WEBVTT - Smart Talks with IBM: How Infrastructure is Powering the Age of AI 0:00:00.040 --> 0:00:02.800 Hey everyone, it's Robert and Joe here. Today we've got 0:00:02.840 --> 0:00:05.080 something a little different to share with you. It's a 0:00:05.080 --> 0:00:08.719 new season of the Smart Talks with IBM podcast series. 0:00:09.160 --> 0:00:11.959 This season, on smart Talks, Malcolm Gladwell and team are 0:00:12.039 --> 0:00:15.240 diving into the transformative world of artificial intelligence with a 0:00:15.240 --> 0:00:18.560 fresh perspective on the concept of open What does open 0:00:18.640 --> 0:00:21.840 really mean in the context of AI. It can mean 0:00:21.960 --> 0:00:25.600 open source code or open data, but it also encompasses 0:00:25.680 --> 0:00:30.760 fostering an ecosystem of ideas, ensuring diverse perspectives are heard, 0:00:31.120 --> 0:00:33.479 and enabling new levels of transparency. 0:00:33.800 --> 0:00:37.040 Join hosts from your favorite pushkin podcasts as they explore 0:00:37.080 --> 0:00:40.879 how opennes in AI is reshaping industries, driving innovation, and 0:00:40.960 --> 0:00:44.800 redefining what's possible. You'll hear from industry experts and leaders 0:00:44.840 --> 0:00:48.320 about the implication and possibilities of open AI, and of course, 0:00:48.640 --> 0:00:50.879 Malcolm Gladwell will be there to guide you through the 0:00:50.920 --> 0:00:52.640 season with his unique insights. 0:00:52.960 --> 0:00:55.520 Look out for new episodes of Smart Talks every other 0:00:55.560 --> 0:00:59.160 week on the iHeartRadio app, Apple Podcasts, or wherever you 0:00:59.200 --> 0:01:02.480 get your podcasts, and learn more at IBM dot com 0:01:02.480 --> 0:01:03.840 slash smart Talks. 0:01:05.560 --> 0:01:08.959 Hello, Hello, Welcome to smart Talks with IBM, a podcast 0:01:09.240 --> 0:01:14.880 from Pushkin Industries, iHeartRadio and IBM. I'm Malcolm Godwell. This season, 0:01:14.920 --> 0:01:17.920 we're diving back into the world of artificial intelligence, but 0:01:17.959 --> 0:01:23.840 with a focus on the powerful concept of open its possibilities, implications, 0:01:23.880 --> 0:01:28.559 and misconceptions. On today's episode, our season finale, I'm joined 0:01:28.560 --> 0:01:32.840 by Rick Lewis, the senior vice president of Infrastructure at IBM. 0:01:33.440 --> 0:01:36.920 Rick has had a remarkable career focused around product innovation. 0:01:37.640 --> 0:01:40.600 He was actually a few years into retirement when IBM 0:01:40.680 --> 0:01:46.080 came calling with an opportunity he just couldn't turn down. Thankfully, 0:01:46.280 --> 0:01:49.600 Rick came out of retirement and today he oversees a 0:01:49.680 --> 0:01:55.160 vast portfolio from storage and software to global customer support operations, 0:01:55.560 --> 0:01:57.880 and he's engaged in one of the key problems facing 0:01:57.920 --> 0:02:02.800 companies today, an explosion of data. In talking with Rick, 0:02:02.880 --> 0:02:05.240 I can see that this problem of having so much 0:02:05.360 --> 0:02:09.440 data is also an incredible opportunity because if you're able 0:02:09.480 --> 0:02:12.160 to leverage that data to get the most value out 0:02:12.160 --> 0:02:14.560 of it, then you can use it to help bring 0:02:14.600 --> 0:02:18.360 your business into the future. We talked about the serious 0:02:18.400 --> 0:02:21.280 computing power needed to scale AI, as well as the 0:02:21.280 --> 0:02:25.480 ways that infrastructure storage solutions can be essential to enabling 0:02:25.560 --> 0:02:30.079 this new world of possibilities. It's a really great conversation, 0:02:30.600 --> 0:02:37.800 so let's get to it. I'm here with Rick Lewis. Rick, Welcome, 0:02:38.240 --> 0:02:41.880 Thank Here. We are in the IBM's New York City 0:02:41.880 --> 0:02:45.240 headquarters at one Madison Avenue. I'm going to start with 0:02:46.040 --> 0:02:47.040 you're a hardware guy. 0:02:47.240 --> 0:02:50.840 I'm a hardware guy. I grew up doing hardware chip engineering. 0:02:51.280 --> 0:02:53.720 But like I tell a lot of people, a chip 0:02:53.800 --> 0:02:57.360 engineering project is actually a giant software project with a 0:02:57.480 --> 0:03:00.160 piece of hardware at the end of the project. I 0:03:00.160 --> 0:03:02.919 think if you have that analytical brain, you like to 0:03:03.000 --> 0:03:06.640 solve problems, you'd like to get things working. You can 0:03:06.680 --> 0:03:07.960 do that in soetwork. 0:03:08.040 --> 0:03:11.400 But as being someone coming from a hardware background mean 0:03:11.440 --> 0:03:14.040 that you think about problems in a different way. 0:03:14.680 --> 0:03:18.320 I think one thing that you do from a hardware background, 0:03:18.320 --> 0:03:21.520 and especially a chip background, is a chip spin and 0:03:21.560 --> 0:03:25.440 costs millions of dollars, So you're a lot more likely 0:03:25.560 --> 0:03:28.639 to make sure everything has a great chance of being 0:03:28.680 --> 0:03:31.160 perfect from the get go. Or if you start kind 0:03:31.160 --> 0:03:34.200 of from a software background, your general mindset is I 0:03:34.200 --> 0:03:35.840 don't know, try this, see if it works. I don't know, 0:03:35.880 --> 0:03:37.440 try that is if it work and you kind of 0:03:37.480 --> 0:03:40.600 it it iterate chip people are a little more uptight 0:03:40.680 --> 0:03:44.200 about okay, if this first round of the chip breaks 0:03:44.440 --> 0:03:47.119 costs us from building another new round of the chip. 0:03:47.280 --> 0:03:50.360 Yeah, so you're a little more You guys are spend 0:03:50.400 --> 0:03:51.160 more time. 0:03:52.000 --> 0:03:56.240 Planning and planning verifying, tons of time verifying. 0:03:57.400 --> 0:04:01.360 So you began your career as you look back, yes, correct, 0:04:01.400 --> 0:04:02.520 and you were there for how many years? 0:04:02.600 --> 0:04:03.920 I was there for thirty two years? 0:04:04.160 --> 0:04:07.920 Yes, And your last job there was I was leading. 0:04:07.600 --> 0:04:10.280 The software defining cloud business. I had grown up a 0:04:10.320 --> 0:04:15.800 hardware guy. I had done all kinds of hardware projects, 0:04:15.840 --> 0:04:20.360 big complicated Unix servers and things like that, and then came, 0:04:20.720 --> 0:04:22.520 you know, grew out of R and D and more 0:04:22.520 --> 0:04:27.279 into the business realm. And then I'm much an innovator 0:04:27.320 --> 0:04:31.760 at heart. I really like innovating new concepts things like that. 0:04:32.040 --> 0:04:35.240 And what I learned is I enjoyed innovating business models 0:04:35.279 --> 0:04:38.400 and software projects as much as I did hardware products 0:04:39.200 --> 0:04:42.920 and projects, and so getting teams inspired towards doing that 0:04:43.320 --> 0:04:45.680 was really a deep fascination for me. So I ended 0:04:45.760 --> 0:04:49.480 up doing a fantastic variety of experiences and had a 0:04:49.520 --> 0:04:54.200 successful run and honestly retired, intending to retire and do 0:04:54.279 --> 0:04:56.440 some of my outside activities and things like that. 0:04:56.960 --> 0:04:59.360 And then how long did you stay retired before IBM 0:04:59.400 --> 0:04:59.839 can close? 0:05:00.120 --> 0:05:04.560 Almost two years? And when I first got at ALL, 0:05:04.800 --> 0:05:07.800 I thought, no, I'm having too much fun. But I 0:05:07.839 --> 0:05:12.719 would say three things really got me thinking hard about it. One, 0:05:13.960 --> 0:05:16.440 the industry that we're in, the IT industry. I think 0:05:16.440 --> 0:05:18.400 it's the golden age. And what I mean by that 0:05:18.600 --> 0:05:22.240 is for twenty years of that career, it is kind 0:05:22.240 --> 0:05:24.640 of in the back office. Hey, make sure that stuff 0:05:24.680 --> 0:05:27.719 doesn't crash, and can you please reduce the cost as 0:05:27.800 --> 0:05:29.800 much as possible, because it's not that important to the 0:05:29.880 --> 0:05:33.320 main business. It's just a back office function. You can 0:05:33.360 --> 0:05:35.599 see it right now. It is at the forefront of 0:05:35.680 --> 0:05:40.200 all business revolution. It happened with the Internet. It happened 0:05:40.200 --> 0:05:44.159 again with cloud and how that changed every ounce of business, 0:05:44.240 --> 0:05:46.920 not just IT business, but all business. And I think 0:05:46.920 --> 0:05:50.159 it's happening again with AI. So to be in that 0:05:50.240 --> 0:05:52.719 career that long and to miss the kind of this 0:05:52.839 --> 0:05:55.159 age where it's like this is front and center. This 0:05:55.320 --> 0:05:59.560 changes everything about all businesses, not just technology businesses. I 0:05:59.640 --> 0:06:03.279 was kind of feeling like, gosh, you you trained to 0:06:03.360 --> 0:06:08.400 be in these really awesome environments. Why wouldn't you do 0:06:08.480 --> 0:06:10.719 that for a little while longer while you still can 0:06:10.800 --> 0:06:14.440 do it. That combined with IBM and IBM seeing the 0:06:14.520 --> 0:06:18.719 talent pool, the brilliant people at IBM, I worked with 0:06:18.760 --> 0:06:21.440 a ton of brilliant people before I saw a chance 0:06:21.480 --> 0:06:24.080 to work with even a larger staff of brilliant people. 0:06:24.200 --> 0:06:27.400 And then the assets that IBM had, which is, you know, 0:06:27.440 --> 0:06:30.160 they'd already been doing a lot of experimentation in AI, 0:06:30.279 --> 0:06:35.480 they're working in quantum, the deep, rich heritage of successful projects. 0:06:35.480 --> 0:06:37.520 I thought, who wouldn't want to kind of see if 0:06:37.520 --> 0:06:40.640 they could be part of that next great wave of IBM. 0:06:40.760 --> 0:06:43.279 And so I kind of decided, all right, I'm going 0:06:43.320 --> 0:06:45.360 to put the outside interest on hold for a while 0:06:45.360 --> 0:06:46.400 and get back in the game. 0:06:46.520 --> 0:06:49.839 Along between the phone call, the first phone call and 0:06:49.920 --> 0:06:50.880 you say, yes. 0:06:50.680 --> 0:06:54.920 It was a while, It was probably six months. Arvin's 0:06:55.000 --> 0:06:57.600 our CEO, teases me about that a lot. Yeah, he 0:06:57.720 --> 0:06:58.080 was like, I. 0:06:58.040 --> 0:07:00.240 Don't think six months is that long? It took a while. 0:07:00.440 --> 0:07:02.800 You're a retirement I know. Yeah. 0:07:02.440 --> 0:07:05.680 It's one thing to compare I'm working here and doing 0:07:05.680 --> 0:07:08.520 this stuff versus working there. It's really hard to compare. 0:07:08.720 --> 0:07:11.240 I'm doing exactly as I want to do every single 0:07:11.320 --> 0:07:13.160 day when I wake up, and now I'm not going 0:07:13.200 --> 0:07:15.960 to get to do that again. It took a while 0:07:15.960 --> 0:07:18.280 for me to get over and I thought, I can't 0:07:18.320 --> 0:07:22.640 miss this wave, and I'm really really happy that I did, 0:07:22.760 --> 0:07:26.440 because we're doing some amazing, fun things and I'm getting 0:07:26.480 --> 0:07:28.960 challenged in ways that I never did, so it's really fun. 0:07:29.200 --> 0:07:32.240 Talk a little bit about your job here at IBM. 0:07:32.440 --> 0:07:36.600 You oversee a kind of massive portfolio. 0:07:36.720 --> 0:07:39.640 It's a big group, so I run the Infrastructure organization. 0:07:39.680 --> 0:07:43.000 There's three main groups of products at IBM. There's the 0:07:43.080 --> 0:07:46.800 Infrastructure group, which I run, the Software group, and the 0:07:46.840 --> 0:07:52.040 Consulting group. And infrastructure is built up of mainframes, which 0:07:52.080 --> 0:07:56.240 is called our Z portfolio, our servers which is our 0:07:56.280 --> 0:08:00.600 power portfolio storage. By the way, those business include the 0:08:00.640 --> 0:08:02.960 supply chain to build all of that stuff, so that's 0:08:02.960 --> 0:08:06.120 in the group. Then I have the worldwide Customer Support 0:08:06.240 --> 0:08:10.520 Organization it's called TLS Technology life Cycle Services, which is 0:08:10.560 --> 0:08:13.920 a network of about thirteen thousand people around the globe 0:08:13.920 --> 0:08:16.040 that make sure that everything runs and works when you 0:08:16.040 --> 0:08:19.760 buy IBM products. And then also our IBM Cloud, which 0:08:19.800 --> 0:08:24.680 is how we host applications and deliver as a service 0:08:24.720 --> 0:08:27.720 products for our client base. So there's a lot. I 0:08:27.760 --> 0:08:30.880 think it's about forty five thousand people total. Do those. 0:08:32.160 --> 0:08:34.959 Components of the Infrastructure group are they aligned in their 0:08:35.000 --> 0:08:38.160 trajectory or are they on different paths? 0:08:38.200 --> 0:08:41.080 And I'm just curious what son little of both. It's 0:08:41.120 --> 0:08:44.000 interesting you would ask that because I think of all 0:08:44.080 --> 0:08:48.079 of the challenges coming to the new company, there were 0:08:48.080 --> 0:08:52.120 things I expected, things that they didn't expect. But getting 0:08:52.200 --> 0:08:57.120 that culture right in that group has been a big challenge. 0:08:57.800 --> 0:09:04.160 IBM has a great culture toward quality products, toward emphasizing 0:09:04.240 --> 0:09:06.800 passion for the client and making sure that the client 0:09:06.920 --> 0:09:10.400 is happy, and for delivering innovation on a scale that 0:09:11.200 --> 0:09:13.880 you know, for more than one hundred years has been 0:09:14.720 --> 0:09:18.760 extremely powerful. But with success comes some challenges. And with 0:09:18.840 --> 0:09:22.959 that success you can tend to get a little bit insular, 0:09:23.160 --> 0:09:25.360 like you don't keep an eye on the competition. As well, 0:09:25.440 --> 0:09:27.960 you can get more siloed, where you know, this is 0:09:28.040 --> 0:09:31.000 my business unit, this is my business unit, I compete 0:09:31.000 --> 0:09:33.400 with the other business unit. That's not a good thing 0:09:33.480 --> 0:09:37.920 when you're a company, and you can get really risk averse, 0:09:38.320 --> 0:09:41.520 meaning you feel like Hey, this is a successful business. 0:09:41.559 --> 0:09:43.280 I don't want to do anything to mess it up, 0:09:43.600 --> 0:09:45.200 so I don't need to try new things. Well, that's 0:09:45.280 --> 0:09:48.520 exactly the recipe to kind of be shrinking, and infrastructure 0:09:48.559 --> 0:09:50.840 had been shrinking for a little while, and so a 0:09:50.880 --> 0:09:54.480 lot of what the challenge was for me was to 0:09:54.520 --> 0:09:58.720 invigorate that risk taking and get to a growth mindset 0:09:58.720 --> 0:10:01.360 where you're trying new things and seeing what works and 0:10:01.400 --> 0:10:03.720 what doesn't work, and changing some of the models, like 0:10:04.040 --> 0:10:06.720 investing a little bit less in hardware for some software 0:10:06.720 --> 0:10:10.480 differentiation that goes into the hardware. So it's been very 0:10:10.480 --> 0:10:12.960 successful so far, and it's been a good journey. It's 0:10:12.960 --> 0:10:14.160 almost four years now. 0:10:14.559 --> 0:10:17.480 Give me an example of what was a really hard 0:10:17.520 --> 0:10:19.240 problem that you've dealt with in those four years. 0:10:19.240 --> 0:10:22.160 So, boy, a really hard problem. 0:10:22.240 --> 0:10:25.200 An interesting and are you interesting is a better word 0:10:25.240 --> 0:10:25.560 than art. 0:10:25.840 --> 0:10:27.840 One of the first things that I kind of chewed 0:10:27.880 --> 0:10:30.520 on a little bit is I talked about how we 0:10:30.559 --> 0:10:33.679 have Z power and storage. The Z and power product 0:10:33.720 --> 0:10:36.480 lines are well known in the industry. Is is really 0:10:36.520 --> 0:10:39.840 fit for purpose computing that have strengths that you know 0:10:39.960 --> 0:10:43.880 Z runs you know most of the world's economic backbone, 0:10:43.920 --> 0:10:47.520 and if you use a credit card. Ninety percent of 0:10:47.559 --> 0:10:51.040 credit card transactions for the globe go through these Z 0:10:51.520 --> 0:10:54.080 mainframes there in every bank there. You know, it's a 0:10:54.080 --> 0:10:56.880 big business. It's well known in the industry. Same with power, 0:10:57.320 --> 0:11:02.240 very tuned and optimized for smaller operations than our giant 0:11:02.360 --> 0:11:08.319 Z mainframes, but really mission critical workloads for retail, for insurance, 0:11:08.320 --> 0:11:11.520 for banking, for all of that. Our storage business not 0:11:11.640 --> 0:11:14.439 so well known. In fact, when I came I thought, 0:11:14.679 --> 0:11:17.079 did they have storage? Well, I have storage when I 0:11:17.240 --> 0:11:19.680 come in too. I So I got online and I thought, 0:11:19.760 --> 0:11:21.560 it's still hard for me to tell did they have 0:11:21.679 --> 0:11:24.000 storage or not? Now I own a storage business. So 0:11:24.080 --> 0:11:25.640 one of the things was not just to get the 0:11:25.679 --> 0:11:29.280 market perception up, but to invest in that business. Because 0:11:29.320 --> 0:11:32.000 if you look at infrastructure overall around the globe, it's 0:11:32.040 --> 0:11:35.600 growing at five percent a year. The infrastructure business had 0:11:35.640 --> 0:11:38.320 been kind of flat to declining, and so a challenge 0:11:38.360 --> 0:11:40.319 was how do we grab onto the growth. Well, one 0:11:40.320 --> 0:11:43.040 of the biggest growth areas due to the explosion of 0:11:43.120 --> 0:11:45.720 data in the world is storage, So what do you 0:11:45.760 --> 0:11:47.960 do to kind of get on that growth rate. So 0:11:47.960 --> 0:11:51.600 we did a lot of reinvigoration of the innovation in 0:11:51.640 --> 0:11:54.520 that a lot of software value, add a lot of 0:11:54.840 --> 0:11:57.920 doubling down on the things that are working. Portfolio rationalization 0:11:58.000 --> 0:12:00.200 where you segment the market and you say, okay, we're 0:12:00.360 --> 0:12:02.160 going to do less of this and really go big 0:12:02.200 --> 0:12:05.079 in these areas. And that's been probably the most dramatic 0:12:05.080 --> 0:12:08.440 turnaround inside the group. Is our storage thing. When you 0:12:08.480 --> 0:12:11.200 say it's a hard problem, it's not just oh, you know, 0:12:11.240 --> 0:12:15.800 how do we do the math? No, it's cultural. It's strategy, 0:12:15.880 --> 0:12:18.040 and how do you get the strategy set. It's segmentation, 0:12:18.160 --> 0:12:21.280 it's product strategy at a granular level across a bunch 0:12:21.320 --> 0:12:24.640 of dimensions, and then putting the investment behind it. It's 0:12:24.679 --> 0:12:26.880 a big challenge. It takes a long time, but it's working, 0:12:26.960 --> 0:12:28.920 so we're happy with Yeah. 0:12:28.600 --> 0:12:30.559 Tell me give me a little bit of perspective on 0:12:31.120 --> 0:12:34.920 you've been there four years. Imagine we're having this conversation 0:12:35.000 --> 0:12:35.640 four years ago. 0:12:36.120 --> 0:12:38.280 Yeah, what sorts. 0:12:38.080 --> 0:12:40.200 Of things have happened over the last four years that 0:12:41.200 --> 0:12:43.680 have surprised you that you didn't see come? At least 0:12:44.480 --> 0:12:46.600 we had exactly the same conversation four years ago. 0:12:48.000 --> 0:12:49.720 No, because I didn't know what was in I'll tell 0:12:49.720 --> 0:12:53.559 you some of the biggest surprises I thought from the outside, 0:12:53.960 --> 0:12:57.199 and you know, you hear from a lot of customers, 0:12:57.360 --> 0:13:00.560 especially ten years ago, we're all going to we're all 0:13:00.840 --> 0:13:03.360 so I thought, well, I wonder if the mainframe business 0:13:03.400 --> 0:13:06.160 is struggling. When I get inside of there, I found 0:13:06.200 --> 0:13:08.720 the opposite to be true. The mainframe business is actually 0:13:08.760 --> 0:13:12.839 flourishing because transaction demand across the globe has done nothing 0:13:12.920 --> 0:13:16.120 but grow. And even more surprising was the level of 0:13:16.200 --> 0:13:19.840 innovation that the team was already doing in mainframes before 0:13:19.880 --> 0:13:24.400 I got here was astounding. For example, we have AI. 0:13:24.880 --> 0:13:29.360 They were building AI technology into the mainframe processors three 0:13:29.440 --> 0:13:32.720 years before chat GPT made everybody talk about it in 0:13:32.760 --> 0:13:36.840 the industry, So that was really pleasantly surprising. So that 0:13:37.000 --> 0:13:41.840 was wonderful. Other surprises I knew about the kind of 0:13:41.840 --> 0:13:44.960 the IP of IBM and the mystique in that, and 0:13:45.040 --> 0:13:47.520 I used to joke with people, especially on the outside, 0:13:47.520 --> 0:13:48.960 I said, I can't wait to get in there and 0:13:48.960 --> 0:13:51.800 see what's in the big blue toolbox? Right, what are 0:13:51.840 --> 0:13:55.160 all the things they have going on? I way underestimated 0:13:55.360 --> 0:13:57.720 the size of the big blue toolbox and what was 0:13:57.760 --> 0:14:02.640 in their meaning amount of really hardcore research that we're 0:14:02.679 --> 0:14:06.040 still doing into how to build chips and how to 0:14:06.080 --> 0:14:09.079 get to things beyond two nanimeter and that kind of 0:14:09.160 --> 0:14:15.160 capability packaging industry leading packaging technologies, and that's in my 0:14:15.320 --> 0:14:18.079 hardware kind of patch quantum. The next thing that will 0:14:18.120 --> 0:14:22.720 come after we're done talking about AI. You know, all 0:14:22.760 --> 0:14:25.560 of those things were surprising, But it wasn't just that. 0:14:25.640 --> 0:14:28.200 It was then the software innovations that are going on 0:14:28.360 --> 0:14:33.160 heavy investment in AI technologies before it was really popular 0:14:33.200 --> 0:14:35.840 to be talking about that. But as I saw that, 0:14:35.920 --> 0:14:38.560 I thought this is going to get really fun. Because 0:14:38.600 --> 0:14:41.080 I had a good feel for where the industry was going. 0:14:41.920 --> 0:14:44.440 I just didn't and I knew, man, I know that 0:14:44.560 --> 0:14:47.200 talent is really good and there's brilliant people there, but 0:14:47.240 --> 0:14:51.160 I didn't know the level of IP frankly that IBM 0:14:51.280 --> 0:14:54.200 had at its disposal. And now you're seeing that in 0:14:54.240 --> 0:14:57.480 things like Watson X and things like AI in mainframes, 0:14:57.480 --> 0:14:57.880 et cetera. 0:14:58.280 --> 0:15:01.520 Building on that. Since you put AI, can you walk 0:15:01.560 --> 0:15:05.360 me through what has to happen from your perspective, from 0:15:05.400 --> 0:15:11.280 the infrastructure perspective to make the AI explosion work? Yeah, 0:15:11.320 --> 0:15:13.320 so everyone wants to do more of this stuff. Yes, 0:15:13.520 --> 0:15:16.000 clearly there has to be some underpinning of it. 0:15:16.440 --> 0:15:20.000 Yeah, I would tell you, you know, I think that 0:15:20.160 --> 0:15:22.440 people feel like where we're at right now in the 0:15:22.480 --> 0:15:25.560 AI journey had to do with one specific piece of software. 0:15:25.920 --> 0:15:30.040 I think the inflection point for that whole thing really 0:15:30.960 --> 0:15:34.680 at its root was around hardware, meaning the algorithms needed 0:15:34.720 --> 0:15:37.280 to do larger language models. And all of that had 0:15:37.360 --> 0:15:40.000 been around, they'd been talked about in the industry, but 0:15:40.040 --> 0:15:43.080 at some point you hit a tipping point of hardware 0:15:43.120 --> 0:15:45.760 capability where it's like, oh, now we can do this 0:15:45.880 --> 0:15:49.280 in a broof force way, massive amounts of matrix math 0:15:49.360 --> 0:15:52.320 to get weights correct so that you can do you know, 0:15:52.360 --> 0:15:55.680 the right level of predictions that enable large language models. 0:15:56.040 --> 0:15:58.600 And once we got to that horsepower, and that's why 0:15:58.640 --> 0:16:01.240 you hear about giant g pus that are driving this 0:16:01.600 --> 0:16:03.720 and the sales of those, et cetera. It's because we 0:16:03.920 --> 0:16:05.760 just barely got over the hump where you can do 0:16:05.800 --> 0:16:10.440 these big hard things in terms of hardware capability to 0:16:10.600 --> 0:16:11.320 do it. 0:16:11.440 --> 0:16:14.280 Give me a layman, give me a sense of when 0:16:14.320 --> 0:16:16.480 you say there was a kind of threshold where suddenly 0:16:16.480 --> 0:16:17.640 these things became possible. 0:16:17.760 --> 0:16:20.560 Yeah, I don't know if there's an exact number, but 0:16:21.720 --> 0:16:23.800 and more basic question that I get from a lot 0:16:23.840 --> 0:16:26.520 of people, you know, my friends and family outside. Is 0:16:26.840 --> 0:16:31.560 why GPUs. What does a GPU, a graphics processor have 0:16:31.680 --> 0:16:36.080 to do with AI. It's not, Well, graphics processors are 0:16:36.120 --> 0:16:39.880 really good at this thing matrix math, because they're figuring 0:16:39.920 --> 0:16:43.280 out how do I map a pixel? And as I 0:16:43.400 --> 0:16:47.560 move an object across the screen, it's essentially matrix math 0:16:47.640 --> 0:16:50.320 to figure out, Okay, what does what does this pixel 0:16:50.400 --> 0:16:53.120 on a screen look like? And what it's doing? And 0:16:53.160 --> 0:16:56.480 as you you know, we've gotten more high resolution graphics, 0:16:56.520 --> 0:16:59.320 more high resolution monitors, et cetera. It's a lot more 0:16:59.320 --> 0:17:01.240 pixels and a lot more math and a lot more 0:17:01.280 --> 0:17:04.359 matrix math about how you compute that. The first big 0:17:04.359 --> 0:17:06.679 thing that kind of started to look like that, it 0:17:06.720 --> 0:17:10.199 turns out, was crypto and crypto mining, and so you 0:17:10.240 --> 0:17:13.959 saw graphics companies starting to sell to crypto. The technology 0:17:13.960 --> 0:17:15.960 got to a certain point and there were use cases 0:17:16.000 --> 0:17:18.480 like bitcoin in that that kind of said, hey, we 0:17:18.520 --> 0:17:20.720 need to do a lot of this matrix math to 0:17:20.760 --> 0:17:23.440 be able to do that. So graphic chips were a 0:17:23.520 --> 0:17:26.440 natural fit and that kind of sustain But meanwhile, behind 0:17:26.480 --> 0:17:29.320 the scenes, a lot of this AI AI is about 0:17:30.000 --> 0:17:34.439 numeric calculations having to do with weights and matrices that say, 0:17:34.880 --> 0:17:38.480 you know, giant consolidated things that predict what's going to 0:17:38.680 --> 0:17:41.159 kind of happen based on what other things have happened, 0:17:41.200 --> 0:17:44.880 just like predicting where pixel goes. But it's really about 0:17:45.480 --> 0:17:48.560 being able to do enough data in jest to be 0:17:48.560 --> 0:17:51.160 able to do and then the calculations to be able 0:17:51.200 --> 0:17:55.640 to simplify things like entire sets of language or giant 0:17:55.920 --> 0:17:58.640 chunks of the Internet, to get enough weightings in there 0:17:58.640 --> 0:18:01.280 to be able to say, okay, we can predict what 0:18:01.320 --> 0:18:04.720 you would say in this language based on all of 0:18:04.720 --> 0:18:07.359 the volumes of stuff that we've seen that when you 0:18:07.520 --> 0:18:10.120 start talking like this, the next word is likely, oh 0:18:10.160 --> 0:18:11.040 it's this. Yeah. 0:18:11.080 --> 0:18:13.320 So, But my point is to get to that point, 0:18:13.440 --> 0:18:17.240 that's threshold. We got there because was it because we 0:18:17.320 --> 0:18:20.159 simply threw a lot more resources at the problem or 0:18:20.240 --> 0:18:24.480 is it because the underlying technology got suddenly or gradually 0:18:24.640 --> 0:18:25.800 so much more efficient. 0:18:25.920 --> 0:18:29.160 It's always yes and yes. But you know, the industry 0:18:29.240 --> 0:18:31.400 for a lot of years would talk about Moore's law. 0:18:31.880 --> 0:18:35.639 Well, quick, will you define for us More's law for 0:18:35.680 --> 0:18:37.240 those of those who's forgotten it. 0:18:37.480 --> 0:18:40.640 Yeah, So Gordon Moore at Intel coined this thing. It 0:18:40.680 --> 0:18:45.399 was basically that the horsepower I'm going to translate it 0:18:45.520 --> 0:18:51.120 roughly of technology will double every couple of years. We're 0:18:51.160 --> 0:18:53.840 still on Moore's law. Moore's law changed a little bit. 0:18:54.240 --> 0:18:57.080 For a while, it was always about frequency. Things would 0:18:57.080 --> 0:19:00.280 go faster, faster, faster. That kind of petered out. But 0:19:00.359 --> 0:19:03.560 what happened is, rather than faster, faster, faster, we did 0:19:03.560 --> 0:19:06.840 more and more and more. So rather than one operating 0:19:06.960 --> 0:19:10.800 unit going a lot faster on its throughput, you put 0:19:10.840 --> 0:19:13.280 ten operating units on a chip, now you put one 0:19:13.359 --> 0:19:16.399 hundred operating units on a chip, now a thousand. Some 0:19:16.480 --> 0:19:21.120 of these problems, the matrix math problems scale parallel extremely well. 0:19:21.119 --> 0:19:23.399 You don't have to do something really fast, you just 0:19:23.520 --> 0:19:25.520 have to do a lot of the similar things in 0:19:25.560 --> 0:19:28.520 parallel at the same time. So again that kind of 0:19:28.520 --> 0:19:31.240 that extension of Moore's law, more and more hardware on 0:19:31.320 --> 0:19:32.720 a chip to be able to do more and more 0:19:32.720 --> 0:19:35.960 of those calculations in parallel and come up with it. 0:19:36.080 --> 0:19:39.200 And we said, yeah, was that threshold predictable? In other words, 0:19:39.200 --> 0:19:42.080 see people in the industry, like you sit down X 0:19:42.119 --> 0:19:44.160 number of years ago and say, when we get here, 0:19:45.160 --> 0:19:48.200 AI is going to become much more of a It's funny. 0:19:48.760 --> 0:19:55.159 The horsepower that very predictable, the use cases not always 0:19:55.240 --> 0:19:57.639 so easy to kind of figure out. That's where the 0:19:58.080 --> 0:20:01.199 human spirit kind of gets involved. I think for some 0:20:01.240 --> 0:20:03.600 people that say, oh, I saw that coming, but people 0:20:03.640 --> 0:20:07.280 have been predicting kind of the rise of AI for 0:20:07.640 --> 0:20:09.600 twenty five years. Oh well, then when we get to 0:20:09.600 --> 0:20:11.800 this next gener oh when we get here, it kind 0:20:11.800 --> 0:20:15.600 of hadn't happened. There's always a magic point where you 0:20:15.720 --> 0:20:18.040 kind of get to where the technology and the use 0:20:18.080 --> 0:20:20.479 case and somebody does something to kind of make it 0:20:21.280 --> 0:20:23.120 catch on. And I think we're at one of those 0:20:23.119 --> 0:20:25.120 moments in AI for sure right now. And I don't 0:20:25.160 --> 0:20:27.080 think it's you know, people that have said, oh, this 0:20:27.200 --> 0:20:30.399 is just the latest wave of you know, I hear 0:20:30.680 --> 0:20:33.199 I've heard this about a lot of technologies, but AI 0:20:33.320 --> 0:20:35.720 is the technology the future, and it always will be. 0:20:35.960 --> 0:20:38.840 I used to hear that. You're not hearing that now, right, 0:20:38.880 --> 0:20:42.560 It's like, no, it's primetime. It will change everything, just 0:20:42.720 --> 0:20:44.760 like some of these other things changed everything. 0:20:45.000 --> 0:20:49.439 I noticed it if personally when I speak somewhere or 0:20:49.480 --> 0:20:53.359 I'm listening an audience somewhere. Over the last let's say 0:20:53.480 --> 0:20:58.600 twelve months, there's always a whole bunch of AI questions. Yes, 0:20:58.760 --> 0:21:00.960 And if I go back to years ago, there were 0:21:00.960 --> 0:21:01.840 no AI questions. 0:21:01.920 --> 0:21:02.120 Yes. 0:21:02.720 --> 0:21:05.040 Now my question is, so there's been this explosion on 0:21:05.080 --> 0:21:09.480 the in popular fascination with what's going on AI. It 0:21:09.560 --> 0:21:12.560 seems like the last year. I agree with you in 0:21:12.640 --> 0:21:18.920 your world, when did the explosion of conversation around this start? 0:21:19.640 --> 0:21:28.960 It's I love this question because IBM had a fairly 0:21:29.119 --> 0:21:34.199 big effort and business called Watson before Watson X. And 0:21:34.240 --> 0:21:37.160