. Can we talk about that?
13:55 Dale Tutt
Yeah, well, definitely. It’s such a good point. Nothing exists in a vacuum. And I’ve had this conversation I guess over many years — I didn’t care if I was designing a landing gear system on an airplane. Valves are interesting, but it’s only when they’re connected to everything else that they actually do something. And so nothing works, you know, nothing lives in a vacuum. And so, you know, the examples that you’re given, like the data centers and, you know, how much, well, there’s so much power being consumed, but even how the external environment is influencing the performance of the chip. So as we think about all this complexity, and this is really where the AI is coming in, so how is AI changing the EDA tools and solutions to help tackle some of these problems that the traditional methods aren’t covering?
15:01 Dale Tutt
Thank you, Ankur. This has been a fantastic discussion. I appreciate your time today. I am sure that our audience enjoyed the episode as much as I have enjoyed talking with you about this topic. And so thank you again to our audience for joining us on Industry Forward Podcast. We will have more from my discussion with Ankur on the feed soon, so please make sure to subscribe to the Industry Forward Podcast on your favorite podcast channel. Take care and we hope to talk to you again next time. Thank you.
Siemens Digital Industries Software helps organizations of all sizes digitally transform using software, hardware and services from the Siemens Xcelerator business platform. Siemens’ software and the comprehensive digital twin enable companies to optimize their design, engineering and manufacturing processes to turn today’s ideas into the sustainable products of the future. From chips to entire systems, from product to process, across all industries. Siemens Digital Industries Software – Accelerating transformation.
