Understanding Hyperscalers: Masters in Business with Ankur Crawford

32:34saturated as in fully priced or saturated as in, hey, that's as big as their market share. Yeah, that's that's as big as their market is going to get. So what you see is oftentimes companies that are, you know, their growth, their growth, they were great growth companies, and all of a sudden their growth has stabilized or growth is starting to kind of approach GDP. I remember that kind of more mature, right? So kind of a more mature company. And then the management has a decision to make. Am I still a growth company or am I going to just milk what we have? And oftentimes that, that begets change.

33:11So a new management comes in and decides we're going to kind of jettison all our low growth businesses and start buying higher growth businesses. And it changes the profile of the business. It could be a regulatory change that makes the, the company a little bit more growthy than it was historically. It could be M &A that, again, re-accelerates top-line growth. It could be a technological change that they really embrace. And this was Microsoft in its early days back when Satya Nadella first came to the helm. So it's almost as if the company had a decision to make And we're looking for changes that the decision from here is to get onto a growth trajectory and then study how they execute such that it drives both top line and bottom line growth.

34:07So that, you know, oftentimes when we buy companies that are on that side of the ledger, people will think they're value names. And they're not really value names. They're actually unidentified and misunderstood growth. And that's how we think of them. And a great example of this is, you know, what's happening to the hard disk drive companies right now, where, you know, they were trading at single digit multiples. But in an era of AI, all of a sudden, you need a lot more data and you need to store all that data. So, you know, hard disk drives all of a sudden became, you know, in shortage. And now they're taking pricing and their earnings power has, you know, gone up three, four, five fold over the last few years.

34:54Even though people thought it was at the tail end of their useful life cycle, they found a second life. That's right. And so that is also a change, and it happens to be a change in the market broadly. So that raises a really fascinating question I have to ask you. There are companies that appear to be on the back end of their life cycle. Their growth has plateaued. Maybe they're not gaining market share. Maybe the market itself isn't growing. How can you identify when something is legitimately fading or potentially at the start? like I know IBM just had a rough quarter but how many times has that company reinvented itself and been left for dead only to surprise everybody and and there's a bunch of other Microsoft you brought up is another example what were they 30 40 years old when Nadella came in that that's a huge turnaround story so so how do you identify when hey these guys are no never gonna be what they once were, or no, there's something real happening.

36:12Okay. So, so there is this publicly traded fintech company that, you know, was just struggling in part because they had saturated their markets and there was nowhere for them to grow. And it was becoming a lot more competitive. CEO and CFO leave, new management comes in, you know, put together a brand new strategy that is fantastic. Our team looks at it, it's like promising. However, the core issues of their business have not been resolved, right? Do you go from a four and 5 % grower to a 10, 12, 15 % grower with the strategy? We couldn't really, you know, resolve that they would be able to get there because the pressures in their markets were so significant.

37:03Competitive, mature, etc. Competitive, maturity, they were just fighting to kind of stay alive or stay at that 3%, 4%, 5 % type growth. So that was one that we looked at. The catalyst was a new CEO, a new management team. The entire management team was different. But to us, it wasn't really logical that it would change or they could change the trajectory of the business. Microsoft, a completely different story. because Satya comes in, he says, we're going to turn the ship. We're going to develop cloud. And then we started to understand what it meant to go to a SaaS-based business. Gosh, it can be, you know, in the near term, it would be depressing their earnings, but longer term, it's really interesting, right?

37:55And they can get to a mid-teens type growth again, which they did get to. I mean, Microsoft, if you remember, everyone thought Google was going to take over Google Sheets was going to take over Excel. And like, why do we all need Microsoft? I ask myself that question every time I launch and look at the annoying new ribbon that they changed a decade ago. But I use both. But you use both. Yeah. And after all these years, and I assume in 10 years, we'll still be using Microsoft. So, you know, Satya then pivoted the ship and got into the cloud business with Azure. And so we watched the actions as well.

38:35So we can dream the dream and then test the hypothesis and see whether or not they're executing against it. Really interesting. Let me reverse the question to you and say, what leads you when you're running a concentrated portfolio to say, I'm going to sell this? Is it the fundamentals deteriorating, the thesis not working out? Sometimes is it based on valuation? Or is it simply, we only have room for X number of companies and this opportunity is here and that opportunity is all the way up here? It's all of the above, right? There is examples of selling a company because there's a better opportunity and you don't want to take double the risk to the same end market, yet the upside of one is greater than the upside in the other.

39:26There are examples of, you know, you sell or at least trim because the price target has been achieved and maybe beyond the price target has been achieved. So the risk reward is simply different. There are examples of disappointments, you know, companies that disappoint relative to our expectations and And they didn't deliver on what we expected them to do. And the hypothesis didn't play out. So I think there's all of the above. And every sale has a different reason. Really, really interesting. Coming up, we continue our conversation with Dr. Ankur Crawford, Executive Vice President and Portfolio Manager at Alger, diving in to her AI thesis.

40:13I'm Barry Ritholtz. You're listening to Masters in Business on Bloomberg Radio.

40:43performance with a low-cost edge, which helps clients earn returns that compound over time. A rigorous approach to risk management holds it all together. The goal? Keeping income instability in and unwelcome surprises out, because fixed income shouldn't feel like a roller coaster. Explore institutional quality bond funds from Vanguard. Learn more at Vanguard.com slash audio. That's Vanguard dot com slash audio. All investing is subject to risk. Vanguard Marketing Corporation Distributor. Get the latest news from around the world delivered to you in an instant. Subscribe to Bloomberg News Now and get news when you want it on your schedule.

41:28These are short audio reports on the day's top stories that you can listen to in just a few minutes. Make Bloomberg News Now your trusted source for the latest global headlines with context. It's available 24 hours a day, anytime you need it, right on your smartphone or smart speakers. Subscribe to Bloomberg News Now today on Apple Podcasts, Spotify, or anywhere you listen. I'm Barry Ritholtz. You're listening to Masters in Business on Bloomberg Radio. My extra special guest today is Dr. Ankur Crawford. She's portfolio manager at Alger, where she co-PMs the large capital appreciation strategy and runs the concentrated ETF for the firm.

42:09So we are legally obligated to discuss artificial intelligence, but you're the perfect person to have this conversation with. There's a quote of yours that I found fascinating. You said, when software begins to write software, innovation becomes exponential. That's already happening. walk us through what this means for earning powers for the semiconductors for the hyperscalers and then for the rest of the s &p 500 okay so that is a very big question um look i think we are at this i mean elon will call it a singularity or um we're at this point in time where we have never seen this kind of innovation.

43:00And imagine everything that, you know, let's take, it's easiest to describe with software. We used to sit and code software, right? And we had to understand the coding. We had to debug it. It would take a long time. Well, when software begins to write software, that whole process is truncated. And imagine what can be done in our largely digital world when software begins to code, decode, and create. So let me push you a little bit there. The large language models that are out there give AI the ability to effectively cut and paste everything that's been done before. How good is AI at creatively innovating code that's never been written before?

43:58So, look, I am not a coder. So I can't tell you whether the code is elegant or, you know, can be taken to production. I will tell you that I was able to build a pretty interesting app inside of a few months. and this is just doing it on the weekends, occasionally on the weekends, not even every weekend.

44:28So, you know, and that was all vibe coded. So it is adding this technology that is highly viable. You talk to coders, they are using it 90 % of the time and are now just instructing and have to have this, the logical framework of how to use the code. And I think that the big picture here is that once the code begins to write the code, then it's not going to necessarily be creative. The creation still has to come from you. The insight still has to come from you. But it can actually innovate, right? The innovation curve for you is significantly higher. So that's what we're seeing today. where these digital assets are becoming more innovative, or they're allowing us to be more innovative.

45:25And we've hit that point in time where we're getting exponential innovation. And we've never really seen anything like this before. You know, humanity hasn't seen this before in such a short period of time. You know, if you look at previous industrial revolutions, they would be, you know, over generations. it wouldn't be coming in the span of five years. And so this is what makes it really interesting because you asked like, how good is it for semiconductors and how good is it for the rest of the S &P and the hyperscalers? The impact on all of these differs. So software is at the, we wrote a paper three years ago called AI and the Declining Cost to Create.

46:11And it was all about how when software begins to write software, the cost to create software goes to zero. And what happens to the incumbents when the cost to create software is zero? Right? One of the moats goes away. And that necessarily means that the operating profit of businesses must change. Not that software is dead. It's just that the operating profile of all of the companies must change because it becomes more competitive, right? And where does that value go? We had$5.5 trillion of spending, now$6 trillion of IT spending. 50 % of that was IT services and software. And our contention was that the value would go from IT services and software into hardware networking because that is really what is driving this innovation curve.

47:08So, you know, there's entire sectors that have been kind of grown a lot and others that are facing their own pressures. I would say the same thing for any sector in the market. You know, we spoke about health care earlier. How can a UnitedHealthcare actually use AI to bend the cost of care? And can there be incumbents that cross the chasm? or there might be some that can't cross the chasm. And there are new companies that begin to use AI to bend the cost of care. So I'm glad you brought up health care. I've been fascinated not so much by bending the curve of cost from somebody like United, But all of the small biotechs and new molecules and huge wealth of existing chemistry and pharmaceuticals and studies we've done that nobody's really had the ability to go back to and say, hey, maybe something's in here that we've missed.

48:15um you know the the most cliched example is uh i never pronounce it right sylidinephal viagra was supposed to be a heart treatment and had this unusual side effect and now it's a multi-billion dollar med i'm same thing with glps and originally for diabetes but hey everyone's losing a lot of weight on these i'm curious not so much on the cost side but there's this giant body of of unexcavated research that just seems like it's waiting for AI to attack it. Yeah. And so, you know, recently I met with the, I was actually on a panel where I was the moderator for a company and I've forgotten, I've forgotten the name of the CEO and the company, but they're, they're basically a new AI company that is taking this compendium of knowledge and taking it to companies and saying, marry it with the data that you have.

49:13And can we start finding not only the solutions, but for your targets, but use this history to get there faster. So there's lots of efforts being made on this right now. I do think that we will accelerate drug discovery and the impact it will have to healthcare. I mean, look, the holy grail is to have personalized healthcare at some point. Well, wasn't DNA testing supposed to give us that a couple of years ago? Well, yes, but DNA testing used to cost a million dollars per sample. And today it's 100. So we're getting to the point where we can actually look at our individual DNA. And it just takes time.

50:01And at some point, can we marry it with some AI insights? So, look, I think healthcare is going to be greatly impacted. I think that I'm most excited, actually, to see how we can democratize healthcare, because really our healthcare system here is kind of broken. Not kind of. To be polite. And how can we take down that cost of care? And really, I would love to have universal healthcare. It just can't be done in the construct of healthcare as it is today. So, you know, can we use AI to provide universal health care? I think we can. It will take a few years, maybe a decade, but I think we can. And this is a global statement.

50:49It's not necessarily just the U.S. It's bringing the cost of care down enough such that, you know, anyone on this planet will have access to health care. So I'm going to assume that you think all the AI bubble talk is wildly overblown. Yes, I do think it's wildly overblown. I think the trade has gotten a bit harder. And in part because the first two to three years of the trade was, oh, you just have to buy the GPUs. And anything that the GPU touched was gold. And then it became more nuanced. Well, agents use CPUs and we have a memory shortage and memory has now gone up 4x in price. So CapEx budgets are going up.

51:39So that question of ROI is coming to the fore. And how much does CapEx have to go up to accommodate the supply chains being as tight as they are? And there's technological differences between a CPU versus a GPU and how they're used. And the Chinese might be coming. And right. So so there is like a lot of different aspects that have made it a little bit harder where, you know, open versus closed source debate, like the open model versus a closed model. That's another debate. The debt and the CDS spreads widening. That's another. So all of a sudden, we've gone from a relatively simple, we're going to need AI, we're going to need compute, to there's a slew of different narratives that one can press on for the bear case.

52:32Now, I structurally believe that we just talked about healthcare and the innovation curve in healthcare and what that can give back to society. that is true value, right? If we can bend the cost of care from X to X minus, that is value that's created for humanity. And we will pay for that value. You know, the other day, there's been this big debate about token maxing. And there was... Define that for the lay listener. Yeah, token maxing was this behavior that companies were encouraging their engineers to basically have leaderboards of who can use the most tokens, which sounds insane, right? It would almost be like, you know, telling your employees to see how much they can spend on lunch and whoever spends the most on lunch gets an award, right?

53:30Well, I imagine if you're a FedEx driver and the company holds a competition to who's going to go through the most amount of gas and tires, meaning making the most deliveries, not a bad thing for the company. Not necessarily a bad thing. But, you know, in this case, the tokens that were being used or what was being used is actually not necessarily tied to deliveries. It was just use the most tokens as you can. It didn't kind of matter what you built with it. Right. So, or there wasn't as much scrutiny as to how many quote unquote deliveries you made. You just burned through your tires. So, so it was kind of inefficient.

54:21But, you know, they came out and they said, you know, we blew through our entire budget in a quarter. For the year, the whole budget. For the year. The entire budget for the year in a quarter. And, you know, and we haven't gotten an ROI. Well, no kidding. Well, they turned around last week and they laid off 10 % of the people that work for the company because of AI. Well, you know, somewhere along the way, the use of artificial intelligence allowed them to kind of refine their workforce. That sounds like they didn't lay off people because of AI. They sound like they laid off people because management was kind of misincentivizing the employees.

55:02Well, I mean, they said that they laid off people because of AI. There's been many companies like Jack Dorsey at XYZ. You know, also, you know, he cut 40 % of the staff blaming AI. Who knows really what the real reason is? It could be AI or it could be they just overhired. Which he has a history of. Which he has a history of. If you track him over his various companies. Right, and many of these companies did, right? So I can't deconvolve that. However, Uber in particular said it was because of AI. They have been very front foot forward on the use of AI. And now they're able to increase productivity enough that they can titrate down their workforce.

55:49So, you know, I do think that there is value that is being created because of AI. I think that it is not necessarily a technology that's plug and play into an enterprise. And there has to be some learnings before you can get to that ROI. And we're seeing those stumbles and that learning curve in all these examples. It doesn't mean that it's never going to work. And my viewpoint is that where there is value, we work in a system of rewarding value. So if you can create value, I believe that whoever uses that system that creates the value, they will pay for it. So you've described the demand for computers insatiable.

56:39What would have to happen for you to say, all right, we're getting to saturation or satiation? What does the top of the cycle look like? Or is it so far off in the future that we can't even think about it? What I would say is that this is not a question that I can say, oh, we'll never, like in 2030, we won't need compute. I think it's a function of how much we put into the ground, right? It's a delicate balance of, you know, if we put X into the ground today, we put, you know, The hyperscalers are spending$650 billion or whatever that number is. It's circular. It's this, it's that. We've heard these complaints now for two years.

57:21But$650 billion seemed like a really big number, yet we are still short compute. You're hearing from the hyperscalers, we do not have enough. The neoclouds are telling you that there are four times as many asks for compute as they have capacity. Wow. So if one says that we are short compute today, I don't really understand the logic. Now, let's fast forward two and three years. If we put$3 trillion into the ground next year or the year after, which we cannot do today because we are short power, we are short people, we are short capacity, we are short, you know, we can't make those chips. But let's hypothetically say we put in$3 trillion of compute into the ground in 2028, I would say that that is overcapacity.

58:09But we can't do it because there is almost a natural limiter to the growth of this market. And we don't have the chips, we don't have the people, we don't have the power. Right? And so the market is being capped. If all normal forces and if we had an infinite supply of everything, I think we would be in overcapacity today because it's such a big market. Everyone would be building at a pace that, you know, they wanted to be that they would want to be first. But the fact is, it's actually a blessing that the market is being capped by all of these supply chain shortages. The fact that we don't have plumbers and electricians to actually work in our data centers, is capping the growth of data centers.

58:59And so it is allowing for duration versus kind of having a one-time growth pop, which you were not going to pay a high multiple for. So I think that oversupply is a function of how much we put into the ground and how we use it. Hi, I'm David Weston. Join me every week for the Wall Street Week podcast to hear stories of capitalism from around the world, from geopolitical tensions and central bank decisions to artificial intelligence, energy and infrastructure. We sit down with the CEOs, economists, policymakers and thought leaders whose decisions are shaping markets everywhere we find them. Subscribe to the Wall Street Week podcast on Apple, Spotify or anywhere you listen.

59:49So, so let's, let's unpack some of that. In the beginning of 25, when when DeepSeek first kind of was released and everyone was startled, and the initial reaction was, oh, we've overbuilt, we don't need this many GPUs, we don't need all these giant data centers, we just need slightly clever software that can do more with less. didn't take long before that just was overrun with, no, we need horsepower. We really need the ability for big problems to not come up with clever little workarounds, but we need the firepower. And then again, more recently, we've seen a number of open source models out of China that seem to be doing a whole lot more with less.

1:00:36at what point does it begin to become, hey, do we really need a$3 trillion worth of capacity? Don't we just need to take a little bit of that working out of the constraints we have the way the Chinese models have? Yeah, so one of the things that I think is well understood is that the Chinese models didn't do this on their own. Mm-hmm. So the way I like to think of it is, you know, you have like these almost like an animal world, right? I just went on safari to Kenya. And, you know, giraffes almost always have like birds sitting on their necks. And those birds are, you know, it's a mutually symbiotic relationship.

1:01:26I suppose it's not that symbiotic to the giraffe. But, you know, the bird gets to rest on the giraffe's neck and benefits from the fact that the giraffe is walking around. So similarly, I think— What does the giraffe get out of that? I suppose the bird might, like— Keep the bugs away? Keep the bugs away or, like, eat at the ticks on the giraffe. I don't know. But similarly, you know, the Kimi model is a little bit like the bird on the giraffe. Whereas, look, and I think that they are incredibly, like, the Chinese are very innovative in their own right. I think they're very good fast followers.

1:02:06However, they need the giraffe, which is our LLMs, in order to survive. And so, you know, I think there are many different ways to address what is happening. You know, my my the scenario that I think is actually most logical, which I'm not quite sure that, you know, the large language models will do is basically to hold the N and N minus one model internal and allow for, you know, certain businesses, certain companies, the U.S. government, other governments who are not going to distill this model and kind of feed a Kimmy type model. and allow for them use of that model and only make public the N-2 model.

1:02:56And that way it keeps any of the distillation at bay. Now, in order for that to happen, all of the frontier models will have to agree to do this. Because if there's any frontier model that is equivalently as good, then it kind of breaks the ecosystem that I'm describing. But, so I think the point is that you need to spend the CapEx for the training in order to get that output so that Kimmy can train on that output. So these open... Training on the output, not creating their own LLM. Well, Kimmy has created their own LLM by feeding off the, it's called distilling, feeding off the output from the large language models.

1:03:48So, you know, but a lot of the spending that is happening is actually coming from the use of the compute. Like, so from the inference aspect. So you train and then you have to infer. So the inference is what we experience as consumers. And so that inference is driving a majority of the spend. And you look at the revenues of OpenAI, Anthropic, they're kind of like, I've never seen growth like this. I don't think we ever have seen growth that is as significant as what we're seeing today. You know, people frequently make a comparison to the dot coms, and I always feel like that's a terrible comparison because these are real companies with real revenue, real potential profits, like almost there, not clicks and eyeballs.

1:04:41But the one thing some of the skeptics have pointed out that almost resonate is, you know, during the Internet era, we had this huge boom where most of that value ended up landing in the consumers laps, not the investors laps because so many of those companies crashed and burned. How similar or different is this environment to that? So I think it's quite different. Look, there may be parallels at some point, i.e., do we overbuild? And how long does it take to actually eat through that overbuild? So you think about 2000s. One of the reasons we overbuilt is because we had dreamed this dream of what the Internet would be.

1:05:31And we had, you know, Pets.com was actually a brilliant idea. Just a little early. It was just early. Now it's chewy. But Chewy became a significant business. Amazon has built a many trillion dollar business off the back of consumers buying on the internet. But we didn't have the internet. We had dial-up. Dial-up is not good enough to increase productivity back then. What I would argue today is that we actually have the tools, all we needed. We had the internet, we had the productivity, or we had the infrastructure that was needed for ubiquitous intelligence. All we needed was the chips. We needed the data centers and the chips.

1:06:27And that's what is happening today. And so if we actually need ubiquitous intelligence and infinite intelligence, to some extent, like if we overbuild, we will eat through that overbuild as well. So what do you think the skeptics misunderstand about AI? Is it the scale, the economics, how durable the investment cycle is? What are the bears getting wrong here? I think it's the duration. I definitively think I think it's the I think maybe it's all of the above really, but it's duration. It's the scale. It's the economics. All three of those is where I think they're pushing on the wrong on the wrong thread.

1:07:14So last question before I get to all of my favorite questions I ask all my guests. What do you think investors aren't talking about or thinking about that perhaps they should be? What is getting overlooked here? And it could be any asset, geography, policy, whatever. But what aren't people talking about but should? Yeah. I think that people aren't really talking about the net positive benefits to humanity from AI. And we talked about healthcare and how we can make healthcare kind of available to any human on this planet. The same goes for education. There's no reason why any child should be, quote unquote, left behind.

1:08:06I mean, I've been shocked at, you know, what I've been reading recently is kids going to college and they can't read. Right. That is a failure of our education system that can be solved using artificial intelligence. You know, this is and this is again, it's a global issue. It is not a local issue. This is something that we can, there's no one that should not be educated. And like the AI kind of, the anti-AIers are climate change, right? I mean, I do think that using AI, will we be able to solve the problems that we have with climate change? Will we be able to engineer things that will help with the rapid rate of climate change?

1:08:57and a lot of the AI kind of doomers or AI naysayers who don't want the data center built in their backyard or data center built anywhere are ignoring the fact that there are many different aspects of AI that will be good for humanity. And does it require great change and is change scary? It is and it will require change. It will require change from all of us. But the end point is actually quite beautiful. I like that you're such a techno optimist. All right, let's jump to our favorite questions, starting with who are your mentors who helped shape your career? Oh, gosh, I think that's a pretty easy one.

1:09:42Our CEO, Dan Chung, has been pivotal in my career and my career growth. And, you know, I told you, like, he hired me from Stanford without knowing anything. And without, I really knew nothing about this business. And he recognized that, you know, why not take a shot on someone who's nontraditional? And he himself is a nontraditional thinker. He was a lawyer and, you know, thinks very much outside of the box. So, you know, over the years, he's challenged me in ways that have been sometimes frustrating, but I learn from. He pushes me in ways that sometimes I don't understand. But again, I learn from and grow from.

1:10:35So, yeah, I think Dan is, Dan's like my number one mentor. Let's talk about books. What are some of your favorites? What are you reading currently? So my favorite book is a book called Think Again. It's by Adam Grant. Oh, of course. Who is an organizational psychologist. And I know it's an odd... I think he was at Harvard. Wharton, I think. And I know it's an odd book to be a favorite book of mine, but in context of business, it definitively is. And in part, it's because it talks about how you can have a hypothesis, but you have to be humble enough to understand that you can also change your hypothesis.

1:11:22But you have to have the confidence enough to hold a hypothesis. And really, intelligence is about the ability to morph and be nimble. And it's not about arrogance. It's not about our business requires a constant questioning of what you think. right and those that become very uh tied to a thesis end up i think on the wrong side of of a lot of trades and so i just loved the book because the way he writes um about intelligence and the humility of questioning and of of holding conversations of with people and i think this is true for society in general right now of having conversations where you may not you may not agree but to hear other people out even if they don't agree with you and any of the books anything you're reading currently um the last book i read was uh the the the recent one by brad jacobs which was how to make a few more billions brad jacobs is the ceo of qxo and he wrote his first book, How to Make a Few Billion, and then he wrote How to Make a Few More Billions.

1:12:42And what I thought was so interesting about the book is the first two chapters is about how he centers himself. And he's an incredibly successful entrepreneur, has built many businesses, really from scratch. He's a self-made billionaire. And he starts every morning meditating, right? and how he finds that center. And it just like, to me, we often don't talk about that aspect of investing in business. It feels sometimes really transactional. But hearing that aspect of Brad, it only puts him in even higher regard for me. Really interesting. What are you streaming these days? What are you either listening to or watching?

1:13:33you? Oh, gosh, I don't, I don't watch much. I don't have that much time. And usually when I do watch something with my kids, I fall asleep. So but, but I am a runner. And so I have a lot of time that I spend running. And I'm constantly listening to podcasts. My favorite ones happen to be macro voices. I love the knowledge project. Oh, Shane Pash? Yeah, yeah, he's, That's a regular on Sunday mornings for me. Yeah. And so, I mean, the variety of conversations that he has with different people from wellness and well-being. And I was listening to one about the Alpha School and how education should be reshaped.

1:14:18There's just like awesome amount of diversity of thought. The circuit, which is all about semiconductors and chips. I kind of think of other ones that I listen to regularly. That's all that comes to mind. That's a nice list to start with. Our final two questions. What sort of advice would you give to a recent college grad interested in a career in either engineering, material science, or investing? Oh, wow. Well, look, I think for any college grad, make sure that you do something that you love. Right. And it doesn't have to be you love it every day, but you spend a lot of your time at work. Your life, a third, more than a third of your life is going to be spent from here on out at work.

1:15:12make sure that you do something that you believe in that gives you great gratification that you feel like you're contributing to society um don't just do it because you're on a treadmill of you know i'm gonna go do this because i set out to do this path and you know i just have to go trotting along. Allow yourself the grace to change and to change your mind. I did. And it was probably the best risk that I ever took. So the best gamble that I ever took on was completely pivoting in my career. So, you know, allow yourself to explore because you change over time as well. You know, what you want today may be different what you want in five and 10 years, but but definitively make sure that you love what you do because once you know that you love what you do, you will be the best at it.

1:16:11Get essential news on the people and companies pushing the tech sector to new frontiers. Hi, I'm Ed Ludlow. Join me for Bloomberg Tech, a daily podcast focused exclusively on technology, innovation and the future of business. Every weekday, we bring you the latest insights on Silicon Valley's top companies and conversations with tech's biggest decision makers. Listen to Bloomberg Tech on your commute home and stay ahead of the news cycle. Subscribe today on Apple, Spotify or anywhere you listen. And our final question. What do you know about the world of investing today might have been useful 20 or so years ago when you were first starting out?

1:16:53So you told me this question would stump me and it is stumped me. Because, you know, the answers that I'm not looking for are, you know, buy Amazon in 2002 when it was$7. Yeah. It's what insight might have been useful way back when. What have you learned? What expensive lessons came along that, you know, I could have saved myself a lot of headache had I figured this out sooner? You know what, Barry? I don't think I would, in my way back machine, I wouldn't go tell myself anything. So it's the path and not necessarily. Yeah, it's the journey. Like my most painful moments as an investor have been the biggest learning moments for me that have had like, like they've branded me in some way with that experience.

1:17:46And so I wouldn't want to shortcut that because it has shaped me. And every single time I've fallen on my face, it has shaped me. And it has reminded me of, you know, the perils of not paying attention to X, Y, or Z. Or I won't make that same mistake again. Because, again, it is so like, it goes back to that first question you asked me about like academic versus learning on the job. You need the real experience. You need the scars. You need the scars. And it's a little bit like your kids, right? You can tell your kids, don't do that. You're going to get hurt. Don't do that. You're going to get hurt.

1:18:30Well, sometimes they just have to fall down and get hurt to realize they're going to get hurt. So it makes a ton of sense. Ankur, thank you so much for being so generous with your time. We have been speaking with Ankur Crawford, portfolio manager at Alger. If you enjoy this conversation, well, be sure and check out any of the 653 we've done over the past. 12 years. We launched July 2014. You can find those at iTunes, Spotify, YouTube, Bloomberg, wherever you find your favorite podcasts. I would be remiss if I didn't thank the crack team that helps put these conversations together each week. Alexis Noriega is my video producer.

1:19:12Sean Russo is my researcher. Anna Luke is my producer. I'm Barry Ritholtz. You've been listening to Masters in Business on Bloomberg Radio.

1:19:26Hi, I'm Tom Keen, inviting you to join me for the Bloomberg Surveillance Podcast. It's about making you smarter every business day. I'm Paul Sweeney. We bring you complete coverage of stocks, bonds, commodities, even crypto. All the information you need to excel in the markets. And I'm Alexis Christophorus. Listen to us for essential conversations with the smartest names in economics, finance, investment, and international relations. That's the Bloomberg Surveillance Podcast. Subscribe today on Apple, Spotify, or anywhere you listen.