Vinci raises $250M to build a physics foundation model for semiconductor and hardware engineering

Speaker 3: Always great

Speaker 2: to see you.

Speaker 1: Congrats to the team on the launch.

Speaker 2: A rest of your day. We'll talk to you soon. Goodbye. We have the founder of Vinci and the CEO, founder and CEO, Hardik Kabaria is in the waiting room. Let's bring him in to the TV. How doing? Are

Speaker 6: Hey. Hey. How are you guys doing? I'm doing

Speaker 2: well. Welcome.

Speaker 1: We're doing great.

Speaker 2: To the show.

Speaker 1: We've had a wide ranging show today.

Speaker 2: Yeah. We've got

Speaker 1: lot Nice of conversation with Dylan, of the National Christmas Tree Foundation as well. Yep. It's been all over the place.

Speaker 2: Did you make it through that interview or not? You actually broke.

Speaker 1: I I I... It broke. It took it took me to deep walk.

Speaker 2: One of us is a professional Yeah. And the other one is an amateur.

Speaker 1: You can go watch... As it gets closer to Christmas, go watch our interview with the... With Rick, the head of the National Christmas Tree Foundation. But today, we're here to talk about your business. But why don't you introduce yourself and then we'll get into the company. Sure.

Speaker 6: Yeah. My name is Hardik. I'm co founder and CEO of Winchie. My background's technical, physics, and geometry software. I've been in that space for some time and started Winchy three years ago.

Speaker 2: Cool. Yeah. PhD at Stanford and eight years at Carbon. What were you doing at Carbon? What what what... I I actually... I'm not... There's a few businesses I wanted to know more about Carbon specifically and how it set you up

Speaker 7: Yeah. On

Speaker 6: Carbon's a three d printing company. Okay. I was a Sequoia, a silver leg back company. I... After my PhD, was looking for something that can use geometry, but would also place closer to the real parts.

Speaker 2: Sure.

Speaker 6: But joined as a software engineer. I was there for eight years. Yeah. I was VP of software for the last couple of years. So ship software on manufacturing type, manufacturing floors. Makes sense. In in Asia, in Eastern Europe, in US, Mexico. So it gave me a lot of perspective on how software can help hardware.

Speaker 2: Yeah. So narrow in on the current mission. Semiconductor design, semiconductor engineering, I noticed you're expanding into, like, vehicles, aircraft, satellites, memory. Those are similar problems, but Yes. It feels like there's enough money to be made, like, narrowly in in semiconductor. So I'm interested in the overall strategy and Yes. And the level of abstraction you're operating at.

Speaker 6: So I'll start with something very basic, which is physics. Yeah. Everything that is around us, whether there is human made parts, nature made parts, it'll be governed by laws of physics. Laws of physics are universal. Something as simple as energy in will always be energy out Yeah. Momentum balance, and so on. Any part we design, we care about physics because ultimately, the parts will start behaving in the real world. And if they don't perform the way you expect it to perform, all goes to hell. Right? What are we building is a foundation model for physics.

Speaker 2: Okay.

Speaker 6: And how it's meaningful... Let's say you create thousands of design ideas. If you cannot grade them with physics, it'll be very hard to pick which one's right.

Speaker 2: Mhmm.

Speaker 6: Right? So physics is the compilation layer for hardware.

Speaker 2: Sure.

Speaker 6: It's been manual. With our product, we are making it automatic and scalable. Yeah. Now this applies to semiconductor, high bandwidth memory. It also applies to satellite. Okay. But when the scope is as big, you gotta start extremely focused somewhere. Yeah. So we said, why don't we pick thermal problems in semiconductor and electronics? And we build a software that allows our users to do it and ship it. Yeah. And, you know, make it true, and then we expand. So that's where we are.

Speaker 2: What is the sales process for a semiconductor company these days? Because I feel like they have something that's working so well. I thought the only job they had was make more of the thing that everyone wants and raise prices, but they have time to implement new software solutions for third party companies.

Speaker 6: A good question. Right. So you're asking... So first, it's top down sales. Yeah. You cannot go to these large, large companies and say, I changed one engineer's mind. I mean, I would say good freaking luck. Okay? If you have to convince a VP Yeah. Saying that, hey. This is something you should try, and it's working for us. Now So to the second question, do they have time to try out, who are we? I'm not even from semiconductor. Yeah. Why would they even give me a time of the day? I've never worked in semiconductor industry before. The problem is so acute Yeah. That if you can really help them, it matters. So we start our sales process with asking what matters to you, and answer is consistently the same thing, time to market and performance.

Speaker 3: Mhmm.

Speaker 6: I do not want to miss my time to market. I have to launch a new chip, new product at CES, end of the

Speaker 1: year Mhmm.

Speaker 6: And it better be performing. And we work with the most complex parts people are making.

Speaker 2: Sure.

Speaker 6: High bandwidth memory. Yeah. TPU, GPU. Right? And their thermal problems are so acute that the existing tooling doesn't take them there. So if you can really be more accurate, higher fidelity automatically, they are definitely interested in trying it.

Speaker 1: Okay. So maybe a month ago at this point, we had the Navier Stokes moment and it was very cool for for people within the tech There was a lot of excitement. But average people, like if you tell them we use a computer to solve a math problem, you know, they're not gonna be super... It didn't like... It didn't feel like it moved the market is another way to put it. Yeah. My question for you is like how how... Are there a bunch of other moments like that happening at like a micro level within businesses where they're using models and they're having breakthroughs, but they're not raising their hand and saying... Like, basically just publishing that because it actually is, can become IP?

Speaker 6: Yes. So I would say that it's... We have been deployed commercially for some time. Mhmm. So I'm gonna ask you a question from... I'm gonna start answering a question from there. Yeah. Would... In in this... Our companies, the IP sensitivity, who they work with, they are extremely, private and secure about it.

Speaker 3: Mhmm.

Speaker 6: So they get start using a new vendor, new tool, new method, and they don't keep it to themselves. Mhmm. And that's the style of the industry, which, it helps us. It hurts us, but I would say we are part of the ecosystem. So I like to go and meet customers where they are. So I would go far as to say, I even... It's embedded in our deployment. We do on prem deployment, and we make our product work with zero forward deployed engineers. Extreme data privacy. We'll meet you where you are Sure. Make sure our software is useful to you Mhmm. And then sort of doesn't matter. Mhmm. What helps us is the industry is so small that everybody talks to each other, wink, wink, works, and that's fine by me.

Speaker 2: Yeah. On the data side, simultaneously, the laws of physics are completely open source but the actual It is open source. The actual design of semiconductors Is not. Is well guarded.

Speaker 1: Discovered physics with s p g but I haven't, you know But

Speaker 2: now it's in the training data. But but but you see what I'm saying like like you have you have the most easy to implement widely, look it up in any textbook And then you have the stuff that, you know, is is sent from NVIDIA to TSMC and is not making it to the open web. It's not in the pre training data. So is there a gap to close? Are you hiring semiconductor designers to do... Or to build RL environments for you? Like, what what actually moves the needle, or is just the laws of physics, is that enough?

Speaker 6: So the laws of physics is enough, actually, because the laws of physics inside the firewall of TSMC Mhmm. Is exactly the same as Yeah. The fire... Outside of the firewall of TSMC. I am yet to meet a company that invented physics, and I don't think I'll meet them.

Speaker 2: Mhmm.

Speaker 6: And we have trained a model on the laws of physics. And time of usage, it will it will produce an answer that abides by it. So if that's the case, we don't need to know the designs that NVIDIA, Micron, TSMC, Samsung are working on.

Speaker 2: Mhmm.

Speaker 6: That's why we deploy and we make it work out of the box. Sure. Does it mean we have to... So we are today training on terabyte of data. Yeah. And we will go to 10 x, 100 x over the next couple of quarters. So what is the expansion? So the expansion is... Oh, from thermal to thermal mechanical to electromagnetic. So basically cover every physics modality that people can write equations about. Mhmm. So to to the point earlier made, there is something new physics that somebody's gonna discover. We haven't embedded that. But the discoverable physics is a freaking fair game. Mhmm.

Speaker 2: Well, congratulations and thank you so much for coming.

Speaker 1: Oh,

Speaker 2: yes. Massive fundraise. 250. Came into the round.

Speaker 1: Quarter bell. Thank you. Thank you.

Speaker 6: I appreciate it. For having

Speaker 1: others. Wow. Amazing progress.

Speaker 2: Coastal Ventures, AMD.

Speaker 1: Yes. Only three years

Speaker 6: And AMD and Applied Materials as well.

Speaker 3: Very, very good.

Speaker 6: AMD and Applied, two corporate ventures.

Speaker 1: Yeah.

Speaker 2: Amazing. Amazing. Well, congratulations and thank you so Great much to meet you. For the

Speaker 1: And congrats to the team on the milestone.

Speaker 2: We'll talk to you later. Have a good one.

Speaker 6: Appreciate it. Thank you. Cheers. Thank you. Goodbye.

Speaker 2: Well, I think it was a massively successful show.

Speaker 1: Massive.

Speaker 2: That was a hilarious interview. I learned a lot genuinely.

Speaker 1: And not not Vinci. Oh,

Speaker 2: I'm going back to the

Speaker 1: Christmas tree one. To be clear. Yeah. That was that was one of the tougher moments on the show for me. I was, you know, trying to look at the monitor.

Speaker 2: The team was not helping.

Speaker 1: The team was... The team was was making Focus. My I was I was in a I was in a dark place. It's October And couldn't keep it together and I had to run off the set.

Speaker 2: The team sent also...

Speaker 1: And and then Nick ran out behind me. And he was he was...

Speaker 2: And the team sent this meme of the sitting forward and that was just the death of me at the time. Was rough. It was a good time. But we learned a lot and that's not the end of our Christmas programming, folks. We got more Christmas related guests coming on the show.

Speaker 1: Didn't ask. We didn't listen.

Speaker 2: Yeah. We got some good stuff.

Speaker 1: No. It's actually fascinating.

Speaker 2: No. It's so interesting. It's genuinely

Speaker 1: Being in a business where you're making something that you're gonna sell in eight years is like an incredible challenge.

Speaker 2: Yo. Totally. Yeah. And... Yeah.

Speaker 1: And competing in a market where there's a there's a war on your industry.

Speaker 2: It's true. Yeah. I mean, there really is. Once he kind of broke down the tariff perspective, it makes a lot more sense. Anyway...

Speaker 1: The only thing I didn't get to ask is do you still hear the bells? But I already knew the answer. Mhmm. Of course he does.

Speaker 2: Of course.

Speaker 1: Of course he does. Of course. We're gonna be making a a meaningful donation Yes. Their foundation and we encourage you to as well.

Speaker 2: Remember that is treesfortroops.org.

Speaker 1: This is financial advice. This is financial advice. Do it. Yeah. You for

Speaker 2: tuning into TBPN today.

Speaker 1: It's been an honor.

Speaker 2: Fantastic show. We'll see you tomorrow

Speaker 1: at I 11AM can't wait.

Speaker 2: Leave us five stars on Apple Podcast and Spotify. Sign up for our newsletter at tbpn.com, and

Speaker 6: we will

Speaker 2: see you tomorrow. Flashback. Have a wonderful Goodbye.