1 00:00:00,280 --> 00:00:02,400 Hi, and welcome to the Neil Ashton Podcast. 2 00:00:03,080 --> 00:00:05,920 In each episode, we explained some of the fascinating ways 3 00:00:05,920 --> 00:00:09,080 that science and engineering are changing the world around us. 4 00:00:09,800 --> 00:00:12,760 We talked to leading engineers from elite level sports like 5 00:00:12,840 --> 00:00:16,920 cycling and Formula One to some of the world's top academics to 6 00:00:16,920 --> 00:00:21,000 understand how fluid dynamics, machine learning, supercomputing 7 00:00:21,360 --> 00:00:23,000 are bringing in a new era of discovery. 8 00:00:23,960 --> 00:00:27,080 We also hear some of their life stories, their career advice, 9 00:00:27,640 --> 00:00:30,080 the lessons they've learned on the way that I hope will be 10 00:00:30,080 --> 00:00:33,840 helpful to you too. So sit back and enjoy this 11 00:00:33,840 --> 00:00:41,880 episode. Hi, and welcome back to the Neil 12 00:00:41,880 --> 00:00:44,680 Ashton Podcast. So today have a very special 13 00:00:44,680 --> 00:00:50,280 guest in Prith Banerjee, who is the CTO of ANSYS, somebody who I 14 00:00:50,280 --> 00:00:56,360 am honored to have on the podcast because he truly is at 15 00:00:56,360 --> 00:01:00,560 the position with the knowledge to answer many of the questions 16 00:01:00,560 --> 00:01:03,840 that I have been wondering myself, but also asking many of 17 00:01:03,840 --> 00:01:07,560 the guests on this podcast. So to ask the CTO of one of the 18 00:01:07,880 --> 00:01:12,680 the largest and most important CAE companies in the world was a 19 00:01:12,680 --> 00:01:15,600 great honour. And I hope it, it's good for you 20 00:01:15,600 --> 00:01:18,800 as well to actually hear from, you know, the person really at 21 00:01:18,800 --> 00:01:20,360 the top of one of these big companies. 22 00:01:21,040 --> 00:01:24,200 And he's an amazing individual. Actually, I, I watched some 23 00:01:24,200 --> 00:01:28,760 videos of interviews with him over the past few months and I 24 00:01:28,760 --> 00:01:34,280 was so impressed by his understanding of these emerging 25 00:01:34,280 --> 00:01:39,440 areas, but also the way that he was able to explain it in such a 26 00:01:39,440 --> 00:01:42,040 simple way. And you'll see him do this in 27 00:01:42,040 --> 00:01:47,440 the in the interview today that really shows that professor in 28 00:01:47,440 --> 00:01:50,440 him. And actually, let's talk about 29 00:01:50,440 --> 00:01:52,520 what his background is. Well, he was a professor for 30 00:01:52,520 --> 00:01:56,280 more than 20 years, publishing more than 350 papers, 31 00:01:56,280 --> 00:01:58,400 supervising, you know, nearly 40 students. 32 00:01:58,560 --> 00:02:02,600 So really had an amazing career on its own as a professor in 33 00:02:02,600 --> 00:02:06,200 Illinois, but then went off to the start up world. 34 00:02:06,440 --> 00:02:09,960 And we discussed a lot about this need for people to 35 00:02:09,960 --> 00:02:12,720 sometimes go from academia to startups to, you know, fully 36 00:02:12,720 --> 00:02:16,840 exploit the ideas they have. But then he went into the 37 00:02:16,840 --> 00:02:20,720 corporate world and became, you know, CTO of companies like 38 00:02:20,720 --> 00:02:26,160 ABBHP Labs, Schneider Electric, and now ANSYS for the past six 39 00:02:26,160 --> 00:02:28,560 years. What an incredible individual to 40 00:02:28,680 --> 00:02:32,160 have gone through those three sort of main stages, I guess, 41 00:02:32,160 --> 00:02:35,680 of, of, of the world that you could be in, you know, academia, 42 00:02:35,680 --> 00:02:39,000 startups and and industry. It's amazing because it's also 43 00:02:39,000 --> 00:02:41,680 one of those questions I've often asked people on the show, 44 00:02:42,120 --> 00:02:44,400 you know what, what do you think about the differences? 45 00:02:44,400 --> 00:02:48,520 So here's somebody who's, you know, done it all and I really 46 00:02:48,520 --> 00:02:52,240 wanted to ask him some of the topics that I personally have 47 00:02:52,240 --> 00:02:55,680 found interesting at, but I think the community at large who 48 00:02:55,680 --> 00:03:00,360 are into fluid dynamics and CFD and HPC and AI are wondering. 49 00:03:00,880 --> 00:03:04,200 So I, I put it to him as the CTO of one of the biggest companies 50 00:03:04,200 --> 00:03:07,600 in the world. So we had a really deep and I 51 00:03:07,600 --> 00:03:10,920 put interesting discussion about the role of machine learning and 52 00:03:10,920 --> 00:03:14,160 artificial intelligence in CAE. We've already dived into some of 53 00:03:14,160 --> 00:03:17,560 the details, discussed quite a length about foundational 54 00:03:17,560 --> 00:03:19,400 models. He came out with some really 55 00:03:19,400 --> 00:03:23,680 interesting stuff and, and the honesty that he had as CTO to 56 00:03:23,680 --> 00:03:26,560 explain to his board that this really is an important thing 57 00:03:26,560 --> 00:03:29,840 that could even see the end or the simulation market as we know 58 00:03:29,840 --> 00:03:33,160 if they don't fully embrace it. So we talked a lot about that. 59 00:03:33,360 --> 00:03:37,360 We talked about quantum computing, how that could be a 60 00:03:37,360 --> 00:03:40,760 sign of things to come, some changes which answers have been 61 00:03:40,760 --> 00:03:42,760 working on. We touched on, you know, 62 00:03:42,760 --> 00:03:48,160 HBCGPUS, but we also talked a lot about the role of startups, 63 00:03:48,200 --> 00:03:51,520 the roles of industry, what startups should be trying to do. 64 00:03:51,720 --> 00:03:55,960 And we talked some advice for students, mid Korea and 65 00:03:55,960 --> 00:03:59,440 everybody about, you know, what they could do to maybe come up 66 00:03:59,440 --> 00:04:03,640 with the next amazing invention. We touched on open source, 67 00:04:03,640 --> 00:04:07,000 closed source and how we need to work with academia. 68 00:04:07,960 --> 00:04:13,800 And then, you know, we finally ended on some advice, I guess, 69 00:04:13,800 --> 00:04:17,160 to, to, to people and, and really finished on what he is 70 00:04:17,160 --> 00:04:19,959 quite an inspiring individual. You know, he's wrote a book. 71 00:04:19,959 --> 00:04:22,720 It's really amazing, the innovation factory. 72 00:04:22,800 --> 00:04:24,960 I can put the link in the YouTube if you're watching it. 73 00:04:25,640 --> 00:04:29,520 And on that note, you know, if you enjoy this, it really would 74 00:04:29,520 --> 00:04:33,080 appreciate it if you did, you know, like it, subscribe it. 75 00:04:33,080 --> 00:04:36,000 The algorithms work that way. If you if you like it but don't 76 00:04:36,000 --> 00:04:38,240 interact, unfortunately, that makes it harder for others to 77 00:04:38,240 --> 00:04:40,240 find it. So I don't often say this, but 78 00:04:40,240 --> 00:04:43,680 I'll I'll say it once every few episodes just because it would 79 00:04:43,680 --> 00:04:46,000 help. And also if you're watching this 80 00:04:46,000 --> 00:04:48,680 on YouTube right now, just to let you know, this is actually 81 00:04:48,680 --> 00:04:52,640 also available in audio only on Spotify and Apple and vice 82 00:04:52,640 --> 00:04:54,080 versa. If you're listening to this and 83 00:04:54,080 --> 00:04:57,520 you weren't aware, there is also a video version on YouTube. 84 00:04:57,520 --> 00:05:02,120 So yeah, I, I, I really was so pleased that he was willing to 85 00:05:02,120 --> 00:05:03,960 speak. I found this conversation so 86 00:05:03,960 --> 00:05:06,200 interesting, and I hope you do too. 87 00:05:06,200 --> 00:05:09,720 So sit back and listen to this interview with Prith Banerjee. 88 00:05:10,160 --> 00:05:13,640 What was your journey to being the CTO of one of the most 89 00:05:13,640 --> 00:05:16,840 important and biggest simulation companies in the world? 90 00:05:16,840 --> 00:05:19,080 How did you how did you get there? 91 00:05:19,080 --> 00:05:21,640 I'm sure others would love to have your position and your job. 92 00:05:21,640 --> 00:05:24,240 So could you tell me a little bit more about your career and 93 00:05:24,240 --> 00:05:26,000 how you got to where you are today? 94 00:05:26,480 --> 00:05:29,640 So, so Neil, first of all, thank you very much for inviting me to 95 00:05:29,640 --> 00:05:31,960 this. So I started my career in 96 00:05:31,960 --> 00:05:34,880 academia. I have got my PhD in Electrical 97 00:05:34,880 --> 00:05:37,080 and Computer Engineering from the University of Illinois 98 00:05:37,200 --> 00:05:41,560 Urbana Champagne, and I started as a professor at Urbana. 99 00:05:41,560 --> 00:05:45,920 I spent the first dozen years going through the ranks becoming 100 00:05:45,920 --> 00:05:51,320 a full professor, and I was the founding Director of 101 00:05:51,960 --> 00:05:55,920 Computational Science and Engineering Program at UIUC. 102 00:05:55,920 --> 00:05:58,760 Illinois is as this National Center for supercomputing 103 00:05:58,760 --> 00:06:01,680 applications, a big place for HPC. 104 00:06:02,560 --> 00:06:05,880 And I used to do and my research was on developing parallel 105 00:06:05,880 --> 00:06:07,640 algorithms and parallel compilers. 106 00:06:07,640 --> 00:06:10,920 So I've always been working in the HPC area. 107 00:06:12,160 --> 00:06:15,280 So my last two years at Illinois, I was a founding 108 00:06:15,280 --> 00:06:17,760 director of computational science and engineering, which 109 00:06:17,760 --> 00:06:21,800 is the field of computing of high performance computing using 110 00:06:22,080 --> 00:06:25,160 HPC to drive sort of science and engineering. 111 00:06:25,160 --> 00:06:27,480 So computational physics, computational chemistry, 112 00:06:27,480 --> 00:06:29,760 computational electromagnetics, all of those things. 113 00:06:29,800 --> 00:06:34,560 And as it turns out, 30 years later, I have landed up at in 114 00:06:34,560 --> 00:06:35,840 this job. So that's sort of the 115 00:06:35,840 --> 00:06:38,440 connection. And then after Illinois, I went 116 00:06:38,440 --> 00:06:40,600 to Northwestern. I was then at the University of 117 00:06:40,600 --> 00:06:43,120 Illinois Chicago. So hardcore academic for about 118 00:06:43,120 --> 00:06:47,880 20 plus years. After that, I made a hard turn 119 00:06:47,880 --> 00:06:52,560 into the corporate world. I was head of HP Labs and in at 120 00:06:52,560 --> 00:06:55,840 HP Labs I used to lead a lot of work on on high performance 121 00:06:55,840 --> 00:06:57,680 computing. We used to build this really 122 00:06:57,680 --> 00:07:01,680 super duper high performance servers, so a lot of cool work 123 00:07:01,680 --> 00:07:04,600 there. And then I became CTO at EBB, a 124 00:07:04,800 --> 00:07:07,800 power and automation company based in Zurich. 125 00:07:07,800 --> 00:07:10,760 And then I became CTO at Schneider Electric and another 126 00:07:10,760 --> 00:07:12,600 power automation company based in France. 127 00:07:13,120 --> 00:07:19,000 About 6 1/2 years ago I I joined ANSYS as the CTO. 128 00:07:19,000 --> 00:07:24,560 So this is my third CTO job. And what Ansys does is we are 129 00:07:24,560 --> 00:07:28,120 the leading modeling and simulation company in the world. 130 00:07:28,240 --> 00:07:31,920 We take the world around us, which is governed by the laws of 131 00:07:31,920 --> 00:07:34,880 physics. And we take that physics, which 132 00:07:34,880 --> 00:07:38,640 is explained as second order partial differential equations. 133 00:07:38,640 --> 00:07:42,120 And we solve those physics through finite element methods, 134 00:07:42,120 --> 00:07:46,640 finite volume methods using things like FLUENT, which is our 135 00:07:46,640 --> 00:07:49,680 fluid score, in things like mechanical, which is our 136 00:07:49,680 --> 00:07:52,280 structural code, in things like HFSS, which is an 137 00:07:52,280 --> 00:07:56,400 electromagnetic score. And my role as CTO is to look at 138 00:07:57,000 --> 00:07:59,920 the all these amazing products, what is the future of 139 00:07:59,920 --> 00:08:02,600 simulation? What kind of technologies can be 140 00:08:02,600 --> 00:08:09,960 used to drive future products? And in my role as CTO, I look at 141 00:08:09,960 --> 00:08:12,560 things like AI, machine learning, right? 142 00:08:12,560 --> 00:08:15,080 How does AIML improve simulation? 143 00:08:15,480 --> 00:08:19,120 HPC, how do you use HPC to accelerate simulation? 144 00:08:19,600 --> 00:08:22,320 How, what do you do with sort of cloud, right? 145 00:08:22,320 --> 00:08:24,360 What do you do with platforms or digital engineering? 146 00:08:24,360 --> 00:08:27,640 So that is I have the coolest job in the company. 147 00:08:27,640 --> 00:08:30,800 Yeah, looking at the future future of simulation. 148 00:08:31,760 --> 00:08:34,600 Yeah, which is why you're absolutely perfect guest on this 149 00:08:34,600 --> 00:08:37,679 podcast, because your job is literally to answer, I guess 150 00:08:37,679 --> 00:08:39,720 some of the questions that that that people have. 151 00:08:40,000 --> 00:08:45,480 But maybe I love the fact that you have had such a a great 152 00:08:45,480 --> 00:08:48,480 academic career and going into industry. 153 00:08:48,480 --> 00:08:50,680 And it's one of the themes I often ask people, you know, 154 00:08:50,720 --> 00:08:53,840 academia or industry, what's the benefits of both? 155 00:08:54,240 --> 00:08:59,840 So what do you now, having done both, what do you see as the 156 00:08:59,840 --> 00:09:02,840 role of academia? Where, where can academia help, 157 00:09:02,840 --> 00:09:07,720 let's say, in advancing CAE and where does industry need to do 158 00:09:07,720 --> 00:09:09,120 it? And where is the overlap? 159 00:09:09,400 --> 00:09:12,080 Absolutely. So so since you're asking a 160 00:09:12,080 --> 00:09:16,600 career question, I I actually bypassed one part of my career. 161 00:09:17,160 --> 00:09:19,680 So I've actually had been 3 phases in my career. 162 00:09:19,680 --> 00:09:23,880 I was in academia for 20 years, but in between Academy in the 163 00:09:23,880 --> 00:09:26,200 large corporate world, I was in the startup world. 164 00:09:26,200 --> 00:09:30,880 I did two start-ups, One was Excel Chip, 1 was Banachip. 165 00:09:30,880 --> 00:09:34,600 And these were companies started out of technologies from the 166 00:09:34,600 --> 00:09:37,400 university, from the one from Northwest and one from the 167 00:09:37,400 --> 00:09:42,280 University of Illinois. And I did those while in 168 00:09:42,280 --> 00:09:44,400 universities you can actually go on sabbatical. 169 00:09:44,400 --> 00:09:47,120 So I left, I took leave from the university, did my first 170 00:09:47,120 --> 00:09:50,240 startup, came back to the university, the second startup 171 00:09:50,240 --> 00:09:54,640 came back to the university. So, and literally the reason I 172 00:09:54,640 --> 00:09:57,760 went from the academic world to the corporate world is because 173 00:09:57,760 --> 00:10:00,880 of the startups, right? So in the, So now let me ask you 174 00:10:00,880 --> 00:10:04,760 the question. In academia, what people do is 175 00:10:04,760 --> 00:10:07,560 to solve fundamental problems, right? 176 00:10:07,560 --> 00:10:12,440 Really, I mean what I call Horizon 3 futuristic research 177 00:10:12,440 --> 00:10:15,000 problems, right? Where we are trying to really 178 00:10:15,000 --> 00:10:20,200 understand what is the absolute the fundamentals of of, of 179 00:10:20,200 --> 00:10:23,600 technology, right And he worked with graduate students and I 180 00:10:23,600 --> 00:10:28,560 have had in my 20 plus career right in academia, I have had 37 181 00:10:28,560 --> 00:10:32,120 PhD students 40 plus masters students with whom I have 182 00:10:32,120 --> 00:10:36,760 published more than 350 technical papers in IEEE 183 00:10:36,760 --> 00:10:39,840 conference in this and IEEE transactions of that and so on 184 00:10:39,840 --> 00:10:42,040 so forth. So that's the world of academia 185 00:10:42,040 --> 00:10:44,880 where you're, you're researching, you're discovering 186 00:10:44,880 --> 00:10:50,240 new things and you're publishing that work in the latest journals 187 00:10:50,240 --> 00:10:53,600 and conferences, right? It's all about creating new 188 00:10:53,600 --> 00:10:57,960 knowledge and then transferring that knowledge to brilliant 189 00:10:58,240 --> 00:11:00,920 students, right? So you are educating the 190 00:11:00,920 --> 00:11:02,800 workforce in the next World, right? 191 00:11:02,800 --> 00:11:07,360 So in academia you have two roles. 1 is invent, create 192 00:11:07,360 --> 00:11:09,200 knowledge, right? Discover knowledge which you 193 00:11:09,200 --> 00:11:13,080 publish and then you train students with the knowledge that 194 00:11:13,080 --> 00:11:15,360 you have created, right? Train undergraduate students, 195 00:11:15,360 --> 00:11:19,280 graduate students and so on, which are the workforces for all 196 00:11:19,280 --> 00:11:22,760 of us, right, in academia and in the corporate world to do. 197 00:11:23,480 --> 00:11:27,360 But what academia does not do is we don't build products, right? 198 00:11:27,680 --> 00:11:31,720 And literally, Neil, the reason I did the startups was I was 199 00:11:31,720 --> 00:11:37,000 frustrated that I was doing all this work, 350 papers, 10 plus 200 00:11:37,000 --> 00:11:38,440 pattern, doing all kinds of stuff. 201 00:11:38,440 --> 00:11:41,920 But nobody cared. Nobody gave a damn right there 202 00:11:41,920 --> 00:11:43,840 was because it was not showing up in any product. 203 00:11:44,960 --> 00:11:49,720 So when I accessed it was actually created when I ended a 204 00:11:50,200 --> 00:11:53,440 DARPA project called the match Compiler and the DARPA PM said, 205 00:11:53,800 --> 00:11:56,760 great, this is really awesome. You should should transfer it to 206 00:11:56,760 --> 00:11:58,520 a company. So I came to the Bay Area, 207 00:11:58,520 --> 00:12:01,880 talked to various companies and say, would you like to use this 208 00:12:01,880 --> 00:12:04,520 technology? I say absolutely, this looks so 209 00:12:04,520 --> 00:12:06,680 good, just leave the the software copy with us. 210 00:12:06,680 --> 00:12:10,080 And I looked at them in the eye and said there's no way they are 211 00:12:10,080 --> 00:12:13,520 going to take this software like the only way this really 212 00:12:13,520 --> 00:12:17,320 commercialize if I were to do it myself with my graduate 213 00:12:17,320 --> 00:12:19,960 students. So that's kind of why I started 214 00:12:19,960 --> 00:12:25,400 the first company at surgery. So startups, what they do is 215 00:12:25,400 --> 00:12:30,120 they actually take a really new idea, something that the world 216 00:12:30,120 --> 00:12:34,480 has not seen before and get laser focused on that idea and 217 00:12:34,480 --> 00:12:38,760 they bring bring that that new product to to the market, right. 218 00:12:39,840 --> 00:12:43,560 And I did two of those startups myself and then I came to the 219 00:12:43,560 --> 00:12:46,280 large corporate world of HPABB and so on, right? 220 00:12:47,160 --> 00:12:52,120 But what I have found is the large companies, they don't have 221 00:12:52,120 --> 00:12:56,160 a single product like ANSYS. We have 70 products, right, in 222 00:12:56,160 --> 00:12:58,120 simulation, right? We have ANSYS Mechanical, we 223 00:12:58,120 --> 00:13:02,400 have LS Dyna, we have Fluent, we have this twin builder, all 224 00:13:02,400 --> 00:13:05,400 kinds of products, right? And the role of a large company 225 00:13:05,400 --> 00:13:08,600 to take these products and evolve their products, right, 226 00:13:08,840 --> 00:13:12,760 doing continuous innovation. What features should I have in 227 00:13:12,760 --> 00:13:16,880 the next release of Fluent, the next release of, of mechanical 228 00:13:16,880 --> 00:13:19,360 and so on? So, but the innovation that 229 00:13:19,360 --> 00:13:23,960 happens in in the corporate world is more incremental, 230 00:13:23,960 --> 00:13:26,640 right? It is what I call Horizon one. 231 00:13:26,760 --> 00:13:30,000 I have a product. So I used to work at HP, right? 232 00:13:30,000 --> 00:13:33,360 You make computers. So, so next version of laptop, 233 00:13:33,360 --> 00:13:35,240 right? Is an incremental very 234 00:13:35,240 --> 00:13:38,120 important, but something that that you need to do. 235 00:13:38,520 --> 00:13:42,200 We have ANSYS, ANSYS mechanical, it's a finite element based 236 00:13:42,440 --> 00:13:45,280 structural solver. We are doing the next version, 237 00:13:45,280 --> 00:13:47,440 right? It's faster, it's a little 238 00:13:47,440 --> 00:13:50,040 better convergence, better meshing, but it's still the same 239 00:13:50,040 --> 00:13:52,840 tool, right? So that's what large companies 240 00:13:52,840 --> 00:13:56,720 do academia, we invent new things, right? 241 00:13:56,720 --> 00:13:59,080 We are doing a hierarchical octree to measure 242 00:13:59,080 --> 00:14:01,200 representation, whatever and you publish a paper and you're 243 00:14:01,200 --> 00:14:03,360 getting a patent and so on. But that is not a product. 244 00:14:04,000 --> 00:14:08,240 What a startups do is take that work in academia and they 245 00:14:08,280 --> 00:14:12,840 package it up into is really brilliant disruptive innovation, 246 00:14:12,840 --> 00:14:15,640 which I call Horizon 3 innovation, right? 247 00:14:15,640 --> 00:14:19,400 The truly disruptive innovation always happens in startups. 248 00:14:19,840 --> 00:14:23,320 Large companies actually struggle with with, with with 249 00:14:23,920 --> 00:14:27,280 disruptive innovation. In fact, Neil, I have done 250 00:14:27,360 --> 00:14:29,680 broadcasts on this. I have written a book called The 251 00:14:29,680 --> 00:14:33,000 Innovation Factory, which your readers may be interested in. 252 00:14:33,000 --> 00:14:37,880 And the whole premise of this book is how does a large company 253 00:14:37,880 --> 00:14:43,440 like ABB or Schneider or or HP or or Ansys, the companies that 254 00:14:43,440 --> 00:14:45,840 have actually worked in the role of CTO, right? 255 00:14:46,120 --> 00:14:52,520 How do these companies try to foster Horizon 3 disruptive 256 00:14:52,520 --> 00:14:54,800 innovation, right? Large companies doing disruptive 257 00:14:54,800 --> 00:14:59,840 innovation and what I say in my book is they they do it through 258 00:14:59,840 --> 00:15:02,960 partnership with academia because academia is where the 259 00:15:02,960 --> 00:15:06,600 research in this future directions is happening and with 260 00:15:06,600 --> 00:15:10,480 startups and bring those so academia and startups to this 261 00:15:10,480 --> 00:15:12,840 thing in a concept called open innovation. 262 00:15:13,320 --> 00:15:16,520 And that is what I'm truly passionate about. 263 00:15:16,880 --> 00:15:18,760 So I know you asked me a question about the difference 264 00:15:18,760 --> 00:15:20,560 between academia and a large world. 265 00:15:21,800 --> 00:15:26,920 Academia does discovery of knowledge Horizon 3, but they 266 00:15:26,920 --> 00:15:30,800 don't actually make products this disruptive innovation. 267 00:15:30,840 --> 00:15:34,720 There are people like me who leave academia and they build a 268 00:15:34,720 --> 00:15:38,520 destructive thing. But in a concept of a startup, a 269 00:15:38,520 --> 00:15:42,560 startup is laser focused on that one product, right? 270 00:15:42,560 --> 00:15:46,080 The world that has not seen right, very destructive, but 271 00:15:46,080 --> 00:15:47,720 that's the only thing that they do right. 272 00:15:47,880 --> 00:15:49,480 So they're they're focused on it. 273 00:15:49,680 --> 00:15:51,760 And then they'll do the second product and the third product. 274 00:15:51,760 --> 00:15:55,400 Ultimately that will also become a large company, at which point 275 00:15:55,520 --> 00:15:58,480 it will stop doing Horizon 3 innovation. 276 00:15:58,480 --> 00:16:02,840 It will become like they AB BS of the world right until they 277 00:16:02,840 --> 00:16:05,240 start. They then start working with 278 00:16:05,240 --> 00:16:07,000 with with other other startup companies. 279 00:16:07,800 --> 00:16:09,200 Yeah. I really like how you put that, 280 00:16:09,200 --> 00:16:11,120 that that's kind of what I was getting at. 281 00:16:11,120 --> 00:16:14,960 And I have to be honest, particularly coming from Europe, 282 00:16:15,560 --> 00:16:20,080 I think, and that it is slightly changing now, there was nowhere 283 00:16:20,080 --> 00:16:23,920 near the same startup culture. And it's felt like you're an 284 00:16:23,920 --> 00:16:26,440 academia. You know, it was almost a dirty 285 00:16:26,440 --> 00:16:29,280 word to try and commercialize what you were doing. 286 00:16:29,600 --> 00:16:31,320 You know, that's not pure academia. 287 00:16:31,320 --> 00:16:35,120 You know, you just publish. And then there was the large 288 00:16:35,120 --> 00:16:38,560 companies, you know, the Rolls Royces or whatever of the world 289 00:16:38,560 --> 00:16:41,600 that I remember were funding it, But it always felt like the 290 00:16:41,600 --> 00:16:44,600 technology transfer wasn't the same. 291 00:16:44,880 --> 00:16:47,720 Now, having worked for AUS company and and being more 292 00:16:47,720 --> 00:16:51,160 exposed to the Bay Area, I'm kind of seeing how you're right. 293 00:16:51,160 --> 00:16:53,960 This start-ups. It seems like it's the it's the 294 00:16:53,960 --> 00:16:59,560 mechanism in between that allows these new ideas to to form. 295 00:16:59,560 --> 00:17:04,040 But what are there any from your time now? 296 00:17:04,119 --> 00:17:07,800 I guess looking at start-ups, but also having run a start up, 297 00:17:08,599 --> 00:17:11,720 what sort of general advice would you give to start-ups? 298 00:17:11,760 --> 00:17:14,160 I know this is a very difficult question to answer, but you 299 00:17:14,160 --> 00:17:21,359 know, yeah, like would you, do you go in with the mindset that 300 00:17:21,359 --> 00:17:23,720 someone's going to buy you? Do you go in the mindset that 301 00:17:23,720 --> 00:17:26,400 you are going to be the next big company? 302 00:17:26,400 --> 00:17:29,080 You know, how do you think about that? 303 00:17:29,080 --> 00:17:32,000 Or advice you would give maybe to start-ups trying to come up 304 00:17:32,000 --> 00:17:34,360 with new ideas. The way I would think about a 305 00:17:34,360 --> 00:17:39,000 start up is if you're doing a start up just to make money, 306 00:17:39,640 --> 00:17:43,120 you've got the wrong motivation. The motivation is really you are 307 00:17:43,120 --> 00:17:47,680 trying to solve a problem that the world has, right? 308 00:17:47,680 --> 00:17:51,280 And you see no solution, right? There's no existing solution 309 00:17:51,280 --> 00:17:54,640 from the large companies, right? I mean, you're trying to do this 310 00:17:54,680 --> 00:17:57,880 fantastic computer that will solve the world's problems, 311 00:17:57,880 --> 00:18:01,000 right? I mean, and the world doesn't 312 00:18:01,000 --> 00:18:04,000 have that tool, that solution today. 313 00:18:05,560 --> 00:18:09,960 And you have you are maybe half the time the startup founders 314 00:18:09,960 --> 00:18:11,560 actually come from large companies, right? 315 00:18:12,440 --> 00:18:15,800 And they say they see a problem and they say, you know what, I'm 316 00:18:15,800 --> 00:18:18,400 going to solve this, right? And typically in a large 317 00:18:18,400 --> 00:18:22,920 company, the manager will allow you to only work on things that 318 00:18:22,920 --> 00:18:25,960 are incremental, right? So you have, as I said, you are 319 00:18:25,960 --> 00:18:27,920 working in HP or making laptops, right? 320 00:18:28,240 --> 00:18:32,160 If you say to HPI want to build a quantum computer, right? 321 00:18:32,160 --> 00:18:34,600 You imagine we say go away, that that's not what we do, right? 322 00:18:35,440 --> 00:18:40,520 So but often times these problems come out and look at 323 00:18:40,520 --> 00:18:42,600 you and say this needs to be solved. 324 00:18:42,600 --> 00:18:45,240 And you are, you are just, you have this burning passion to 325 00:18:45,360 --> 00:18:49,560 solve that problem. And you sometimes your company 326 00:18:49,560 --> 00:18:51,160 manager will allow you to do it right. 327 00:18:51,160 --> 00:18:52,920 Then you're lucky. Then the company is actually 328 00:18:52,920 --> 00:18:54,800 allowing you to do Horizon 3 innovation. 329 00:18:55,200 --> 00:18:59,240 But 90% of the time you will not be able to do it right. 330 00:18:59,240 --> 00:19:00,960 And then you say, what, what choice do I have? 331 00:19:00,960 --> 00:19:04,320 You should then do a start up, try to follow your passion, 332 00:19:04,320 --> 00:19:05,800 follow your dreams and do it right. 333 00:19:05,800 --> 00:19:09,440 That's how most entrepreneurs start startups, right? 334 00:19:11,160 --> 00:19:14,520 The other way is for academia, academic people, right? 335 00:19:14,520 --> 00:19:16,960 And so literally startups come from 2 ends. 336 00:19:17,000 --> 00:19:21,000 Either it's an academic who has solved a really hard problem and 337 00:19:21,000 --> 00:19:24,320 say, OK, now we want to commercialize it like me. 338 00:19:24,320 --> 00:19:27,440 And again, I am just a a very small person, but there's so 339 00:19:27,440 --> 00:19:30,880 many more famous people who came from Academy and some absolutely 340 00:19:30,880 --> 00:19:33,720 wonderful companies, right? And I mentioned them in my book. 341 00:19:34,160 --> 00:19:37,680 And and then there's this startup that happened from. 342 00:19:37,680 --> 00:19:41,320 So I would say 80% of the startup founders actually come 343 00:19:41,320 --> 00:19:42,840 from the large corporate world, right? 344 00:19:42,840 --> 00:19:46,360 And then they have found a problem solve it and then they 345 00:19:46,680 --> 00:19:50,960 start one company, they start a second company now with you 346 00:19:50,960 --> 00:19:55,200 asked a question, what shop does ultimately, yes, so you, you get 347 00:19:55,200 --> 00:19:57,080 motivated by solving the world's problems. 348 00:19:57,080 --> 00:19:59,120 But of course there is a second motivation. 349 00:19:59,120 --> 00:20:01,440 I I would like to make some money out of it, right. 350 00:20:02,520 --> 00:20:07,600 So the way you pick a problem, right, you should pick a problem 351 00:20:07,600 --> 00:20:13,800 that has a large market, right? And so how do you establish the 352 00:20:13,800 --> 00:20:16,080 market? That is the hardest thing for a 353 00:20:16,080 --> 00:20:18,320 startup entrepreneur to do right and. 354 00:20:18,760 --> 00:20:22,280 So often times you say, well, what's the market for GPUs? 355 00:20:22,280 --> 00:20:26,680 Well, you can take the look at, look at NVIDIA and and AMD and 356 00:20:26,680 --> 00:20:28,560 so on. And it's OK, These are people 357 00:20:28,560 --> 00:20:31,320 who are making GPUs, they are selling this many GPUs. 358 00:20:31,320 --> 00:20:32,720 And so the market for GPU is this. 359 00:20:32,720 --> 00:20:36,200 And if you are a new startup and you do another GPU, you know 360 00:20:36,200 --> 00:20:37,800 exactly what that market is, right? 361 00:20:39,240 --> 00:20:41,360 What's the market for, for eyeglasses? 362 00:20:41,360 --> 00:20:43,400 Eyeglasses. You look at all the people who 363 00:20:43,400 --> 00:20:45,960 are wearing eyeglasses. You can say that, but suppose 364 00:20:45,960 --> 00:20:51,720 you are a startup you have you are inventing a device such as 365 00:20:51,720 --> 00:20:57,960 blind men can see. OK, that device does not exist. 366 00:20:58,360 --> 00:21:02,600 You do a Google search of market for device for blind men per C 367 00:21:02,960 --> 00:21:05,720 is 0 because there is no product in that area. 368 00:21:05,760 --> 00:21:07,120 I mean, I'm just giving an example. 369 00:21:07,120 --> 00:21:09,840 Maybe today there is, but there isn't, right? 370 00:21:10,440 --> 00:21:13,320 So then you say, oh, the market is 0, therefore it's a bad idea. 371 00:21:13,320 --> 00:21:16,280 I should not do it because those marketing things done by 372 00:21:16,280 --> 00:21:18,440 companies like Gartner or Dataquest, right? 373 00:21:18,640 --> 00:21:22,240 They are only looking at at markets where products exist, 374 00:21:22,240 --> 00:21:23,760 right? What's the market for the cloud? 375 00:21:24,120 --> 00:21:30,040 It is $100 billion, right? The market for cloud before Jeff 376 00:21:30,040 --> 00:21:35,960 Bezos invented AWS was 0 right? So right it it took a person of 377 00:21:35,960 --> 00:21:40,760 Jeff's imagination says that the market is this if I could build 378 00:21:40,760 --> 00:21:44,200 it, right? So then for that, that device 379 00:21:44,200 --> 00:21:47,360 that blind men can see, right? I'm the entrepreneur. 380 00:21:47,360 --> 00:21:48,840 I'm trying to to find the market. 381 00:21:48,840 --> 00:21:53,440 I say, well, how many blind men are there in the world, right, 382 00:21:53,520 --> 00:21:56,040 that I know I have 10 billion people on the planet. 383 00:21:57,360 --> 00:21:59,600 I don't know, maybe 3 million people are blind. 384 00:22:00,880 --> 00:22:04,080 How much would they pay for it? Well, I pay, I go to Lens 385 00:22:04,080 --> 00:22:06,480 Trafters and buy these glasses for $200.00. 386 00:22:06,840 --> 00:22:10,400 So at least I'm not blind. But I'm paying something to 387 00:22:10,400 --> 00:22:14,400 improve my vision. So my at least I'll pay 200, 388 00:22:14,400 --> 00:22:18,040 maybe 300. So 300 times 100 million blind 389 00:22:18,040 --> 00:22:20,880 people. That's the $303 billion market. 390 00:22:21,160 --> 00:22:24,880 That's how you size the market. So you have a choice of making a 391 00:22:24,880 --> 00:22:26,800 device such that blind men can see. 392 00:22:27,000 --> 00:22:32,080 The market is 3 billion versus a chair with 9 legs, right? 393 00:22:32,360 --> 00:22:34,320 And the market for that is only $2.00. 394 00:22:34,680 --> 00:22:37,560 You should pick the first one, even though that is a harder 395 00:22:37,560 --> 00:22:41,360 problem to work on because if you're successful, you will 396 00:22:41,360 --> 00:22:44,760 solve the world's problem. And it's a large problem versus 397 00:22:45,240 --> 00:22:48,400 inventing a chair with 9 legs, which is a simple thing because 398 00:22:48,400 --> 00:22:50,480 you know, I have a chair with four legs. 399 00:22:50,600 --> 00:22:54,600 It is easy to do with 9 legs, but the market is only only two. 400 00:22:55,600 --> 00:23:00,240 That's the simplistic way that I can I can I can explain the 401 00:23:00,240 --> 00:23:01,960 world of. Start and I see that a little 402 00:23:01,960 --> 00:23:08,280 bit with simulation is that it is difficult probably in the CFD 403 00:23:08,280 --> 00:23:13,360 world or the CAE world to really appreciate the difference I 404 00:23:13,360 --> 00:23:17,520 guess between theoretical and would anybody actually use it? 405 00:23:18,080 --> 00:23:20,360 You know, like there's a difference between saying, oh, 406 00:23:20,360 --> 00:23:26,320 we could make CFD 10 times faster, but even if it was 10 407 00:23:26,320 --> 00:23:29,040 times faster, it doesn't mean everybody's going to pick your 408 00:23:29,040 --> 00:23:30,760 software because they may not trust you. 409 00:23:30,760 --> 00:23:35,000 They may prefer, you know, So I guess this is where the IT 410 00:23:35,000 --> 00:23:38,600 becomes even harder, doesn't it? When you're, you know, the cloud 411 00:23:38,600 --> 00:23:41,880 was such a massive new thing. It's so clear. 412 00:23:41,880 --> 00:23:47,040 I guess most start-ups are more are not as revolutionary, you 413 00:23:47,040 --> 00:23:48,880 know, and they're probably the harder ones, aren't they? 414 00:23:48,880 --> 00:23:54,200 Because there is a value, but it's sort of harder to to figure 415 00:23:54,200 --> 00:23:56,160 out. And I guess maybe this leads 416 00:23:56,160 --> 00:23:58,800 nicely because one of the things that a lot of people have seen 417 00:23:59,600 --> 00:24:05,720 is a huge growth now in the AIML world, you know, obviously for 418 00:24:05,720 --> 00:24:08,360 large language models. But I think personally, what's 419 00:24:08,360 --> 00:24:13,720 excited me is seeing how much of this is now slowly moving into 420 00:24:13,720 --> 00:24:20,680 the scientific world and the potential impact it has on 421 00:24:21,040 --> 00:24:24,640 accelerating traditional, you know, CAE codes. 422 00:24:24,640 --> 00:24:29,880 And I know you have your own product as well, SIM AI, but I 423 00:24:29,880 --> 00:24:32,880 was just wanting to get maybe some of your thoughts on where 424 00:24:32,880 --> 00:24:39,760 you see the use of AIML today short term and you know, what's 425 00:24:39,760 --> 00:24:42,840 the what's the think big? What's the art of the possible 426 00:24:43,240 --> 00:24:45,440 that you think this could become? 427 00:24:46,320 --> 00:24:48,160 It's great. That's a great question. 428 00:24:48,160 --> 00:24:53,960 So let me explain the my my thought, right, just by going in 429 00:24:53,960 --> 00:24:55,600 the area of simulation itself, right. 430 00:24:55,600 --> 00:24:56,800 So I want to explain the problem. 431 00:24:56,960 --> 00:25:00,280 So when you're looking at simulation of say, a fluids 432 00:25:00,280 --> 00:25:05,920 problem, right, the problem is formulated in the ideal world as 433 00:25:05,920 --> 00:25:09,120 Navier Stokes equations, right? You have the, the governing 434 00:25:09,120 --> 00:25:11,280 equations, you have energy conservation, so on. 435 00:25:11,280 --> 00:25:13,360 And those are second order PDS, right? 436 00:25:13,960 --> 00:25:18,000 So you can write those PDS. And when you went to college, 437 00:25:18,000 --> 00:25:19,840 right? You can take a very simple 438 00:25:20,400 --> 00:25:25,400 differential equation, right? Linear, whatever the simplest 1 439 00:25:25,400 --> 00:25:28,960 you could analytically solve, right, is E to the power -2 440 00:25:28,960 --> 00:25:31,120 whatever sum. This is how the equations go, 441 00:25:31,120 --> 00:25:33,960 right? But in the practical world, 442 00:25:33,960 --> 00:25:37,360 right, these problems have the CAD geometries are so 443 00:25:37,360 --> 00:25:41,280 complicated by the time you take the CAD, define the boundary 444 00:25:41,280 --> 00:25:43,960 conditions and so on, and you have the Navier Stokes equations 445 00:25:44,560 --> 00:25:47,760 to solve it, right? It is impossible to solve it 446 00:25:48,040 --> 00:25:49,840 analytically. So you have to solve it 447 00:25:49,840 --> 00:25:54,120 numerically. So you take those Pdes and you 448 00:25:54,120 --> 00:25:57,880 discretize them, right? So you do say finite elements, 449 00:25:57,880 --> 00:26:01,640 right? You take this whatever kind of 450 00:26:01,640 --> 00:26:04,360 thing and you break it up into 1000 elements, right? 451 00:26:04,520 --> 00:26:07,840 And in each, the finite element method says on each element 452 00:26:07,840 --> 00:26:09,440 those governing equations will work. 453 00:26:09,600 --> 00:26:14,720 So you solve it on that element with the boundary conditions of 454 00:26:14,720 --> 00:26:18,760 the other nodes that are next to you and you keep iterating on 455 00:26:18,760 --> 00:26:20,560 you. And that's how all our numerical 456 00:26:20,560 --> 00:26:24,040 methods work, right? The trouble with these numerical 457 00:26:24,040 --> 00:26:28,320 methods is the trade off become between accuracy and speed, 458 00:26:28,320 --> 00:26:30,720 right? So suppose you solve that 459 00:26:30,720 --> 00:26:37,000 problem, the CFD with whatever, with say 1000 elements, right? 460 00:26:37,440 --> 00:26:42,200 And you get an accuracy which is about 10% error, which may be 461 00:26:42,200 --> 00:26:44,200 fine for you. I said yeah, I like it, right? 462 00:26:44,200 --> 00:26:48,840 And you solve that in an hour. Say I don't like 10% error, I 463 00:26:48,840 --> 00:26:52,760 wanted to be more accurate. It is very easy in our world to 464 00:26:52,760 --> 00:26:57,040 just instead of 1000 elements do 100,000 elements, right? 465 00:26:57,040 --> 00:27:02,840 You do finer meshes and it will be 1% error, right? 466 00:27:03,040 --> 00:27:06,600 But then in instead of 1000 hours to run, it will take you 467 00:27:06,840 --> 00:27:11,000 100,000 hours to run, right? So the trade off of accuracy and 468 00:27:11,000 --> 00:27:15,480 speed in our world of CAE simulation CFD is, is this 469 00:27:15,480 --> 00:27:17,480 problem right, the accuracy versus speed. 470 00:27:18,160 --> 00:27:21,560 And we want both. We want both accuracy and speed. 471 00:27:22,840 --> 00:27:27,600 And then furthermore, the third thing is these things are so 472 00:27:28,000 --> 00:27:29,800 complicated in terms of convergence. 473 00:27:29,800 --> 00:27:33,400 Sometimes you do these crazy things with the meshing, it 474 00:27:33,400 --> 00:27:35,440 doesn't convert. So wow, my God, I didn't 475 00:27:35,720 --> 00:27:38,560 conserve. Oh, it didn't convert because of 476 00:27:38,560 --> 00:27:40,920 this. I should use mosaic machine. 477 00:27:40,920 --> 00:27:44,160 I should use towel machine. So there are these zillion tools 478 00:27:44,160 --> 00:27:49,360 that I have at my disposal and the the CAE analyst is using all 479 00:27:49,360 --> 00:27:52,160 of these things and sometimes it works, sometimes it doesn't. 480 00:27:52,440 --> 00:27:57,840 So it's not that easy to use. Imagine a tool out there that'll 481 00:27:57,840 --> 00:28:02,920 say, hey, me, run this thing for a a external aerodynamics of a 482 00:28:02,920 --> 00:28:06,800 Boeing 777 airplane, right? You just give it in English and 483 00:28:07,280 --> 00:28:12,200 automatically it sets the settings for star CCM plus or or 484 00:28:12,200 --> 00:28:16,040 EXA from DASO or fluent. It just does it like that's the 485 00:28:16,600 --> 00:28:22,240 ultimate Holy Grail. So in our world, the problem is 486 00:28:22,240 --> 00:28:27,040 you have to go be accurate. You have to be fast, it has to 487 00:28:27,040 --> 00:28:30,280 be easy to use and converge all the time. 488 00:28:30,280 --> 00:28:35,200 That is the Holy Grail. So in my role as CTO, I look at 489 00:28:35,200 --> 00:28:39,480 all the solvers, I say how can I get to that current state to 490 00:28:39,480 --> 00:28:42,600 make it more accurate, faster, easy to use and so on, right. 491 00:28:42,880 --> 00:28:47,840 So I have, one of the things is I have a pillar on numerical 492 00:28:47,840 --> 00:28:51,400 methods and we are just with advanced numerical methods, 493 00:28:51,400 --> 00:28:54,400 right? We are doing without using high 494 00:28:54,400 --> 00:28:56,640 performance computing, without using AIML. 495 00:28:56,920 --> 00:29:00,560 I'm trying to make it faster, accurate, for example doing 496 00:29:00,560 --> 00:29:03,920 better meshing, for example using higher order methods, 497 00:29:03,920 --> 00:29:06,200 right? How about using hierarchical 498 00:29:06,200 --> 00:29:08,040 octree? So it's a sequential algorithm, 499 00:29:08,040 --> 00:29:12,080 but just using smart things in the numerical method itself you 500 00:29:12,080 --> 00:29:15,400 make it faster, easy to use, converge all the time and so on. 501 00:29:16,440 --> 00:29:21,280 The second pillar is HPC, right? I mean, again, you work out AW 502 00:29:22,080 --> 00:29:25,040 AWS and you have all those high performance computing, right? 503 00:29:25,040 --> 00:29:29,560 So we have we, we, we take, we, we take an algorithm and we 504 00:29:30,040 --> 00:29:33,600 parallelize it, put it on 100 processors using shared memory 505 00:29:33,600 --> 00:29:37,040 or message passing with distributed sort of data 506 00:29:37,040 --> 00:29:39,960 decomposition, all with GPU. So there are all these different 507 00:29:39,960 --> 00:29:43,520 things basically. But this is what I call brute 508 00:29:43,520 --> 00:29:47,200 force acceleration, right? I have a job that I have decided 509 00:29:47,400 --> 00:29:50,600 that I will use 1,000,000 elements, right? 510 00:29:50,600 --> 00:29:54,600 So because of accuracy I have and it's taking me 1000 hours to 511 00:29:54,600 --> 00:29:57,760 run. If I had 100 processors, the 512 00:29:57,760 --> 00:30:02,440 best I can get is get 100 times, speed up and run it in 10 hours, 513 00:30:02,440 --> 00:30:04,440 right? So within that I use shared 514 00:30:04,440 --> 00:30:08,640 memory message passing GPU's XYZ and now we are looking at 515 00:30:08,640 --> 00:30:10,720 quantum computing also to speed things up, right? 516 00:30:11,000 --> 00:30:14,360 But that's what I call brute force parallelism, right? 517 00:30:15,680 --> 00:30:19,240 The third pillar that we have is AIML, which is sort of your 518 00:30:19,240 --> 00:30:22,520 question. So AIML has been used in a 519 00:30:22,520 --> 00:30:28,000 variety of fields, but we and it has been used in, as you know, 520 00:30:28,000 --> 00:30:30,360 for, for recommendation engines for this and so on. 521 00:30:30,360 --> 00:30:31,680 Hey, which restaurant should I go to? 522 00:30:31,880 --> 00:30:33,480 It's wonderful for those things, right? 523 00:30:33,480 --> 00:30:36,800 Or chat GPD allowing you to write wonderful poetry and text. 524 00:30:37,040 --> 00:30:41,520 But the question that we asked is, can AIML be applied to 525 00:30:41,520 --> 00:30:43,280 numerical method simulation, right? 526 00:30:43,280 --> 00:30:47,000 And that's when I joined the company six years ago, my CEO 527 00:30:47,000 --> 00:30:48,000 said, what do you want to work on? 528 00:30:48,000 --> 00:30:53,080 I said, I want to work on AI. And the early work on AI that we 529 00:30:53,080 --> 00:30:59,760 did was to say, OK, let's take a black box solver like Fluent 530 00:31:00,160 --> 00:31:02,240 from which is a fluid solver, right? 531 00:31:02,560 --> 00:31:06,920 Give it an initial condition, boundary condition and you get 532 00:31:06,920 --> 00:31:10,000 the output. And with this input and output 533 00:31:10,000 --> 00:31:13,720 you train in AI model, right? And you see you have this new 6 534 00:31:13,720 --> 00:31:16,120 stage neural network, right? And you are you have these 535 00:31:16,120 --> 00:31:18,320 weights of the neural networks. You don't know what the weights 536 00:31:18,320 --> 00:31:20,240 are. So you start with some random 537 00:31:20,240 --> 00:31:24,040 weights with some random weights on the neurons, right? 538 00:31:25,040 --> 00:31:28,680 You, you here is an input, here is the output. 539 00:31:28,680 --> 00:31:32,480 So with random weights you will predict an output which will be 540 00:31:32,480 --> 00:31:35,080 completely wrong. There is an error at the output. 541 00:31:35,360 --> 00:31:36,760 You say now that there is an error. 542 00:31:37,040 --> 00:31:40,840 How do I minimize the error? I do back propagation to adjust 543 00:31:40,840 --> 00:31:44,960 the weights of neural networks so that my error is 0 for this 544 00:31:44,960 --> 00:31:49,440 input output combination. Then I give it a second input 545 00:31:49,920 --> 00:31:51,920 with a different boundary condition, different whatever, 546 00:31:52,000 --> 00:31:55,240 and with now the previous set of weights. 547 00:31:56,000 --> 00:31:59,160 I run it, I get an A predicted output. 548 00:31:59,520 --> 00:32:02,640 I have a new output from fluent. Again there is an error. 549 00:32:02,640 --> 00:32:04,360 I said, oh, I need to fix the error. 550 00:32:04,640 --> 00:32:07,960 So I do back propagation to change the weights again. 551 00:32:09,680 --> 00:32:13,960 And then I do the third input with the first two set of 552 00:32:13,960 --> 00:32:17,200 weights and my third input. I keep iterating. 553 00:32:17,400 --> 00:32:23,680 After about 102 hundred cases, I kind of get, I converge on the 554 00:32:23,680 --> 00:32:25,680 set of weights on the neural network, right? 555 00:32:25,680 --> 00:32:28,240 And within that there's all kinds of there's choices, right? 556 00:32:28,400 --> 00:32:30,760 Should I have a six stage network? 557 00:32:30,760 --> 00:32:32,240 Should I have a eight stage network? 558 00:32:32,440 --> 00:32:33,600 How many? What's the depth? 559 00:32:33,600 --> 00:32:36,880 What's the depth, right? And that ties to the parameter 560 00:32:36,880 --> 00:32:39,600 size of your, of your network, right? 561 00:32:39,720 --> 00:32:44,520 But assuming you have done all that, right, that's what SIM AI 562 00:32:44,520 --> 00:32:49,680 does. So SIM AI is a platform which 563 00:32:49,680 --> 00:32:56,880 allows a customer to take their problem their sets of designs. 564 00:32:57,720 --> 00:33:03,760 Use our tool Fluent for fluid dynamics or Ansys Mechanical for 565 00:33:03,760 --> 00:33:06,200 structures or HFSS for electromagnetics. 566 00:33:06,440 --> 00:33:11,800 And you, Mr. Customer, use CMI platform to train the AI models 567 00:33:11,800 --> 00:33:17,080 on your problem and then train it for the 1st 100 designs that 568 00:33:17,080 --> 00:33:21,520 you have and the 101st design instead of taking 100 hours, 569 00:33:21,520 --> 00:33:25,040 we'll be we'll run in a minute. That's the value proposition. 570 00:33:26,680 --> 00:33:31,720 Now the AI is only as good as the data you train it with, 571 00:33:31,800 --> 00:33:34,560 right? So if you train it with this 572 00:33:34,880 --> 00:33:37,080 picture of you have an SUV, right? 573 00:33:37,200 --> 00:33:41,280 You, you train it with this SUV from Toyota, there are 10 574 00:33:41,280 --> 00:33:44,360 different versions of Highlander, the, the, the 575 00:33:44,360 --> 00:33:46,600 forerunner, the this Rav or whatever. 576 00:33:46,600 --> 00:33:51,080 And then also the SUVs from, from Hyundai and the SUVs from 577 00:33:51,280 --> 00:33:53,360 the four. So you are training it with 578 00:33:53,400 --> 00:33:56,760 SUVs, it learns, then you give it an airplane. 579 00:33:57,760 --> 00:34:02,160 I have not seen this before. And AI is only good as the data 580 00:34:02,160 --> 00:34:07,040 it has been trained on. But you may say, oh, therefore 581 00:34:07,040 --> 00:34:10,320 it's not, not not useful. It is actually useful because if 582 00:34:10,320 --> 00:34:13,120 you work for a company like Airbus, right, you're making 583 00:34:13,120 --> 00:34:15,480 airplanes or Boeing, you're making airplanes. 584 00:34:15,800 --> 00:34:19,080 You're not going to go from 1 airplane to tomorrow doing a 585 00:34:19,080 --> 00:34:20,840 submarine, right? So you're actually doing only 586 00:34:20,840 --> 00:34:23,320 airplanes. So it is actually work. 587 00:34:23,400 --> 00:34:27,600 There is value in subtle variations and that's what 588 00:34:27,600 --> 00:34:29,679 designers do, right? There have been thousands of 589 00:34:29,679 --> 00:34:32,920 designs of slightly different airplanes or slightly different 590 00:34:32,920 --> 00:34:35,280 cards and so on. So there is value in CMAI. 591 00:34:38,120 --> 00:34:40,880 But then you ask the question, right, So where is the future? 592 00:34:40,880 --> 00:34:49,199 The future is foundational models for AI where the customer 593 00:34:49,199 --> 00:34:54,600 will not have to train any set of things. 594 00:34:54,600 --> 00:34:56,800 There is no need for a semi platform. 595 00:34:57,840 --> 00:35:03,160 We, Ansys, will take the world of physics, of fluids around us 596 00:35:03,720 --> 00:35:06,640 and we'll train it and that is what we will. 597 00:35:07,000 --> 00:35:13,640 So we will train the AI, just like ChatGPT has trained all the 598 00:35:13,640 --> 00:35:15,320 words in the English language, right? 599 00:35:15,640 --> 00:35:18,760 And has learned how to speak, how to write poetry. 600 00:35:20,320 --> 00:35:24,920 The grand vision of AI with foundational models for physics 601 00:35:24,920 --> 00:35:27,760 is to do that. It is an incredibly hard 602 00:35:27,760 --> 00:35:30,480 problem, but that's what we are working on. 603 00:35:31,720 --> 00:35:34,760 But you, that's interesting because I've often had this 604 00:35:34,760 --> 00:35:40,360 debate on the commercial or the economics of that. 605 00:35:41,120 --> 00:35:48,640 As in, if you are a car company, you probably have your own cars. 606 00:35:48,640 --> 00:35:51,720 Like you said, it's quite incremental and you'll train, 607 00:35:52,320 --> 00:35:56,360 you could train using your own data that is proprietary to you. 608 00:35:57,040 --> 00:36:00,520 And you, you would have assumed that the model you would train 609 00:36:00,520 --> 00:36:03,360 would be as accurate as possible because it's your cars and your, 610 00:36:03,440 --> 00:36:08,520 your, your iterations. Same if you're an aircraft 611 00:36:08,520 --> 00:36:12,120 designer. I guess what you're alluding to 612 00:36:12,120 --> 00:36:18,120 is if your company or another company could run their own 613 00:36:18,120 --> 00:36:21,760 simulations of all of these different things and then train 614 00:36:21,760 --> 00:36:27,840 a massive model, will that model be more accurate than the model 615 00:36:27,840 --> 00:36:30,120 that the car company has trained themselves? 616 00:36:32,320 --> 00:36:34,880 And. Yes, yes, and and and here's 617 00:36:34,880 --> 00:36:37,840 why. I'll go back to the Google 618 00:36:37,840 --> 00:36:40,840 example. Like Google search is so good 619 00:36:42,040 --> 00:36:45,880 because it's a free tool, right? You and I type things on Google 620 00:36:46,720 --> 00:36:49,520 and based on it, they are creating this massive database, 621 00:36:49,520 --> 00:36:51,160 right? This page rank algorithms of 622 00:36:51,160 --> 00:36:53,920 this, tied to that and so on. And based on I'm clicking this, 623 00:36:53,920 --> 00:36:57,800 right, gives you the left list of 100 things, and you click 624 00:36:57,800 --> 00:37:00,040 this. And the more you click, that is 625 00:37:00,040 --> 00:37:04,360 a more important thing, right? And so if Google were to be only 626 00:37:04,360 --> 00:37:09,440 limited to the searches that Prith and Neil only did, that's 627 00:37:09,440 --> 00:37:13,080 the only data they looked at, the search would not be as good. 628 00:37:14,280 --> 00:37:17,760 The reason Google is good is because they're looking at the 629 00:37:17,760 --> 00:37:20,880 10 billion people on the planet banging on our, on our 630 00:37:20,880 --> 00:37:25,040 keyboards, right, for free. They, they, we think it is a 631 00:37:25,040 --> 00:37:27,960 free thing. They're not paying us to give 632 00:37:27,960 --> 00:37:31,880 us, give them the data, right? They're using all our 633 00:37:31,880 --> 00:37:34,360 information to make the search be better. 634 00:37:36,200 --> 00:37:42,520 So in exchange for us getting a free tool, we are giving Google 635 00:37:42,520 --> 00:37:47,800 back the knowledge in our head that after I type who is Neil 636 00:37:47,880 --> 00:37:51,080 Ashton from AWS, right? That somebody in the world is 637 00:37:51,080 --> 00:37:54,000 actually interested in the question of who is Neil Ashton 638 00:37:54,000 --> 00:37:56,560 from AWS? And the other question is who is 639 00:37:56,560 --> 00:38:00,160 Prit Banerjee Francis, right? That is knowledge that is being 640 00:38:00,160 --> 00:38:05,840 captured by Google. So now my knowledge is right. 641 00:38:05,840 --> 00:38:10,280 Suppose I went to train it only on Airbus designs, right? 642 00:38:12,280 --> 00:38:15,440 The think of it as the Google search for only people within 643 00:38:15,440 --> 00:38:21,120 Airbus typing their searches versus letting the searches go 644 00:38:21,120 --> 00:38:23,520 to all engineering companies, right? 645 00:38:23,800 --> 00:38:29,120 To Airbus and Boeing and Pratt, Whitney and G, it will be 646 00:38:30,240 --> 00:38:32,400 clearly richer. Yeah. 647 00:38:33,000 --> 00:38:36,920 That's the value now to make it happen, right? 648 00:38:36,920 --> 00:38:39,400 So, so I'll, I'll anticipate the question, right. 649 00:38:39,560 --> 00:38:42,440 So where do you get the data from, right? 650 00:38:42,440 --> 00:38:45,440 So this is something I'm actually thinking of, right? 651 00:38:45,840 --> 00:38:50,920 So to build these foundational models, I will have to get all 652 00:38:50,920 --> 00:38:55,720 the CAD files from Airbus and all the CAD files from Boeing. 653 00:38:57,000 --> 00:39:00,320 Will Boeing and Airbus be willing to give it to us? 654 00:39:02,040 --> 00:39:04,760 And this is the whole thing about Google, right? 655 00:39:04,760 --> 00:39:06,520 I just give it. The reason I give the Google 656 00:39:06,520 --> 00:39:11,160 example is they created a business model which was free, 657 00:39:11,840 --> 00:39:15,880 perceived free, but in exchange for free they are sucking their 658 00:39:15,880 --> 00:39:19,520 stuff, right? So can I create a model which is 659 00:39:19,520 --> 00:39:27,000 an opt in model where all ANSYS customers would opt to give 660 00:39:27,000 --> 00:39:31,920 their data in an anonymized way two ANSYS to collect the data, 661 00:39:31,920 --> 00:39:35,760 run all those things for example on the AWS cloud, right? 662 00:39:36,040 --> 00:39:41,960 And the cloud is a great way to train all these models because 663 00:39:42,560 --> 00:39:45,400 if it is on Prem, you actually cannot have access to it. 664 00:39:45,400 --> 00:39:49,320 But if you are going on the cloud, if every customer, if all 665 00:39:49,320 --> 00:39:51,800 CAD designs are done on the cloud, right, it is actually 666 00:39:51,800 --> 00:39:54,800 possible. All you need is for Airbus and 667 00:39:54,800 --> 00:39:59,280 Boeing and Ford and GM to say you can train the model. 668 00:39:59,960 --> 00:40:05,200 Just don't attribute it to Ford. Yeah, that's. 669 00:40:06,360 --> 00:40:08,840 So now we are getting into policy to things. 670 00:40:09,040 --> 00:40:11,840 So it is possible. And so if if that were to 671 00:40:11,840 --> 00:40:14,440 happen, it would be more accurate than the Airbus 672 00:40:14,440 --> 00:40:17,560 specific result that is the long answer to that question. 673 00:40:17,560 --> 00:40:19,960 Yeah, no, no. And I think this is actually 674 00:40:19,960 --> 00:40:23,920 something that Max Welling, when I spoke with him, he brought up 675 00:40:23,920 --> 00:40:29,640 which was the incentivizing people to share data, you know, 676 00:40:29,840 --> 00:40:34,280 exactly, you know, that having some mechanism where either they 677 00:40:34,280 --> 00:40:37,760 get paid for it or they get something in return, something 678 00:40:37,760 --> 00:40:40,680 that will allow them to overcome the barrier. 679 00:40:41,000 --> 00:40:44,560 The, the sort of traditional position of this is our data. 680 00:40:44,560 --> 00:40:47,120 I'm not going to, you know, let anybody else use it to the point 681 00:40:47,120 --> 00:40:49,160 where they see a benefit from doing it. 682 00:40:49,440 --> 00:40:54,000 I guess the technology piece is the making it anonymous. 683 00:40:54,080 --> 00:40:56,400 You know, that's probably the challenging bit is to sort of 684 00:40:56,400 --> 00:40:59,360 figure out how to do it. But I guess this is not limited 685 00:40:59,360 --> 00:41:01,640 to simulation. This is a broader question, 686 00:41:01,640 --> 00:41:04,640 isn't it, to and? And for example, people are now 687 00:41:05,160 --> 00:41:11,280 suing Dally right for hey, you, you are generating a image based 688 00:41:11,280 --> 00:41:14,440 on my I'm an artist. I'm I'm whatever, right? 689 00:41:15,040 --> 00:41:16,920 I'm Preet Banerjee. I've written a beautiful 690 00:41:17,120 --> 00:41:19,640 picture, right? And you took that picture into 691 00:41:19,640 --> 00:41:22,720 Delhi and now you generate a new picture, right, based on his 692 00:41:22,720 --> 00:41:26,040 Indian knowledge. That's not fair if Preet 693 00:41:26,040 --> 00:41:31,320 Banerjee where to say I will give 10 of my pictures to Delhi 694 00:41:32,160 --> 00:41:37,160 or 10 of my poems to open AI right and I get one cent for 695 00:41:37,160 --> 00:41:40,640 everything that I give every token I give to contribute to 696 00:41:40,640 --> 00:41:44,680 this thing, I get one cent. Hey, I am I incentivize in that 697 00:41:44,680 --> 00:41:47,880 case, I will not sue Delhi. So my thing is, I think the 698 00:41:47,880 --> 00:41:52,720 whole world of gene AI, all the governance mechanism, so on, is 699 00:41:52,800 --> 00:41:56,080 that people getting sued because they feel like their 700 00:41:56,080 --> 00:41:58,680 intellectual property is not getting recognized, right? 701 00:41:59,480 --> 00:42:02,680 Why do people have patterns? Well, they have patterns so that 702 00:42:02,680 --> 00:42:04,880 they can have royalties based on the patterns, right? 703 00:42:05,320 --> 00:42:08,120 That royalties based on the pattern, which is this big thing 704 00:42:08,720 --> 00:42:11,640 in the world of AI. You have to figure it out how to 705 00:42:11,640 --> 00:42:15,920 take that big thing into small, small sunk chunks and to figure 706 00:42:15,920 --> 00:42:18,960 a royalty of 1 cent per pixel, right? 707 00:42:19,680 --> 00:42:21,680 Literally. I mean, the world will actually 708 00:42:21,680 --> 00:42:25,040 go in that area. I I think the world will really 709 00:42:25,040 --> 00:42:29,360 figure out it's governance and fairness so that everybody wins. 710 00:42:30,000 --> 00:42:32,920 Yeah. No, no, I, I think that's a 711 00:42:32,960 --> 00:42:36,960 really good analogy. I, I, I and I agree with you. 712 00:42:36,960 --> 00:42:40,200 I think if the data maybe this leads to the other one because 713 00:42:40,200 --> 00:42:46,960 often the question is around data-driven versus physics 714 00:42:46,960 --> 00:42:51,280 driven with the logic being that you know, we operate in a 715 00:42:51,280 --> 00:42:54,600 scientific world, we should include physics in the models. 716 00:42:55,560 --> 00:42:58,080 But often the argument is we need to include physics because 717 00:42:58,080 --> 00:43:02,680 we don't have enough data. And I have followed the progress 718 00:43:02,680 --> 00:43:04,840 of you know physics informed etcetera. 719 00:43:04,840 --> 00:43:10,200 But what struck me is to, to my knowledge anyway, most of the 720 00:43:10,280 --> 00:43:14,400 successful examples in the public domain have been with 721 00:43:14,400 --> 00:43:20,680 data-driven approaches typically and not so much from the 722 00:43:20,760 --> 00:43:25,320 theoretically better, but often practically not as convenient. 723 00:43:25,600 --> 00:43:28,720 So do you think that is just because it's harder and it will 724 00:43:28,720 --> 00:43:32,720 take more time to sort of develop the more physics 725 00:43:32,720 --> 00:43:36,520 informed physics inspired, you know, how much do you think that 726 00:43:36,520 --> 00:43:42,600 is a needed science step to really overcome the data 727 00:43:42,600 --> 00:43:45,920 challenge and the generalization challenge when it comes to AI 728 00:43:45,920 --> 00:43:49,480 for, you know, computer aided engineering? 729 00:43:50,800 --> 00:43:53,120 That is a great question actually. 730 00:43:53,120 --> 00:43:55,920 And the answer is there is not enough research that has 731 00:43:55,920 --> 00:43:57,920 happened. There's more research that has 732 00:43:57,920 --> 00:43:59,080 happened in the data world, right? 733 00:44:00,000 --> 00:44:02,760 Purely data-driven method is more general. 734 00:44:02,920 --> 00:44:05,560 That's the advantage, right? And and you don't need anything, 735 00:44:05,560 --> 00:44:07,960 just do do the data and you do your stuff, right. 736 00:44:08,640 --> 00:44:12,360 But it requires enormous amount of data, right to do the right 737 00:44:12,360 --> 00:44:16,440 level of accuracy for the models where physics team firm gets you 738 00:44:16,480 --> 00:44:20,240 is so so I mean, I just I know you know this, but to your 739 00:44:20,360 --> 00:44:22,520 readers, I will give you a very simple explanation, right? 740 00:44:22,960 --> 00:44:26,400 So suppose you are trying to look at fluids data, right? 741 00:44:26,760 --> 00:44:30,240 And you are trying to put it into an AI model for fluids, 742 00:44:30,240 --> 00:44:31,840 right? You will take the the fluids 743 00:44:31,840 --> 00:44:35,160 data here is this velocity, pressure, etcetera, temperature 744 00:44:35,160 --> 00:44:36,560 and so on. And this is the distribution, 745 00:44:36,560 --> 00:44:39,200 right? And you think the whole thing is 746 00:44:39,200 --> 00:44:42,720 is random. It is not because the fluids 747 00:44:43,160 --> 00:44:46,160 physics says there is Navier Stokes equation, there is energy 748 00:44:46,160 --> 00:44:49,160 conservation, all the stuff that you know from a physics point of 749 00:44:49,160 --> 00:44:52,800 view. So the data will not be 750 00:44:52,800 --> 00:44:57,480 completely uncorrelated. The data is actually going to be 751 00:44:57,480 --> 00:45:01,880 this thing with this turbulence will be constrained to only this 752 00:45:01,880 --> 00:45:04,520 set of things, right? So if the, if there are three 753 00:45:04,520 --> 00:45:06,440 things here you are, you are measuring, right? 754 00:45:06,880 --> 00:45:10,400 If it is, if you know these two points, you can deduce the Third 755 00:45:10,400 --> 00:45:11,880 Point. It is actually not an 756 00:45:11,880 --> 00:45:15,880 independent variable, right? That's the, the, the, the thing 757 00:45:15,880 --> 00:45:18,560 about statistics, right? You think not all the things are 758 00:45:18,560 --> 00:45:22,160 going to be independent. So the pure data-driven approach 759 00:45:22,200 --> 00:45:24,320 assumes everything is independent and it's not. 760 00:45:24,920 --> 00:45:28,640 So if you can insert the knowledge of the physics, you 761 00:45:28,640 --> 00:45:30,920 can constraint. You say you don't have to search 762 00:45:30,920 --> 00:45:35,600 for millions of data points, you can do it with only 1000 data 763 00:45:35,600 --> 00:45:37,560 points. That's the power of physics 764 00:45:37,560 --> 00:45:39,280 informed, right? And the work was, as you know, 765 00:45:39,280 --> 00:45:42,920 done by John, I mean Karnatakis, George and I say at Brown 766 00:45:42,920 --> 00:45:46,000 University, and we did a lot of work at AT and says we've done a 767 00:45:46,000 --> 00:45:48,480 lot of work at NVIDIA on these things. 768 00:45:48,760 --> 00:45:51,720 The trouble is when we started doing the physics informed to 769 00:45:51,720 --> 00:45:56,600 incorporate the physics, the competition needed in this, we 770 00:45:56,600 --> 00:45:58,920 have not quite figured that out, right? 771 00:45:58,920 --> 00:46:01,720 And then then people went towards graph neural networks 772 00:46:01,920 --> 00:46:04,280 and then animated the work on F&O. 773 00:46:04,280 --> 00:46:08,640 So I saw this this lot of research being done and 774 00:46:08,640 --> 00:46:11,560 somewhere in there in the Holy Grail is foundational models. 775 00:46:12,200 --> 00:46:17,360 And once that thing is invented is like the Einstein theory of 776 00:46:17,360 --> 00:46:21,520 relativity, the the universal thing, something like this will 777 00:46:21,520 --> 00:46:26,200 happen where we'll merge the areas of numerical methods and 778 00:46:26,200 --> 00:46:29,880 AI. And that I have told my board is 779 00:46:29,880 --> 00:46:34,880 when the whole market for ANSYS will completely collapse because 780 00:46:34,880 --> 00:46:38,400 we have the last 50 years we have worked on, on the fact that 781 00:46:38,400 --> 00:46:40,280 it's all numerical methods and so on, right? 782 00:46:40,960 --> 00:46:43,280 Numerical methods will no longer be needed. 783 00:46:43,400 --> 00:46:48,120 It will all be done with AI, with the accuracy and the speed 784 00:46:48,320 --> 00:46:50,760 much, much better than than numerical methods. 785 00:46:51,080 --> 00:46:53,080 But we're not there yet. That's where the research is 786 00:46:53,080 --> 00:46:56,280 needed. And that actually brings me on 787 00:46:56,280 --> 00:47:01,480 to the point then of how, how can we enable that research to 788 00:47:01,480 --> 00:47:04,080 happen. As you said yourself, a very 789 00:47:04,080 --> 00:47:09,080 large enterprise will struggle to dedicate resources to 790 00:47:09,360 --> 00:47:12,600 fundamental problems because of the pressures of headcount, of 791 00:47:12,600 --> 00:47:17,720 incremental product improvement. A start up can do that, but 792 00:47:17,720 --> 00:47:20,640 they're not, they have pressure from their VCs to actually 793 00:47:20,640 --> 00:47:23,960 deliver something within a, you know, relatively small amount of 794 00:47:23,960 --> 00:47:26,960 time usually. So it falls down to academia. 795 00:47:27,880 --> 00:47:34,160 But if, let's say foundational models is, as you rightly say, 796 00:47:34,480 --> 00:47:37,400 could be a, you know, Eureka moment, you know, a massive 797 00:47:37,400 --> 00:47:41,240 moment for the field. The bit that I've noticed is 798 00:47:41,240 --> 00:47:45,640 data, you know, you could incentivize people to give you 799 00:47:45,640 --> 00:47:48,960 data through, you know, and CIS and, and mechanisms. 800 00:47:49,480 --> 00:47:53,800 But I wonder, therefore, what's your opinion of the open source 801 00:47:53,800 --> 00:47:56,680 versus closed source? You know, how much should we be 802 00:47:56,680 --> 00:47:59,160 trying to create some open source data that's to help the 803 00:47:59,160 --> 00:48:02,720 academia, but not do too much that you give all your IP away 804 00:48:02,720 --> 00:48:04,760 and you know, you, you, you lose an advantage. 805 00:48:04,760 --> 00:48:08,600 So where does that balance? That is a great point. 806 00:48:08,600 --> 00:48:14,760 In fact, let me tell you, the reason that the AI world has 807 00:48:17,040 --> 00:48:20,840 worked so fast is because of open source, right? 808 00:48:21,680 --> 00:48:25,320 You have things like Tensorflow and Pytor, these are absolutely 809 00:48:25,320 --> 00:48:29,600 open source ways of doing neural networks, right? 810 00:48:29,600 --> 00:48:32,840 I mean, you, they could have like Google and so on could have 811 00:48:32,840 --> 00:48:35,360 kept all of those and Facebook could have kept it completely 812 00:48:35,360 --> 00:48:38,360 closed, right? And then the world wouldn't have 813 00:48:38,520 --> 00:48:39,840 done all this kind of stuff, right? 814 00:48:40,960 --> 00:48:44,880 NVIDIA opened up CUDA, right? So CUDA became sort of not, and 815 00:48:44,880 --> 00:48:47,240 the code is not open. So, but they have open 816 00:48:47,280 --> 00:48:49,720 framework, right? So the combination of open 817 00:48:49,720 --> 00:48:54,760 source things like CUDA, open source things like Bytorch and 818 00:48:54,760 --> 00:49:01,160 Tensorflow, etcetera, and open source models for the data like 819 00:49:01,160 --> 00:49:03,960 Imagenet and all those things that are out there, right? 820 00:49:04,520 --> 00:49:09,080 It has accelerated the pace of innovation in the world of AI, 821 00:49:09,680 --> 00:49:13,520 unlike other fields like in the world of numerical methods, we 822 00:49:13,520 --> 00:49:15,560 are, we know something doesn't know something. 823 00:49:15,560 --> 00:49:17,840 He seems so and we don't sort of share stuff, right. 824 00:49:17,960 --> 00:49:22,040 So there's a paper they'll come from CMU or Stanford, some 825 00:49:22,040 --> 00:49:26,520 wonderful people and we, we say, ah, but they are working on, on 826 00:49:26,560 --> 00:49:29,800 trivial fraud to eye problems. They cannot work on ANSYS, but 827 00:49:29,800 --> 00:49:33,520 ANSYS will not give those tough problems that did not happen in 828 00:49:33,520 --> 00:49:36,480 the AI world, that did not happen in the map reduce world, 829 00:49:36,720 --> 00:49:40,440 in the map reduce world, right? The map reduce thing was 830 00:49:40,840 --> 00:49:44,240 actually openly given away, right? 831 00:49:44,240 --> 00:49:48,200 Open source by both Google and and Yahoo. 832 00:49:48,640 --> 00:49:53,320 Now why did they do that? That map reduce framework is a 833 00:49:53,320 --> 00:49:57,880 framework that Google needs to improve their searches, right? 834 00:49:57,880 --> 00:50:02,280 So it was a brilliant business move for them to open source map 835 00:50:02,280 --> 00:50:05,640 reduce to the world, right where the smartest graduate students 836 00:50:05,640 --> 00:50:10,880 at MIT and Oxford and and CMU, they all work to improve map 837 00:50:10,880 --> 00:50:13,560 reduce and it's open source. So all the innovations that 838 00:50:13,560 --> 00:50:17,640 coming from the from the open source world academic world, 839 00:50:18,080 --> 00:50:21,160 Google could put in and make the search even better, right? 840 00:50:21,760 --> 00:50:25,040 They did not say here is a search algorithm that we open 841 00:50:25,040 --> 00:50:29,000 source that they took a core part of their search algorithm 842 00:50:29,560 --> 00:50:32,480 which they are making money off, right, with ads. 843 00:50:33,800 --> 00:50:36,600 So I thought that was an absolute brilliant strategy. 844 00:50:36,960 --> 00:50:39,760 Linux is another brilliant strategy, right, for for 845 00:50:39,760 --> 00:50:41,520 advancing operating systems, right? 846 00:50:41,640 --> 00:50:44,440 Which is so we have to actually learn. 847 00:50:44,440 --> 00:50:50,400 So the in our world of CAE simulation, right, there is 848 00:50:50,400 --> 00:50:52,960 obviously one code called Openform. 849 00:50:52,960 --> 00:50:55,440 Since you know, fluids, you know Openform, right? 850 00:50:55,840 --> 00:51:00,560 So it's open source, actually Openform plus Open AI. 851 00:51:00,680 --> 00:51:04,480 I mean sort of is sort of where I think things will happen. 852 00:51:04,480 --> 00:51:08,920 But for that you also need that data for the CAD models, right? 853 00:51:09,160 --> 00:51:13,200 And so just like Imagenet has created this thing for 2D and 3D 854 00:51:13,200 --> 00:51:19,880 images, we need in this area some work on 3D geometries of 855 00:51:19,880 --> 00:51:23,440 all kinds of things, on gears and this and and propellers and 856 00:51:23,440 --> 00:51:25,960 airplanes and so on. If you can do that, I think that 857 00:51:25,960 --> 00:51:27,680 will advance the state-of-the-art. 858 00:51:28,880 --> 00:51:31,920 Yeah. And we, you know, we, we 859 00:51:31,960 --> 00:51:35,280 ourselves published a couple of data that's these Driver ML and 860 00:51:35,560 --> 00:51:37,560 Ahmed ML. That I I am aware so. 861 00:51:38,800 --> 00:51:41,920 The which have helped a little bit, but I. 862 00:51:43,000 --> 00:51:45,560 Not at the level of image net, not at the level of of. 863 00:51:46,400 --> 00:51:47,760 Exactly. Exactly. 864 00:51:48,480 --> 00:51:52,000 But the the other bit that I always and maybe I just need to 865 00:51:52,000 --> 00:51:58,240 get my brain around this, which is if you're training like a 866 00:51:58,240 --> 00:52:02,880 large language model, the text and the data that you scrape up 867 00:52:02,880 --> 00:52:07,320 the Internet is in some ways it's sort of the ground truth. 868 00:52:07,400 --> 00:52:11,880 Yes, someone has wrote it, but it isn't a simulation, right? 869 00:52:12,200 --> 00:52:15,320 It is someone writing it. If you transfer now to, let's 870 00:52:15,320 --> 00:52:20,760 say CFD, I could run simulations of thousands of cars and planes, 871 00:52:21,400 --> 00:52:27,000 but the model will only, well, this is my question. 872 00:52:27,720 --> 00:52:30,920 Using, let's say, just a standard approach, it's only 873 00:52:30,920 --> 00:52:36,320 going to learn the equivalent simulation settings if it's a 874 00:52:36,320 --> 00:52:40,200 RANS approach or an LES approach, or a mesh that is 875 00:52:40,200 --> 00:52:43,640 coarse or fine. If I do all my simulations with 876 00:52:43,640 --> 00:52:46,520 a RANS, the model's going to learn. 877 00:52:48,480 --> 00:52:54,760 So does that mean you have a foundational model of this 878 00:52:54,760 --> 00:52:57,840 simulation approach and then you have another foundational model? 879 00:52:57,840 --> 00:53:01,120 Or is there some way? Or or you have to curate your 880 00:53:01,120 --> 00:53:06,080 data so that you take 10% of the data from RANS, 10% data from 881 00:53:06,120 --> 00:53:09,680 LES, 10% data from DNS. That would be the real 882 00:53:09,680 --> 00:53:12,240 foundational model. So getting back to your issue, 883 00:53:12,240 --> 00:53:14,680 right? And so the reason, so I'm glad 884 00:53:14,680 --> 00:53:18,360 you asked this question because with large language models for 885 00:53:18,360 --> 00:53:20,840 words as tokens, you are absolutely right. 886 00:53:20,840 --> 00:53:25,200 They have scraped all the words from all the books that people 887 00:53:25,200 --> 00:53:28,000 have written, right? Of course, they are not giving 888 00:53:28,000 --> 00:53:29,960 the royalty back to the people who have written those things, 889 00:53:29,960 --> 00:53:35,120 right, Pop chap, right. They have not generated those. 890 00:53:35,360 --> 00:53:40,440 Suppose a next version of chat GPD is take chat GPD to generate 891 00:53:40,720 --> 00:53:43,360 all those tests, right? And then you feel it, that would 892 00:53:43,360 --> 00:53:44,720 be what your problem would be, right? 893 00:53:45,360 --> 00:53:49,320 So and there and, and Delhi has taken the same approach of 894 00:53:49,320 --> 00:53:52,920 taking all the images on the Internet and use those images to 895 00:53:52,920 --> 00:53:56,120 program Delhi, right? And Sora has done the same thing 896 00:53:56,120 --> 00:54:00,680 for for videos, right? So in our world to do the 897 00:54:00,680 --> 00:54:03,440 foundational models, it has to be the 3D fields. 898 00:54:03,680 --> 00:54:07,920 Now you can actually go and measure 3D fields, right? 899 00:54:08,200 --> 00:54:13,080 You you take a car right Google way more those cars have videos 900 00:54:13,080 --> 00:54:15,800 right. So imagine you you sensorize 901 00:54:15,800 --> 00:54:21,680 your car to measure the fluid flow at every small microsecond 902 00:54:21,680 --> 00:54:24,400 or micro whatever millimeter of your car right. 903 00:54:24,760 --> 00:54:28,400 That is an actual measurement. I mean this is the air how the 904 00:54:28,400 --> 00:54:31,520 airflow actually happened on the on the car right that you have 905 00:54:31,520 --> 00:54:34,360 to take thousands of cars millions of cars and so on. 906 00:54:34,360 --> 00:54:37,840 It's just ridiculous right. So what I am saying what I have 907 00:54:37,840 --> 00:54:41,920 told my board is we ANSYS will create the synthetic data 908 00:54:43,080 --> 00:54:46,480 through simulation right of all the fluids model. 909 00:54:46,480 --> 00:54:50,800 But you are absolutely right, we ANSYS fluent is ran simulation. 910 00:54:50,800 --> 00:54:52,760 So it will not be generating the LES. 911 00:54:52,920 --> 00:54:57,120 So we will also have to do the LES and the DNS and for external 912 00:54:57,120 --> 00:55:00,880 fluid and for for cars and trucks and so on. 913 00:55:00,880 --> 00:55:04,880 So only if you do all of that will be a true foundational 914 00:55:04,880 --> 00:55:12,040 model, which is why if ChatGPT took six months and 1.8 trillion 915 00:55:12,040 --> 00:55:16,360 parameters, in our world, it is probably a billion trillion 916 00:55:16,360 --> 00:55:17,920 parameter. I don't even know what the size 917 00:55:17,920 --> 00:55:20,440 of the model is, but it is possible. 918 00:55:20,480 --> 00:55:23,840 I absolutely it is possible and we'll eventually get there. 919 00:55:24,560 --> 00:55:27,960 Yeah, it is. It is so fascinating, isn't it? 920 00:55:27,960 --> 00:55:31,000 Because that it would be a transformational change. 921 00:55:31,040 --> 00:55:34,160 Like you say, you're right to tell your board that it's the 922 00:55:34,160 --> 00:55:38,280 honest truth that the tradition of, you know, running your own 923 00:55:38,280 --> 00:55:42,400 simulations, if the model was accurate enough and you know, 924 00:55:42,400 --> 00:55:47,160 there's a big if I guess on that, it would certainly become 925 00:55:47,160 --> 00:55:51,440 a very widely, it would disrupt the market, that's for sure in 926 00:55:51,440 --> 00:55:53,280 a, in a big, big way. Let me. 927 00:55:54,000 --> 00:55:56,840 Let me make this statement to your readers. 928 00:55:56,880 --> 00:56:01,280 I know you know this, but the whole world of physics, right? 929 00:56:02,480 --> 00:56:08,080 When Newton observed an apple, he dropped it and it fell down 930 00:56:08,080 --> 00:56:09,320 and he drove something else, right? 931 00:56:10,040 --> 00:56:13,680 Just by a bunch of observations in the real world. 932 00:56:14,080 --> 00:56:20,080 That data that he fed into his engine right determined the law 933 00:56:20,080 --> 00:56:26,040 which is force equals mass times acceleration which is a 934 00:56:26,120 --> 00:56:33,880 differential equation right. He deduced the law of gravity by 935 00:56:33,920 --> 00:56:40,360 observations. It is therefore possible to 936 00:56:40,360 --> 00:56:44,800 observe the world around us and actually we have got work going 937 00:56:44,800 --> 00:56:48,800 on with Google DeepMind right. DeepMind is using the ANSYS 938 00:56:48,800 --> 00:56:53,680 tools to observe the physics and learn the physics. 939 00:56:53,960 --> 00:56:59,840 So that imagine you're trying to balance a long pen on your head, 940 00:56:59,840 --> 00:57:00,800 right? Yeah, I'm sure when you're a 941 00:57:00,800 --> 00:57:02,320 kid, you did that and you're balancing, right. 942 00:57:02,600 --> 00:57:05,480 When you ball, the thing goes on the other side. 943 00:57:05,480 --> 00:57:08,840 You you move your hand, right. So they have actually taken that 944 00:57:08,840 --> 00:57:13,680 as an example to use ANSYS mechanical to model the world of 945 00:57:13,920 --> 00:57:17,560 structures, right? And just by say and thereby 946 00:57:17,560 --> 00:57:24,080 train the Google, Google Mind has trained the robot arm to do 947 00:57:24,080 --> 00:57:26,800 the balancing of this by learning the physics. 948 00:57:27,520 --> 00:57:30,760 So it is possible and that is how I think foundational models 949 00:57:30,760 --> 00:57:37,160 will work because Isaac Newston generated the physics model of 950 00:57:37,160 --> 00:57:40,360 force equals minus some acceleration by looking at the 951 00:57:40,360 --> 00:57:43,400 data. By observing the data, AI is 952 00:57:43,400 --> 00:57:48,160 going to observe the physics around us and train the physics 953 00:57:48,160 --> 00:57:51,280 models. Every one of those equations can 954 00:57:51,280 --> 00:57:54,000 be deduced. Navier Stokes equations can be 955 00:57:54,000 --> 00:57:58,320 reverse engineered by AI. And that would be, it's 956 00:57:58,320 --> 00:58:00,240 interesting to bring up the robotics angle to that, because 957 00:58:00,240 --> 00:58:03,560 I guess this is the, the, the bigger picture, isn't it the 958 00:58:03,600 --> 00:58:06,520 sort of future of manufacturing, the future of robotics? 959 00:58:06,520 --> 00:58:10,200 You know, CAE, you don't just simulate for the sake of it, do 960 00:58:10,200 --> 00:58:12,000 you? You simulate it to do something. 961 00:58:13,040 --> 00:58:16,160 And so you're right, there is a much, you know, broader, broader 962 00:58:16,160 --> 00:58:22,400 picture around. However, one thing a sort of 963 00:58:22,400 --> 00:58:24,960 counter example to that, I guess, which goes back to your, 964 00:58:25,720 --> 00:58:27,920 well, not a counter example, but another way of thinking about 965 00:58:27,920 --> 00:58:29,960 it. You said right at the beginning, 966 00:58:30,480 --> 00:58:34,960 accuracy and speed or cost, It's true that all engineering 967 00:58:34,960 --> 00:58:39,240 companies are so, you know, focused on that, aren't they? 968 00:58:39,360 --> 00:58:43,480 Accuracy, speed, cost. So if your traditional 969 00:58:43,480 --> 00:58:47,080 simulation could be fast enough and cheap enough, you don't 970 00:58:47,080 --> 00:58:50,800 necessarily need AI, do you? That could be a normal approach. 971 00:58:51,560 --> 00:58:56,000 So I was just wondering the quantum piece, everybody brings 972 00:58:56,000 --> 00:59:00,480 this up and I would love to get your perspective on how 973 00:59:01,040 --> 00:59:05,040 realistic is it and of what parts of you think that CAE 974 00:59:05,040 --> 00:59:06,920 could speed up the quantum side of things? 975 00:59:07,040 --> 00:59:10,080 Could could quantum speed up the CAE side of things? 976 00:59:10,440 --> 00:59:16,160 Or do you see it still as being too niche and yeah, not? 977 00:59:16,600 --> 00:59:19,920 No, no, it's a great question. And actually in my city office, 978 00:59:19,920 --> 00:59:23,600 so the the first thing I did in my city office was to work on AI 979 00:59:24,240 --> 00:59:28,400 and now I've got now that that's AI thing is sort of not solved, 980 00:59:28,400 --> 00:59:30,520 but at least we have some products out in this area, 981 00:59:30,520 --> 00:59:33,120 right? We have started working on 982 00:59:33,120 --> 00:59:37,680 quantum for exactly that reason. And the the beauty of quantum is 983 00:59:37,880 --> 00:59:40,760 it is a potential for exponential speed UPS, right? 984 00:59:41,560 --> 00:59:45,800 Because if you have N cubits, your speed, your, your, your 985 00:59:45,800 --> 00:59:48,400 runtime, your, your speed up is 2 to the power N, right? 986 00:59:48,840 --> 00:59:52,680 And so as you typically quantum computing algorithms have been 987 00:59:52,680 --> 00:59:56,840 used on problems that are sort of NP complete to begin with, 988 00:59:56,840 --> 01:00:00,440 right, exponential problems like materials discovery, your 989 01:00:00,440 --> 01:00:02,560 optimization and travelling salesman and so on. 990 01:00:03,200 --> 01:00:06,520 And you would think that in our world, right our our problems 991 01:00:06,520 --> 01:00:10,880 are polynomial is n ^3. Except that N is huge, N is a 992 01:00:10,880 --> 01:00:13,080 million million cubed is a large number, right? 993 01:00:13,360 --> 01:00:19,200 So if you can throw a 2 to the power P kind of exponential 994 01:00:19,200 --> 01:00:23,600 capacity to solve a polynomial problem order N cube where N is 995 01:00:23,600 --> 01:00:26,640 very large, yes there is benefits. 996 01:00:26,920 --> 01:00:31,920 So we are looking at ways these algorithms called the HHL 997 01:00:31,920 --> 01:00:36,440 algorithm that you may have heard of that can take a because 998 01:00:36,440 --> 01:00:40,640 in all our things like we ultimately we will take our PDS 999 01:00:40,640 --> 01:00:46,560 and and make it into some AX equals B, some matrix vector. 1000 01:00:46,640 --> 01:00:48,960 So you are trying to solve some linear system equations. 1001 01:00:49,520 --> 01:00:54,120 Turns out that quantum computers can solve linear systems of 1002 01:00:54,120 --> 01:00:56,240 equations like with exponential speed up. 1003 01:00:56,240 --> 01:00:59,680 So we are looking into those kind of methods at ANSYS. 1004 01:01:00,280 --> 01:01:05,280 It is not going to be next year. It'll not be two years, but 1005 01:01:05,280 --> 01:01:08,160 definitely within 10 years we'll see quantum computing 1006 01:01:08,160 --> 01:01:12,600 accelerating CE simulation. And that'll be interesting. 1007 01:01:12,600 --> 01:01:18,480 It's almost like you never clear what technology will be the one 1008 01:01:18,800 --> 01:01:21,040 that transforms. You know, the I know it's not a 1009 01:01:21,040 --> 01:01:23,960 great analogy, but you don't remember 3D glasses. 1010 01:01:24,440 --> 01:01:27,040 The televisions came out and everyone thought that would be 1011 01:01:27,040 --> 01:01:28,440 it. We'll all wear 3D glasses. 1012 01:01:28,880 --> 01:01:33,800 And at least to my knowledge, it sort of died away because in 1013 01:01:33,800 --> 01:01:35,960 reality, nobody wants to put those on. 1014 01:01:36,280 --> 01:01:39,440 Now maybe there's a future warm, you know, with the Apple vision, 1015 01:01:39,440 --> 01:01:41,640 etcetera. But it's amazing how resilient 1016 01:01:42,000 --> 01:01:45,760 we have been to watching a normal TV, even with all the 1017 01:01:45,760 --> 01:01:50,680 sort of technology changes. So it does often make me wonder, 1018 01:01:51,960 --> 01:01:55,200 is it ML, is it quantum? You know will one of them. 1019 01:01:55,800 --> 01:01:59,040 And actually this quantum ML people are now working on 1020 01:01:59,080 --> 01:02:02,040 quantum machine learning. So, so, so it is this 1021 01:02:02,040 --> 01:02:05,360 combination of things that means to your broader question, what I 1022 01:02:05,400 --> 01:02:10,840 I would say is Nansys is now a company of about 2 1/2 billion 1023 01:02:10,840 --> 01:02:14,480 dollars, right? And our software originated from 1024 01:02:14,480 --> 01:02:18,240 CE simulation written in Fortran in 1970, right? 1025 01:02:18,880 --> 01:02:22,080 But then as newer technologies like HPC came in, we took that 1026 01:02:22,080 --> 01:02:24,760 Fortran code and said, OK, let's put it on a shared memory and 1027 01:02:24,840 --> 01:02:26,600 with this directive, paralyze it. 1028 01:02:27,040 --> 01:02:30,120 And then the GPUs came in, well, with this directive, make it run 1029 01:02:30,120 --> 01:02:32,720 on CUDA and so on. So it has been always 1030 01:02:33,320 --> 01:02:37,120 retrofitting a piece of thing. And somehow we have OK, now 1031 01:02:37,120 --> 01:02:41,120 let's work with this. And now the AIML came in and 1032 01:02:41,120 --> 01:02:42,520 let's do it with with Tensorflow. 1033 01:02:42,520 --> 01:02:44,840 And now let this cloud came in. Let's try to make it on the 1034 01:02:44,840 --> 01:02:47,520 cloud, right? What if, and this is sort of a 1035 01:02:47,520 --> 01:02:50,000 thought experiment. I, I asked of my technology, I 1036 01:02:50,000 --> 01:02:55,400 said, what if you knew that you have access to quantum, to AIML, 1037 01:02:55,400 --> 01:03:00,640 to cloud, to GPUs, all of those technologies and you to start 1038 01:03:00,640 --> 01:03:03,840 writing ANSYS mechanical, How would you write it? 1039 01:03:04,200 --> 01:03:07,880 You would definitely and you have languages like Julia and 1040 01:03:07,880 --> 01:03:11,440 Python And so on, right? Would you write it in Fortran 1041 01:03:11,560 --> 01:03:13,280 with the linear is not at all right. 1042 01:03:14,880 --> 01:03:19,480 And this is sort of the advantage that startups have. 1043 01:03:19,520 --> 01:03:23,680 A startup has no legacy. This is what how I try to 1044 01:03:23,680 --> 01:03:26,240 motivate people. I said if you are a startup, you 1045 01:03:26,240 --> 01:03:31,560 have the latest widgets, right With with ARVRIOTI don't know 1046 01:03:31,560 --> 01:03:33,520 what it is right. Put all of that in the blender 1047 01:03:33,520 --> 01:03:37,440 and out will come something that a large company against this has 1048 01:03:37,440 --> 01:03:39,240 not had the luxury of doing right. 1049 01:03:39,840 --> 01:03:41,680 That's the power of the Horizon 3. 1050 01:03:42,840 --> 01:03:46,040 And that that's a great sort of circle back to our original 1051 01:03:46,040 --> 01:03:50,360 point that essentially I think what you're saying is you want 1052 01:03:50,440 --> 01:03:53,880 the startups to take that challenge to say, I want to 1053 01:03:53,880 --> 01:03:58,680 start from scratch. Prove to me and Sis that a new 1054 01:03:58,680 --> 01:04:04,240 written code using the latest technology right now could be, 1055 01:04:04,360 --> 01:04:07,840 you know, exponentially better than retrofitting another code. 1056 01:04:08,280 --> 01:04:10,720 And that is what would make a start up attractive to you. 1057 01:04:10,720 --> 01:04:12,160 That's what would change the market. 1058 01:04:12,160 --> 01:04:14,600 That's what would do it. And it's only really a start up. 1059 01:04:14,640 --> 01:04:17,000 And the business model is the following, right? 1060 01:04:17,920 --> 01:04:21,800 Often times I am asked, Aaron says, hey, you are doing this 1061 01:04:21,800 --> 01:04:23,640 right with AIML, why are you doing this? 1062 01:04:23,680 --> 01:04:25,200 Is you'll kill your cash cow, right? 1063 01:04:25,440 --> 01:04:28,880 Because if with AI you know things run 100 times faster, 1064 01:04:28,880 --> 01:04:32,360 it's not good for the ANSYS mechanical business or fluid 1065 01:04:32,360 --> 01:04:36,640 business, right? But the response is if I don't 1066 01:04:36,640 --> 01:04:40,360 do it myself, a startup would do it and destroy me anyway. 1067 01:04:40,480 --> 01:04:42,400 So I might as well do it myself, right. 1068 01:04:42,520 --> 01:04:46,120 So it is like the classic case of Kodak. 1069 01:04:47,480 --> 01:04:51,040 They actually knew of digital printing. 1070 01:04:52,120 --> 01:04:53,480 It's not like those guys are stupid. 1071 01:04:53,480 --> 01:04:54,560 They actually know digital thing. 1072 01:04:54,560 --> 01:04:58,160 But the cash cow from analog was so good that they did not want 1073 01:04:58,160 --> 01:05:01,960 to disrupt it, right? Motorola new about digital 1074 01:05:01,960 --> 01:05:06,200 phones, but they, they, they were afraid of it, right? 1075 01:05:06,200 --> 01:05:09,120 So it's always the innovator's dilemma, right? 1076 01:05:09,240 --> 01:05:10,720 I have a cash cow business, right? 1077 01:05:11,040 --> 01:05:14,240 Should I do it? And so companies like Apple 1078 01:05:15,040 --> 01:05:20,360 where Steve Jobs said the iPhone is going to disrupt iPod, but I 1079 01:05:20,360 --> 01:05:23,400 would rather disrupt iPod myself than be disrupted with somebody 1080 01:05:23,400 --> 01:05:26,040 else. And whereas a startup has got 1081 01:05:26,040 --> 01:05:28,000 nothing to lose, right? They're starting with zero 1082 01:05:28,000 --> 01:05:30,920 revenue, right? So that's the advantage of a 1083 01:05:30,920 --> 01:05:33,360 startup. And what I write in my book is 1084 01:05:33,360 --> 01:05:38,320 therefore companies like ANSYS, like Kodak, like Motorola need 1085 01:05:38,320 --> 01:05:42,400 to work with the digital printers, the digital phones, 1086 01:05:42,400 --> 01:05:49,360 the digital, the quantum or AIB simulation and embrace them and 1087 01:05:49,360 --> 01:05:51,080 bring them into your thing, right? 1088 01:05:51,080 --> 01:05:55,080 That's the way a company like Ansys can actually stay on top 1089 01:05:55,080 --> 01:05:56,960 of Horizon 3 Innovations, right? Yeah. 1090 01:05:57,840 --> 01:06:01,640 And I think that's probably where I think that, you know, 1091 01:06:01,640 --> 01:06:05,120 Jeff Bezos with his original Amazon leadership principles, 1092 01:06:05,160 --> 01:06:08,920 the one of custom obsession makes sense if you just focus on 1093 01:06:08,920 --> 01:06:13,600 what, what would a customer want that typically always works. 1094 01:06:13,960 --> 01:06:15,600 And that always works. That always. 1095 01:06:15,640 --> 01:06:17,600 And that is has been mind defining philosophy. 1096 01:06:18,440 --> 01:06:20,640 Yeah. Well, maybe to close out, I 1097 01:06:20,640 --> 01:06:25,840 would love just to get some of your, you know, summarized 1098 01:06:25,840 --> 01:06:27,720 advice. I guess there's people listening 1099 01:06:27,720 --> 01:06:31,640 to this who are maybe, you know, coming to the end of their PhDs, 1100 01:06:31,640 --> 01:06:33,280 They're in industry. They have ideas. 1101 01:06:33,280 --> 01:06:35,400 What would be your if you had to summarize? 1102 01:06:35,400 --> 01:06:39,560 What have helped you to get to such an illustrious position 1103 01:06:39,560 --> 01:06:41,880 where, where you are now? How What advice would you give 1104 01:06:41,880 --> 01:06:46,040 to people at the end of this to motivate them in their in their 1105 01:06:46,040 --> 01:06:49,000 careers? So, so the motivational thing is 1106 01:06:49,000 --> 01:06:52,640 exactly what I kind of covered in the last minute, right? 1107 01:06:52,640 --> 01:06:56,880 That they have to just in fact, in my book, I talk about the 1108 01:06:56,920 --> 01:06:59,040 digital technology that we have today, right? 1109 01:06:59,040 --> 01:07:01,800 I talk about quantum, I talk about AI, talk about IoT, talk 1110 01:07:01,800 --> 01:07:04,080 about platforms, all the stuff that is there, right? 1111 01:07:04,600 --> 01:07:09,360 So here is a kid graduating with a PhD in 2024. 1112 01:07:09,840 --> 01:07:13,040 When I graduated the PhD in 1984, forty years ago, I didn't 1113 01:07:13,040 --> 01:07:17,040 have those things, right? So you guys have so much more 1114 01:07:17,040 --> 01:07:19,320 exciting technology at your disposal, right? 1115 01:07:19,640 --> 01:07:24,960 You have to figure out to solve a world's problem of simulation 1116 01:07:24,960 --> 01:07:29,000 or base or whatever, I mean healthcare or, or, or whatever 1117 01:07:29,000 --> 01:07:32,520 problem, right? Try to figure out the 1118 01:07:32,520 --> 01:07:36,000 combination of quantum plus ML plus cloud plus whatever, right? 1119 01:07:36,000 --> 01:07:38,680 I mean, how can I solve the world's problem, right? 1120 01:07:38,920 --> 01:07:42,360 You start with a problem that you're going to solve, right? 1121 01:07:42,760 --> 01:07:45,280 And you have a choice. You could either work in a large 1122 01:07:45,280 --> 01:07:48,480 company and just join that and run on that treadmill and, and 1123 01:07:48,480 --> 01:07:52,080 you'll be guaranteed 100,000 dollar 200,000 salary. 1124 01:07:52,080 --> 01:07:54,680 You have a home, you, you, you whatever, right? 1125 01:07:54,680 --> 01:07:59,280 Or you are passionate. You see, I could take a risk do 1126 01:07:59,280 --> 01:08:04,640 do something interesting. And, and my son who graduated 1127 01:08:04,640 --> 01:08:09,560 from Berkeley, he actually chose the path of the startup, right? 1128 01:08:09,560 --> 01:08:12,320 I mean, he, he, he could have joined many. 1129 01:08:12,320 --> 01:08:14,440 He had a computer science degree from Berkeley. 1130 01:08:14,440 --> 01:08:17,399 He had who have interviewed, he had interviewed at all the large 1131 01:08:17,399 --> 01:08:20,120 companies in that behavior, but he chose to do a startup and 1132 01:08:20,120 --> 01:08:22,880 he's working on a startup in the healthcare area. 1133 01:08:23,279 --> 01:08:25,520 And I, I, I wish him all the luck, right? 1134 01:08:25,520 --> 01:08:29,600 And, and so he has taken a risk. He has taken a much less 1135 01:08:29,600 --> 01:08:33,120 compensation in, during the years of a startup with the hope 1136 01:08:33,120 --> 01:08:36,399 of transforming the world in this healthcare startup called 1137 01:08:36,399 --> 01:08:39,640 Sempra Health that he's doing right along with his wife, 1138 01:08:40,000 --> 01:08:44,040 Anurati. So, so that is the the message. 1139 01:08:44,040 --> 01:08:45,399 I would like to end it with that. 1140 01:08:45,680 --> 01:08:49,160 Follow your passion. You have to take some risks in 1141 01:08:49,160 --> 01:08:51,000 life, right? It is not easy, the life of 1142 01:08:51,000 --> 01:08:53,960 startup, but if it is successful, you will transform 1143 01:08:53,960 --> 01:08:57,479 the world and will transform your personal financial 1144 01:08:57,479 --> 01:08:59,720 situation. But you shouldn't do a startup 1145 01:08:59,960 --> 01:09:02,640 for the financial. You should do the startup 1146 01:09:02,800 --> 01:09:06,720 because you really want to solve a really hard problem that the 1147 01:09:06,720 --> 01:09:10,600 world doesn't know how to solve using the latest technologies 1148 01:09:10,600 --> 01:09:13,240 that you have. And people have not figured out 1149 01:09:13,240 --> 01:09:18,000 how to combine quantum plus AI plus HPC plus cloud plus IoT, 1150 01:09:18,000 --> 01:09:19,720 right? And you are the first guy who 1151 01:09:19,720 --> 01:09:22,600 did it, right? It's that interdisciplinary 1152 01:09:22,600 --> 01:09:24,840 thing. The ability to assimilate is 1153 01:09:24,840 --> 01:09:29,920 what is unique in the startup. So if take a problem that is 1154 01:09:29,920 --> 01:09:33,479 really hard and solve it with the gadgets that you have today, 1155 01:09:33,920 --> 01:09:38,720 which Prince Banerjee in 1984 did not have access to, you guys 1156 01:09:38,720 --> 01:09:41,160 in 2024 have so much more to work on. 1157 01:09:42,800 --> 01:09:48,040 Now that that's great advice and that, yeah, the I agree the the 1158 01:09:48,040 --> 01:09:50,640 passion you need to have, it's a bit like doing a PhD. 1159 01:09:50,640 --> 01:09:52,720 There's no point in doing a PhD just for the sake of it, or 1160 01:09:52,720 --> 01:09:54,920 you'll fail, or you'll. You can still attacking my 1161 01:09:55,040 --> 01:09:55,800 voice, right? I I. 1162 01:09:58,960 --> 01:10:00,840 Thank. You very much for inviting me 1163 01:10:00,840 --> 01:10:02,600 for for the blog. I really, really enjoyed it. 1164 01:10:03,400 --> 01:10:04,600 Yeah. Thank you so much. 1165 01:10:04,600 --> 01:10:08,440 This has been great. ANSYS is very lucky to have you 1166 01:10:08,480 --> 01:10:10,400 as their CTO, so thank you again. 1167 01:10:10,800 --> 01:10:11,360 Thank you, Neil.