Episode Transcript
[00:00:03] Speaker A: Welcome to AINA Insights where prominent leaders and influencers shaping the industrial and industrial technology sector discuss topics that are critical for executives, boards and investors. INA Insights is brought to you by INA AI, a firm focused on working with industrial companies to make them unrivaled segment of ONE leaders. To learn more about INA AI, please visit our website at www.ina.AI.
[00:00:40] Speaker B: Today we are joined by Bhavna, SVP of New Ventures at Synopsys. Synopsys is the backbone of modern chip design. With 35 billion ANSYS acquisition now complete, Synopsys has expanded from electronic design automation into full silicon to system engineering combining physics based simulation, AI and deep domain expertise across industries.
Bhubna leads new ventures at Synopsys including the health and life sciences business. The mission is to improve patient and population level outcomes by applying AI powered simulation to the hardest problem in healthcare. Before this role she was at intel, previously at Synopsys and held various executive positions in Applied Materials at Renaissance and now back to Synopsys as a head of new venture.
Bhubna, welcome to have you on the podcast here.
[00:01:29] Speaker C: Thank you Vinny. I'm very excited to be here. Thank you.
[00:01:32] Speaker B: Perfect. Bhubna, you first joined Synopsys in 97 and you've spent your career across Intel, Applied Materials and Renaissance.
And for our listeners who don't know about Synopsys, can you tell us what does Synopsys do and how is today's synopsis different from the one you left?
[00:01:52] Speaker C: I joined synopsys back in 2002. Actually before that I was at Intel.
Let me start with Synopsys today. Synopsys Today is number 12 software company by revenue. It's number one EDA supplier in terms of interface. IP also number one and now with the acquisition of Ansys, it is a leader also in simulation and analysis tools.
Back in the day I would say that Synopsys was focused more the scale of Synopsys was much different, focused more on EDA and ip. Now we have other businesses and just the role that EDA in general and Synopsys in particular plays in enabling. All of the technology innovations that have happened in the past decade or so are fueled by by the EDA software, the IP business that Synopsys has.
So the scale is actually much, much different. Not just from a perspective of revenue, number of employees, but also in terms of the impact and the ability to fuel all of the technology advancements that have happened in the past decade.
[00:03:11] Speaker B: Thanks for sharing that information. So now in your new stint you are leading The New Ventures group.
Can you tell us a little bit about the charter, the problems that you're trying to solve and how does it fit within the broader Synopsys strategy?
[00:03:25] Speaker C: Our mission at Synopsys is to empower innovators to drive human advancements.
New Ventures is attempting to do exactly that. So if I look at the assets, the unique assets that Synopsys has, both in terms of scale of software, multiphysics simulations in AI and scalability, these are all things that essentially can be applied to silicon to systems and silicon to any systems actually. So while we've been focusing prior to the Ansys acquisition on semiconductors, this really applies, the stack really applies to any system.
And healthcare, health and Life Sciences is one of those verticals that the scale and the assets that Synopsys has can be very uniquely applied for human advancement.
[00:04:23] Speaker B: That's very interesting to hear. Right, because when you think about Synopsys, people think about chip design and here you are solving the hardest problems in the healthcare. Can you elaborate a little bit more on what you just talked about? What is this connection between this, the simulation capabilities that Synopsys has built and these hard problems in the healthcare that you are trying to solve?
[00:04:45] Speaker C: So if you peel away what we think of as Synopsys in the industry, as an EDA software provider, simulation and analysis and actually look at kind of the core technology, it's really about this ability to optimize across multiple dimensions. So across the dimensions, power, performance area and in a manner that you keep costs and time, time in check on the multiphysics simulation.
Again, this is applying the concepts of solving physics based problems to various systems. Then on top of it, the core assets include AI models, insights, orchestration agents, et cetera.
Now that's one aspect of the answer. The second is really looking at the trends that are happening in healthcare.
First of all, if you looked at healthcare about a few years, maybe a decade ago, healthcare was really perceived as this vertical that was very, very slow to adopt. Change mired in regulatory, long, onerous regulatory processes and almost a generation behind in terms of technology adoption.
Today the scenario is very different. According to a report by Menlo Ventures, a few months ago, they were saying that AI adoption, enterprise scale, AI adoption, healthcare was actually leading in that the pace of the adoption is actually two times faster than any of the other economies.
From a growth perspective, health care makes up about 18% of the GDP and wellness is another 5, 7%.
Combined health and life sciences with health care, wellness and all of these subcategories is greater and the cagr, the growth rate is greater than any of the other vertical sectors that we're used to talking about in technology, in, in manufacturing, in aerospace and defense, energy and utilities, any of those sectors. So it's a very fast growing market.
AI adoption is increasing.
And AI adoption today is primarily in health systems focused on improving efficiency, addressing labor shortages. About 22% of all health systems have adopted enterprise level AI in their solutions. But primarily this is for things like ambient clinical documentation, it is for patient engagement, it is for coding and billing automation, all targeted at efficiency and addressing, like I said, human labor shortage. But that's just a tip, because where we want to get to is the adoption of AI for human outcomes. So to improve patient outcomes, for clinical decision, support for better design of devices, better design of clinical trials, for better risk assessment and outcome prediction. That's where we want to get to. So the opportunity is there. And then the final layer is really about multiphysics simulation. So oftentimes people get confused between AI and simulations. These are two separate things. Also you do have AI powered simulations.
There is increased acceptance and adoption of simulation in health care.
FDA has announced its plans, for example, to both reduce and at some point maybe eliminate animal testing altogether, replaced by evidence from computational modeling and AI powered simulation.
Those trends all coming together and the fact that the need is really for assets that are very similar to what Synopsys has.
So healthcare makes sense.
[00:09:00] Speaker B: Perfect. There are a couple of topics that you talked about that I want to dig a little deeper into. One was you talked about the mission to improve the human outcomes. And I think that's a very core of at least what you're trying to do here. And the other piece that you talked about is how is regulators accepting this more and more. So we'll come to that. But let's talk about the human outcome piece, right? And especially through AI powered simulation.
And you talked about some of the places where this technology is being deployed today. Can you talk, give us like one more concrete example in a little bit more detail of what does it actually look like in practice and the places where this technology is being deployed today.
[00:09:41] Speaker C: AI powered simulations. Yeah. So one example I can give, I think where there is actually the most mature use cases are probably in cardiovascular applications.
So in cardiovascular applications, both from the design of the device. So basically your valves, your stents, et cetera. So from the device perspective, AI powered simulations are being used to design those devices better.
Then from a clinical perspective at the deployment stage, also in the clinics, they're being used modeling and AI powered Simulations are also being used for assessment. What is the best sizing of a particular device based on a patient's specific anatomy? What is the best positioning? What is the best approach, the actually imaging and the catheter through the vasculature.
There are simulations to do all of that, but not only that, after the placement to also assess risk for, for example, for restenosis or for the event to reoccur, what is the risk of that? So in all of those layers, simulations are being used.
[00:10:58] Speaker B: Got it. That's actually very helpful to, at least to understand. Right. Because when I think about it, chip design itself is very difficult. But simulating a biological system like a heart that you talked about or how a drug flows through the body, they're completely different problems and they have their own, own unique challenges and difficulties. In terms of how you actually model those, can you talk us through what has AI made possible that was just not possible before in terms of looking at some of these problems and solving them?
[00:11:35] Speaker C: Yeah. So first let me address the earlier point, your comparison about EDA and chip design and biology and healthcare care. Right.
So applying multiphysics simulations to these domains is completely different.
So especially in biology, we're really talking about, you know, you rarely have this concept of a completely closed equation system to represent whatever it is that you're modeling. So oftentimes what it is really is complex human physiology requiring multi physics simulations. You know, structures, fluid dynamics, mass transport that really spans the simulation and interactions from molecular cell level, tissue level, organ level, and then complete system level.
In terms of the difference, it's really multiphysics at a very large scale.
Nonlinear and applied to magnitude of degrees of difference in complexity, that's number one. Number two is the high degree of variability.
What I mean by that is every patient is different, anatomy is different, physiology is different, responses to drugs is different.
Really, there are no accurate models. Whereas if you look on the eda, the chip design, the silicon and software side in general, the model performance and accuracy continues to rise meteorically. That's not the case here because of the high degree of variability.
What we have is population distribution and we have patient approximation. We don't have patient specific digital twins. Even though the aspiration for everybody is to get there.
The next thing is really about data and the ethical. First of all, the data fidelity is not the same as it is in software. We have sparse data coming from images from health care records, and there's really incomplete longitudinal data in health care.
Ethical considerations are another thing. You can't go about freely recording and Getting data from everywhere. You can also not do free experimentation as you please on humans.
Ethical considerations also completely different, which slows down the application of AI.
The way I think about it really is in engineering you're solving an equation to predict an outcome, whereas in biology you're modeling a reality in the hope of giving guidance in the real world under uncertain situations and environmental conditions. So that's kind of the difference. Now how AI can help is number one, if you look at the confluence of things that have happened that actually enable the application of simulations, applications of AI into health care, it's really this threefold concept of speed up in simulations. And that comes from all the developments that have happened in compute, in algorithmic efficiency, in software, et cetera. The power of AI to really accelerate all of this.
And simultaneously on the acceptance side, recognition, broad recognition of the power, and also from the power of simulation, but also from an approval process starting to accept it and FDA and other regulatory bodies actually accepting simulation based evidence. All of those coming together help with this AI in particular, I would say simulations still are very compute intense.
So things like surrogate models or reduced order models can really help with speeding up the outcomes.
The challenge of course is to maintain the high fidelity.
[00:15:48] Speaker B: Yeah, that's very interesting. And I think the parallels that you draw right in terms of the comparison, in engineering it's about solving the problem and here it's like to get to a best possible outcome. I think that that sort of resonates very well. You talked about regulatory and now a couple of times. And at least it's good to know that in the regulatory approval process using simulation instead of animal testing, that's a huge positive step.
Where are we on this journey on the approval side of using more and more simulations? Is it the first phase of the journey where I think the regulators are still building trust with the tool or are we like towards the latter half of that journey?
[00:16:34] Speaker C: So I would say very early, very early in the journey. So at a high level, basically simulation is used as supportive evidence. It is not the decision grade evidence, meaning that FDA recognizes and can accept, especially in the 510 approval path, substantial evidence versus in clinical trials.
First of all, acknowledging that there's been tremendous progress in this area. For example, in 2015 FDA approved six AI ML devices. Six.
Today that number is close to 1500.
Majority of these device AI device approvals have been related to imaging and signal analysis. The applications have been more in radiology, followed by radiology is majority of, I think about 75% or so, followed by cardiology, neurology, hematology and other applications that are trailing.
But all of these are still supportive evidence. Really what we want to get to is decision.
Great acceptance, but we're on the right trajectory.
Perfect.
[00:18:01] Speaker B: Let's talk a little bit about the agentic AI.
You've compared Agent K adoption to levels of autonomous driving L1 to L5. And I think in the simulation world, in last November you talked about that we are somewhere in the L1 to L3.
Now agent engineer was launched and Synopsys is calling it as a L4 workflow. Do you think we are moving much faster than you expected?
And where does the human expertise still
[00:18:24] Speaker C: matter the most in healthcare? In silicon or in health care? I think in healthcare we're nowhere, nowhere close to it. Healthcare is a very different scenario.
The way I would describe health care. And that option AI is really that we have a very rich data set and not so rich insights.
We have fragmented tools, fragmented workflows, long regulatory process, very expensive clinical trial processes. So while AI can help in all of those areas, I would say we're not even maybe at L1 if I were to compare it with the journey of EDA and silicon and system design. But having said that, I do see a lot of parallels. So for example, if you go back to early days of EDA and chip design, every large semiconductor company had their own system design company, had their own local EDA CAD groups and their own EDA flows.
Today, EDA is that layer that sits across all of silicon development to system development, all the way from your request requirements to manufacturing to life cycle management. Right. EDA is sitting on top of that layer.
But that happened over a few decades and a lot of innovation.
So that's sort of a evolution I see coming to healthcare. And I think the pace of adoption and the pace of innovation is going to be much faster today just because we have the advantage. We have the advantage of superior compute, we have the advantage today of superior algorithmic excellence of software scale and AI. And so all of these are going to fuel that innovation and the pace in healthcare.
[00:20:30] Speaker B: Oh, that's very good. Heartening to hear.
There's one more thing that I want to talk about at this year's Converge 2026. Your CEO said this is the year one of new synopsis, which I think sort of resonates in many ways. Right. With the ANSYS acquisition, there's a lot of capabilities that have been added to Synopsys from your seat. What does this combination unlock, especially when it comes to healthcare? In terms of solving problems, that was not possible before.
[00:21:03] Speaker C: I think a combination of everything that I kind of talked about, first of all, from why now? If you look at that question, why now? What is different today?
What are we able to do today, to do everything digitally that we're doing physically in healthcare? Well, it is this confluence, like I talked about, software advances, AI advances, compute advances that enable us to think even about the possibility of solving for the multiphysics, coupled domains that apply to healthcare to represent complex physiological events that happen in the human system.
So first of all, that is different and the scale that Synopsys is at today, and having solved this, built the expertise and built that muscle of being able to scale the software, the vertical integration, horizontal connection, putting all of that together, all of the domain expertise, I think is very transferable and applicable to health care.
And in a way I feel at an advantage because I have seen that evolution in ed, I have seen how that has happened, and I feel that it is the same playbook, the same blueprint that can be applied to healthcare.
[00:22:22] Speaker B: Well, that's good to hear.
I want to pivot and talk a little bit about you. You've had a very fascinating journey as well. And building new things inside a big company is not new to you.
You've done it at many of your previous stints, like at first time at Synopsys, you build the IP systems for automotive. Then at Applied Materials, you build material to systems platform. And then at you build the software and digitization. Org. So what have you learned about building something new inside a company with an established identity? And how are you leveraging that in the new venture group today?
[00:23:01] Speaker C: For me, I think the commonality amongst all of that and in building new things is really finding the ability to connect dots, to find analogies and be able to apply it to the new problems that you see. So really, first of all, seeing the opportunity, identifying the opportunity, following something that you're passionate about because you got to love what you're doing, and then finding those connections because you have those learnings, all of us have lived through that. And how do you take that learning and apply to it?
What are some concrete examples of that from an EDA perspective? Again, if you look at the EDA journey, companies like Synopsys, we started with parts of the workflow.
You had RTL synthesis, you had timing engines, you had placement, routing, et cetera, et cetera. Today you look at it, it's a beautifully vertically integrated workflow system. Now, not only that, I mean, that was not enough. Then you had to go and have this horizontal connectivity to maintain that integrity. So for example, what do I mean? If you look at verification, you have to have functional verification.
Then you also need verification for safety, power aware verification. All of these need to connect and tie together while you're doing this vertical integration.
I see a similarity in healthcare where the systems today are completely disjointed. Data is fragmented, analysis is fragmented, workflows are disjointed.
There is no standardization. There is some data interchange formats, but there is no standardization. None of the things that we've seen in the chip design and silicon design world.
For example, if I think about system specification, the functional specification, this is the biological intent. If I think about the sign off process, this is equal into the FDA approval. If you think about post silicon validation, this is the real world clinical feedback.
Really there's a lot of similarities and for me it's about connecting, finding those problems, connecting the dots and the ability to apply what you've learned to what you're seeing in front of you. And that's what I've done at my past new ventures in other companies.
[00:25:37] Speaker B: So you actually had a very unusual career path.
You have done chip design at Intel EDA software in your first int at synopsis, you've done semiconductor equipment at Applied Materials, you've done software at Renaissance. There are very few players who have seen the industry through several of these layers. What's the through line connecting all of this?
[00:26:01] Speaker C: Curiosity, curiosity and not being satisfied with accepting that I don't know about this, the curiosity to know how does what I do connect with other parts of the ecosystem that are upstream, downstream and what can I do in the space that I'm working on to have positive outcomes upstream and downstream.
And at the bottom of this is really about human advancement. That resonates for me. And with healthcare I just feel like the time is perfect. We're at the beautiful inflection point where we can really marry all the advancements that we in technology, apply it to healthcare for the advancement of humanity.
[00:26:54] Speaker B: The other fascinating thing about you also is that you're coming from a very deep technical background. You have many technical publications, multiple patents, very deeply rooted in, at least on the technical side.
How was the leap from the deep research to the business side for you?
[00:27:15] Speaker C: Felt very natural. Because at the end of the day, whatever technology we built, the satisfaction, the fulfillment comes when it is deployed and when it is used.
And again, when I think about health care, there are many angles of the problems to be solved. There is advancing clinical research is Advancement of in silico trials to replace or to complement animal testing, early human testing. But then there's also the economics part of it, because at the end of the day, the economic incentives have to align to adoption of the technology that you're creating. To me, it feels very natural to think about the business aspect of it as well.
[00:28:07] Speaker B: Helpful.
One quick question, or at least to sum up a lot of the discussion that we have had.
You talked about AI, you talked about multiphysics simulation, you talked about domain specific data, especially in the healthcare, which is difficult to get. And you also talked about these are converging in ways that was just not possible before, even like two, three years back.
What does it look like for the next chapter and what has to go right for all of this to be successful?
[00:28:37] Speaker C: We're on the right trajectory, I would say so, for example, all the regulatory bodies, and using supportive evidence from simulations, from AI powered simulations, the trajectory is the right one.
But I think there's a lot of work to be done, really, if we want broader acceptance of this, then simulations have to be embedded much deeper into workflows. When I look, for example, again, drawing the parallel to what we know, chip design and silicon design, the adoption happened because the ecosystem kind of came together, connecting these fragmented data pipelines and workflows, orchestration of those workflows, et cetera. In healthcare, I think a similar journey is needed today. Everybody is off doing their own innovation, their own experimentation.
The ecosystem needs to come together. What do I mean by that? So if you look at eda, for example, what really fueled all of this is ultimately this recognition that there has to be broad collaboration across the ecosystem. So data interchange formats, whether that's RTL or how you pass your netlist from one part of the workflow to another, your verification workflows, what is acceptable to really stitch together, enable that workflow integration, that's one part of it. But also if you look at the concept of reusable silicon ip, there was broad recognition that there is no need for everybody to go repeat this because these are based on standardized specifications for the interfaces, your PCI Express or USB or Ethernet or whatever.
I believe that kind of a thing needs to come into healthcare, where you have models, even with the in silico trials and all the progress that's happening today, those models are very difficult to reproduce with accuracy. And it's really difficult to represent the whole body, complex physiology and biology and the interplay of cells, tissue, organs, et cetera. But if you can pivot for a minute to this concept of reusable models. So if the FDA has approved certain models, and if those models can then be available to the broader ecosystem and you can adopt that, then all of a sudden you've really accelerated the pace of innovation. So I believe that similar thinking and collaboration has to come in the healthcare ecosystem as well. But at the end of the day, the incentives and the economic incentives have to align as well in order to encourage this behavior.
So there are a lot of positive things happening. Like I said, the FDA approval process, but also this concept of per case reimbursement gaining traction. So for every situation where computational modeling, AI powered simulations is applied successfully with evidence, there is a possibility of a reimbursement per case which kind of encourages that behavior. So I would say we're on the right trajectory, but early in the game, early in the journey,
[00:32:10] Speaker B: now in the second stint, right, like you're leading this, the new ventures group, especially for healthcare, and your passion for the human outcome, right. Comes out in the conversation that you have had so far. What does it mean for you personally?
[00:32:24] Speaker C: So I've always wanted to be in health and life sciences. In fact, from career and profession choice, I wanted to be a doctor, but here I am an engineer.
Love my career, love engineering, but my passion is in health and life sciences.
From again, the conclusions of where we are. I just feel super fortunate where we're living at this intersection of all the advances that are happening and the pervasive intelligence everywhere. AI and technology seeping into our lives from every aspect.
For example, self driving cars just a decade, 15 years ago was a concept and today there's a common sighting on the streets.
But even though we have had all of this advancement in technology, unfortunately health care is only available to a select sect of the population.
It's not universally accessible because of many factors, the cost, shortage of labor, et cetera, et cetera. I really believe that good health care is a human right. Everybody needs to have access to that technology.
I truly believe where we are with the impact of AI, multiphysics, simulation, automation, software excellence, we can bring all of those together to make healthcare more universally accessible.
[00:34:03] Speaker B: That's very commendable.
Last question.
If a new engineer comes to you and says, hey, I want a career like you with the technical lab, with the business acumen, with the impact across industries and maybe industries in this particular case, what would be the one advice you'll get them?
[00:34:19] Speaker C: Follow your heart.
Follow your heart. Don't be afraid to experiment. If your heart tells you you can make an impact here, or you want to learn something.
Follow your heart, but also be aware that not all your ventures are going to be successful. You're going to have a few failures before you figure out what it takes to succeed. And so to be able to reframe failure as a learning experience towards success and not being caught up in failure as failure, I think both of those are equally important. The curiosity, the passion, following your passion, but also having resilience to failures and chalking those to learning experiences.
[00:35:07] Speaker B: Perfect. Gurma, it was a pleasure talking to you today and thanks a lot for your time.
[00:35:10] Speaker C: Thank you. Thank you for having me.
[00:35:17] Speaker A: Thanks for listening to INA Insights. Please visit INA AI for more podcasts, publications and events on developments shaping the industrial and industrial technology sector.
Sam.
