Episode #497: Ulrike Hoffmann-Burchardi, Tudor Investments – AI, Digital, Data & Disruptive Innovation

Episode #497: Ulrike Hoffmann-Burchardi, Tudor Investments – AI, Digital, Data & Disruptive Innovation

Guest: Ulrike Hoffmann-Burchardi is a Portfolio Manager at Tudor Investment Corporation where she oversees a global equity portfolio inside Tudor’s flagship fund focusing on Digital, Data & Disruptive Innovation.

Recorded: 8/17/2023  |  Run-Time: 44:23

---

Summary: In today’s episode, she starts by lessons learned over the past 25 years working at a famed shop like Tudor. Then we dive into topics everyone is talking about today: data, AI, large language models. She shares how she sees investment teams incorporating AI and LLMs into their investing process in the future, her view of the macro landscape, and finally what areas of the market she likes today.

---

Sponsor: Future Proof, The World’s Largest Wealth Festival, is coming back to Huntington Beach on September 10-13th! Over 3,000 finance professionals and every relevant company in fintech, asset management and wealth management will be there. It’s the one event that every wealth management professional must attend!

---

Comments or suggestions? Interested in sponsoring an episode? Email us Feedback@TheMebFaberShow.com

Links from the Episode:

• 0:00 – Welcome Ulrike to the show
• 0:33 – Learning the value of micro and macro perspectives during her 25 years at Tudor
• 8:04 – How large language models may eclipse the internet, impacting society and investments
• 10:18 – AI’s impact on investment firms, and how it’s creating investment opportunities
• 13:19 – Public vs. private opportunities
• 19:21 – Macro and micro aligned in H1, but now cautious due to growth slowdown
• 24:04 – Trust is crucial in AI’s use of data, requiring transparency, ethics, and guardrails
• 26:53 – The importance of balancing macro and micro perspectives
• 33:47 – Ulrike’s most memorable investment opportunity
• 37:43 – Generative AI’s power for both existential risks and climate solutions excites and concerns
• Learn more about Ulrike: Tudor; LinkedIn

 

Transcript:

Welcome Message:

Welcome to The Meb Faber Show, where the focus is on helping you grow and preserve your wealth. Join us as we discuss the craft of investing and uncover new and profitable ideas, all to help you grow wealthier and wiser. Better investing starts here.

Disclaimer:

Meb Faber is the Co-founder and Chief Investment Officer at Cambria Investment Management. Due to industry regulations, he will not discuss any of Cambria’s funds on this podcast. All opinions expressed by podcast participants are solely their own opinions and do not reflect the opinion of Cambria Investment Management or its affiliates. For more information, visit cambriainvestments.com.

Meb:

Welcome, podcast listeners. We have a special episode today. Our guest is Ulrike Hoffmann-Burchardi, a Portfolio Manager at Tudor Investment Corporation, where she oversees a global equity portfolio inside Tudor’s flagship fund. Her area of focus is around digital, data, and disruptive innovation. Barron’s named her as one of the 100 most influential women in finance this year. In today’s episode, she starts by lessons learned over the past 25 years working at a fame shop like Tudor. Then we dive into topics everyone is talking about today, data AI, large language models. She shares how she sees investment teams incorporating AI and LLMs into their investing process in the future, her view of the macro landscape, and finally what areas of the market she likes today. With all the AI hype going on, there couldn’t have been a better time to have her on the show. Please enjoy this episode with Ulrike Hoffmann-Burchardi.

Meb:

Ulrike, welcome to the show.

Ulrike:

Thank you. Thank you for inviting me.

Meb:

Where do we find you today?

Ulrike:

New York City.

Meb:

What’s the vibe like? I just went back recently, and I joke with my friends, I said, “It seemed pretty vibrant. It smelled a little different. It smells a little bit like Venice Beach, California now.” But other than that, it feels like the city’s humming again. Is that the case? Give us a on the boots review.

Ulrike:

It is. And actually our offices are in Astor Place, so very close to the Silicon Alley of Manhattan. It couldn’t be more vibrant.

Meb:

Yeah, fun. I love it. This summer, a little warm but creeping up on fall time, my favorite. All right, so we’re going to talk all sorts of different stuff today. This generation, I feel like it’s my dad, mom, full career, one place. This generation, I feel like it’s like every two years somebody switches jobs. You’ve been at one company this entire time, is that right? Are you a one and doner?

Ulrike:

Yeah, it’s hard to believe that I’m in year 25 of investing as a career, and I’ve been fortunate, as you say, to have been with the same company for this period of time and also fortunate for having been in that company in many different investing capacities. So maybe a little bit like Odyssey, at least structurally, multiple books within a book.

Meb:

I was joking the other day where I feel like a more traditional path. You see so many successful value managers, like equity managers who do fantastic in the equity world for a number of years, and then they start to drift into macro. I say it’s almost like an impossible magnet to avoid where they start talking about gold and the Fed and all these other things that are like politics and geopolitics. And very rarely do you see the progression you’ve had, which is almost everything, but also macro moving towards equities. You’ve covered it all. What’s left? Short selling and I don’t know what else. Are you guys do a little shorting actually?

Ulrike:

Yeah, we call it hedging as it actually gives you staying power for your long-term investments.

Meb:

Hedging is a better way to say it.

Ulrike:

And yes, you’re right. It’s been a somewhat unique journey. In a sense, book one for me was macro investing, then global asset allocation, then quant equity. And then finally over the last 14 years, I’ve been lucky to forge my own way as a fundamental equity investor and that all inside a firm with this exceptional macro and quantitative band. It’s been terrific to have had these different types of exposures. I think it taught me the value of different perspectives.

There’s this one famous quote by Alan Kay who said that perspective is worth more than 80 IQ points. And I think for equity investing, it’s double that. And the reason for that is, if you look at stocks with perfect hindsight and you ask yourself what has actually driven stock returns and can do that by decomposing stock returns with a multifactor model, you find that 50% of returns are idiosyncratic, so things that are company specific related to the management teams and also the objectives that they set out to achieve, then 35% is determined by the market, 10% by industry and actually only 5% is everything else, including style factors. And so for an equity investor, you need to understand all these different angles. You need to understand the company, the management team, the industry demand drivers, and what’s the regulatory backdrop. And then finally, the macro picture.

And maybe the one arc of this all, and also maybe the arc of my professional career, is the S&P 500. Believe it or not, but my journey at Tutor actually started out with a forecasting model for the S&P 500, predicting the S&P one week and also one month ahead when I joined tutor in 1999. And predicting S&P is still frankly key to what I’m doing today when I try to figure out what beta to run in the various equity portfolios. So I guess it was my first task and will probably be my forever endeavor.

Meb:

If you look back at that time, the famous joke the media loves to run with is what butter in Bangladesh or something like that. Things that are most, like the famous paper was like what’s most correlated with S&P returns? Is there anything you remember in particular either A, that worked or didn’t work or B, that you thought worked at the time that didn’t work out of sample or 20 years later?

Ulrike:

Yes, that’s such a great question Meb, correlation versus causation. You bring me right back to the lunch table conversations with my quant colleagues back in the early days. One of my former colleagues actually wrote his PhD thesis on this very topic. The way we tried to prevent over fitting in our models back then was to start out with a thesis that’s anchored in economic theory. So rates should impact equity prices and then we would see whether those actually are statistically important. So all these forecasting models for the S&P 500 or predicting the prices of a thousand stocks were very much purpose-built. Thesis, variables, data, and then we would take these and see which variables actually mattered. And this whole chapter of classical statistical AI is all about human control. The chance of these models going rogue is very small. So I can tell you butter production in Bangladesh did not make it into any of our models back then.

But the other lesson I learned during this time is to be wary of crowding. You may remember 2007, and for me the biggest lesson learned from the quant crisis is to be early and to be convicted. When your thesis floods your inbox, then it’s time to make your way to the exit. And that’s not only the case for stocks, but also for strategies, because crowding is especially an issue when the exit door is small and when you have too much money flowing into a fixed sized market opportunity, it just never ends well. I can tell you from firsthand experience as I lived right through this quant unwind in August 2007.

And thereafter, as a reminder of this crowding risk, I used to have this chart from Andrew Lo’s paper on the quant crisis pinned to my office wall. Those were the analog times back then with printouts and pin boards. The chart showed two things. It showed on the one hand the fund inflows into quant equity market neutral over the prior 10 years, and it showed something like zero to a hundred funds with ultimately over a hundred billion in AUM at the very end in 2007. And then secondly, it showed the chart with declining returns over the same period, still positive, but declining. So what a lot of funds did during this time was say, “Hey, if I just increase the leverage, I can still get to the same type of returns.” And again, that’s never a recipe for much success because what we saw is that almost all of these strategies lost within a few days the amount of P&L that they had made over the prior year and more.

And so for me, the big lesson was that there are two signs. One is that you have very persistent and even sometimes accelerating inflows into certain areas and at the same time declining returns, that’s a time when you want to be cautious and you want to wait for better entry points.

Meb:

There’s like five different ways we could go down this path. So you entered around the same time I did, I think, if you were talking about 99 was a pretty crazy time in markets obviously. But when is it not a crazy time in markets? You’ve seen a few different zigs and zags at this point, the global financial crisis, the BRICs, the COVID meme stock, whatever you want to call this most recent one. What’s the world like today? Is it still a pretty interesting time for investing or you got it all figured out or what’s the world look like as a good time to talk about investing now?

Ulrike:

I actually think it could not be a more interesting time right now. We are in such a maelstrom of different currents. We’ve seen the fastest increase in rates since 1980. The Fed fund rate is up over 5% in just a little over a year. And then we’ve seen the fastest technology adoption ever with ChatGPT. And you’re right that there’s some similarities to 99. ChatGPT is in a lot of ways for AI what Netscape was for the internet back then.  And then all at the same time right now, we are facing an existential climate challenge that we need to solve sooner rather than later. So frankly, I cannot think about a time with more disruption over the last 25 years. And the other side of disruption of course is opportunity. So lots to talk about.

Meb:

I see a lot of the AI startups and everything, but I haven’t got past using ChatGPT to do anything other than write jokes. Have you integrated into your daily life yet? I have a friend whose entire company’s workflow is now ChatGPT. Have you been able to get any daily utility out of yet or still playing around?

Ulrike:

Yes. I would say that we are still experimenting. It will definitely have an impact on the investing process though over time. Maybe let me start with why I think large language models are such a watershed moment. Unlike any other invention, they’re about developing an operating system that is superior to our biological one, that’s superior to our human brain. They share similar features of the human brain. They’re both stochastic and they’re semantic, but they have the potential to be much more powerful. I mean, if you think about it, large language models can learn from more and more data. Llama 2 was trained on 2 trillion tokens. It’s about a trillion words and the human brain is only exposed to about 1 billion words during our lifetime. So that’s a thousand times less information. And then large language models will have more and more parameters to understand the world.

GPT4 is rumored to have close to 2 trillion parameters. And, of course, that’s all possible because AI compute increases with more and more powerful GPUs and our human compute peaks at the age of 18.

And then the improvements are so, so rapid. The number of academic papers that have come out since the launch of ChatGPT have frankly been difficult to keep up with. They range from prompt engineering, there was the Reflexion paper early in the year, the Google ReAct framework, and then to completely new fundamental approaches like the Retentive architecture that claims to have even better predictive power and also be more efficient. So I think large language models are a foundational innovation unlike anything we’ve seen before and it’ll eclipse the internet by orders of magnitude. It’ll have societal implications, geopolitical implications, investment implications, and all on the scale that we have not seen before.

Meb:

Are you starting to see this have implications in our world? If so, from two seats, there’s the seat of the investor side, but also the investment opportunity set. What’s that look like to you? Is it like 1995 of the internet or 1990 or is it accelerating much quicker than that?

Ulrike:

Yes, it’s for sure accelerating faster than prior technologies. I think ChatGPT has broken all adoption records with 1 million users within five days. And yes, I also think we had an inflection point with this new technology when it suddenly becomes easily usable, which often happens many years after the initial invention. IBM invented the PC in 81, yet it was Windows, the graphical user interface in 85 that made PCs easily usable. And the transformer model dates back to 2017 and now ChatGPT made it so popular.

And then like you say, there are two things to think about. One is the how and then the what. How is it going to change the future of investment firms and what does it mean for investing opportunities? I think AI will have an impact on all industry. It targets white collar jobs in the very same way that the industrial revolution did blue collar work.

And I think that means for this next stage that we’ll see more and more intelligent agents in our personal and our professional lives and we’ll rely more on those to make decisions. And then over time these agents will act more and more autonomously. And so what this means for institutions is that their knowledge base will be more and more tied to the intelligence of these agents. And in the investing world like we are both in, this means that in the first stage building AI analysts, analysts that perform different tasks, research tasks with domain knowledge and technology and healthcare and climate and so on. And then there’ll be a meta layer, an investor AI and a risk manage AI. And those translate insights from research AIs into a portfolio of investments. That’s clearly the journey we are on. Obviously we are in the early beginnings of this, but I think it’ll profoundly affect the way that investment firms are being run.

And then you ask about the investment opportunity set and the way I look at AI. I think AI will be the dividing line between winners and losers, whether it’s for companies, for investors, for nations, maybe for species.

And when I think about investing opportunities, there’ve been many times when I look with envy to the private markets, especially in those early days of software as a service. But I think now is a time where public companies are so much more exciting. We have a moment of such high uncertainty where the best investments are often the picks and shovels, the tools that are needed no matter who succeeds in this next wave of AI applications.

And those are semiconductors as just one example in particular, GPUs and also interconnects. And then secondly, cloud infrastructure. And most of these companies now are public companies. And then when you think about the application layer where we’ll likely see lots of new and exciting companies, there’s still a lot of uncertainty. Will the next version of GPT make a new startup obsolete? I mean, it could turn out that just the new feature of GPT5 will completely subsume your business model like we’ve already seen with some startups. And then how many base large language models will there really need to be and how will you monetize those?

Meb:

You dropped a few mic drops in there very quietly, talking about species in there as well as other things. But I thought the comment between private and public was particularly interesting because usually I feel like the assumption of most investors is a lot of the innovation happens in the Silicon Valley garage or it’s the private startups on the forefront of technology. But you got to remember that the Googles of the world have a massive, massive war chest of both resources and cash, but also a ton of thousands and thousands of very smart people. Talk to us a little bit about the public opportunities a little more. Expand a little more on why you think that’s a good place to fish or there’s the innovation going on there as well.

Ulrike:

I think it’s just the stage we’re in where the picks and shovels happen to be in the public markets. And it’s the application layer that’s likely to come out of the private markets, and it’s just a little early to tell who is going to be the winner there, especially as these models are becoming so much more powerful and domain specific. It’s not clear for example, if you say have a specific large language model for lawyers, I guess an LLM for LLMs, whether that’s going to be more powerful than the next version of GPT5, once all the legal cases have been fed into the model.

So maybe another way to think about the winners and losers is to think about the relative scarcity value that companies are going to have in the future. And one of the superpowers of generative AI is writing code. So I think there’ll be an abundance of new software that is generated by AI and the physical world just cannot scale that easily to keep up with all this processing power that’s needed to generate this code. So again, I think the physical world, semiconductors, will likely become scarcer than software over time, and that opportunity set is more in the public markets than the private markets right now.

Meb:

How much of this is a winner take all? Someone was talking to me the other day and I was trying to wrap my head around the AI opportunity with a reflexive coding or where it starts to build upon itself and was trying to think of these exponential outcomes where if one dataset or AI company is just that much better than the others, it quickly becomes not just a little bit better, but 10 or a hundred times better. I feel like in the history of free markets you do have the massive winners that often end up a little monopolistic, but is that a scenario you think is plausible, probable, not very likely. What is the more likely path of this creative destruction between these companies? I know we’re in the early days, but what do you look out to the horizon a little bit?

Ulrike:

I think you’re right that there are probably only going to be a few winners in each industry. You need three things to be successful. You need data, you can need AI expertise, and then you need domain knowledge of the industry that you are operating in. And companies who have all three will compound their strength. They’ll have this positive feedback loop of more and more information, more learning, and then the ability to provide better solutions. And then on the large language models, I think we’re also only going to see a few winners. There’re so many companies right now that are trying to design these new foundational models, but they’ll probably only end up with one or two or maybe three that are going to be relevant.

Meb:

How do you stay abreast of all this? Is it mostly listening to what the companies are putting out? Is it sell side research? Is it conferences? Is it academic papers? Is it just chatting with your network of friends? Is it all the above? In a super-fast changing space, what’s the best way to keep up with everything going on?

Ulrike:

Yes, it’s all of the above, academic papers, industry events, blogs. Maybe one way we are a little different is that we are users of many of the technologies that we invest in. Peter Lynch use to say invest in what you know. I think it’s relatively straightforward on the consumer side. It’s a little bit trickier on the enterprise side, especially for data and AI. And I’m lucky to work with a team that has skills in AI, in engineering and in data science. And for the majority of my career, our team has used some form of statistical AI to help our investment decisions and that can lead to early insights, but also insights with higher conviction.

There are many examples, but maybe in this recent case of large language model, it’s realizing that large language models based on the Transformer architecture need parallel compute both for inference and for training and realizing that this would usher in a new age of parallel compute, very much like deep learning did in 2014. So I do think being a user of the technologies that you invest in gives you a leg up in understanding the fast moving environment we are in.

Meb:

Is this a US only story? I talked to so many friends who obviously the S&P has stomped everything in sight for the past, what is it, 15 years now. I think the assumption when I talk to a lot of investors is that the US tech is the only game in town. As you look beyond our borders, are there other geographies that are having success either on the picks and shovels, whether it’s a semiconductors areas as well, because in general it seems like the multiples often are quite a bit cheaper outside our shores because of various concerns. What’s the perspective there? Is this a US only story?

Ulrike:

It’s mainly a US story. There are some semiconductor companies in Europe and also Asia that are going to profit from this AI wave. But for the core picks and shovels, they’re very US centric.

Meb:

Okay. You talk about your role now and if you rewind, going back to the skillset that you’ve learned over the past couple of decades, how much of that gets to inform what’s going on now? And part of this could be mandate and part of it could be if you were just left to your own designs, you could incorporate more of the macro or some of the ideas there. And you mentioned some of what’s transpiring in the rest of the year on interest rates and other things. Is it mostly driven company specific at this point or are you in the back of your mind saying, “Oh no, we need to adjust maybe our net exposure based on these variables and what’s going on in the world?” How do you put those two together or do you? Do you just separate them and move on?

Ulrike:

Yes, I look at both the macro and the micro to figure out net and gross exposures. And if you look at the first half of this year, both macro and micro were very much aligned. On the macro side we had a lot of room for offside surprises. The market expected positive real GDP growth of close to 2%, yet earnings were expected to shrink by 7% year over year. And then at the same time on the micro side, we had this inflection point which generative AI as this new foundational technology with such productivity promise. So a very bullish backdrop on both fronts. So it’s a good time to run high nets and grosses. And now if we look at the back half of the year, the micro and the macro don’t look quite as rosy.

On the macro side, I expect GDP growth to slow. I think the weight of interest rates will be felt by the economy eventually. It’s a little bit like the damage accumulation effect in wood. Wood can withstand relatively heavy load in the short term, but it will get weaker over time and we have seen cracks. Silicon Valley Bank is one example. And then on AI, I think we may overestimate the growth rate in the very short term. Don’t get me wrong, I think AI is the biggest and most exponential technology we have seen, but we may overestimate the speed at which we can translate these models into reliable applications that are ready for the enterprise. We are now in this state of excitement where everybody wants to build or at least experiment with these large language models, but it turns out it’s actually pretty difficult. And I would estimate that they’re only around a thousand people in the world with this particular skillset. So with the risk of a longer wait for enterprise ready AI and a more challenging macro, it seems now it’s time for lower nets and gross exposure.

Meb:

We talk about our industry in general, which when I think of it is one of the highest margin industries being asset management. There’s the old Jeff Bezos phrase that he loves to say, which is like “Your margin is my opportunity.” And so it’s funny because in the US there’s been this massive amount of competition, thousands, 10,000 plus funds, everyone entering the terradome with Vanguard and the death star of BlackRock and all these giant trillion dollar AUM companies. What does AI mean here? Is this going to be a pretty big disruptor from our business side? Are there going to be the haves and have-nots that have adopted this or is it going to be a nothing burger?

Ulrike:

The dividing line is going to be AI for everyone. You need to augment your own intelligence and bandwidth with these tools to remain competitive. This is true as much for the tech industries as it is for the non-tech industries. I think it has the potential to reshuffle leadership in all verticals, including asset management, and there you can use AI to better tailor your investments to your clients to communicate better and more frequently.

Meb:

Well, I’m ready for MEB2000 or MebGPT. It seems like we asked some questions already. I’m ready for the assistant. Honestly, I think I could use it.

Ulrike:

Yes, it will pre generate the perfect questions ahead of time. It still needs your gravitas though, Meb.

Meb:

If I had to do a word cloud of your writings and speeches over the years, I feel like the number one word that probably is going to stick out is going to be data, right? Data has always been a big input and forefront on what you’re talking about. And data is at the center of all this. And I think back to daily, all the hundred emails I get and I’m like, “Where did these people get my information?” Thinking about consent and how this world evolves and you think a lot about this, are there any general things that are on your brain that you’re excited or worry about as we start to think about kind of data and its implications in this world where it’s sort of ubiquitous everywhere?

Ulrike:

I think the most important factor is trust. You want to trust that your data is treated in a confidential way in line with rules and regulations. And I think it’s the same with AI. The biggest factor and imperative going forward is trust and transparency. We need to understand what data inputs these models are learning from, and we need to understand how they’re learning. What is considered good and what is considered bad. In a way, training these large language models is a bit like raising children. It depends on what you expose them to. That’s the data. If you expose them to things that are not so good, that’s going to affect their psyche. And then there is what you teach your kids. Don’t do this, do more of that, and that’s reinforcement learning. And then finally, guardrails. When you tell them that there are certain things that are off limits. And, companies should be open about how they approach all three of these layers and what values guide them.

Meb:

Do you have any thoughts generally about how we just volunteer out our information if that’s more of a good thing or should we should be a little more buttoned down about it?

Ulrike:

I think it comes down again to trust. Do you trust the party that you’re sharing the information with? Certain companies, you probably do so and others you’re like, “Hmm, I’m not so sure.” It’s probably the most valuable assets that companies are going to build over time and it compounds in very strong ways. The more information you share with the company, the more data they have to get insights and come up with better and more personalized offerings. I think that’s the one thing companies should never compromise on, their data promises. In a sense, trust and reputation are very similar. Both take years to build and can take seconds to lose.

Meb:

How do we think about, again, you’ve been through the same cycles I have and sometimes there’s some pretty gut-wrenching drawdowns in the beta markets, S&P, even just in the past 20 years, it’s had a couple of times been cut in half. REITs went down, I don’t know, 70% in the financial crisis, industries and sectors, even more. You guys do some hedging. Is there any general best practices or ways to think about that for most investors that don’t want to watch their AI portfolio go down 90% at some point if the world gets a little upside down. Is it thinking about hedging with indexes, not at all companies? How do you guys think about it?

Ulrike:

Yeah. Actually in our case, we use both indices and custom baskets, but I think the most important way to avoid drawdowns is to try to avoid blind spots when you are either missing the micro or the macro perspective. And if you look at this year, the biggest macro drivers were in fact micro: Silicon Valley Bank and AI. In 2022, it was the opposite. The biggest stock driver was macro, rising interest rates since Powell’s pivot in November 2021. So being able to see the micro and the macro perspectives as an investment firm or as an investment team gives you a shot at capturing both the upside and protecting your downside.

But I think actually this cognitive diversity is key, not just in investing. When we ask the CEOs of our portfolio companies what we can be most helpful with as investors, the answer I’ve been most impressed with is when one of them said, help me avoid blind spots. And that actually prompted us to write research purpose-built for our portfolio companies about macro industry trends, benchmark, so perspectives that you are not necessarily aware of as a CEO when you’re focused on running your company. I think being purposeful about this cognitive diversity is key to success for all teams, especially when things are changing as rapidly as they’re right now.

Meb:

That’s a good CEO because I feel like half the time you talk to CEOs and they surround themselves by yes people. They get to be very successful, very wealthy, king of the castle sort of situation, and they don’t want to hear descending opinions. So you got some golden CEOs if they’re actually thinking about, “Hey, I actually want to hear about what the threats are and what are we doing wrong or missing?” That’s a great hold onto those, for sure.

Ulrike:

It’s the sign of those CEOs having a growth mindset, which by the way, I think is the other factor that is the most relevant in this world of change, whether you’re an investor or whether you’re a leader of an organization. Change is inevitable, but growing or growth is a choice. And that is the one leadership skill that I think ultimately is the biggest determinant for success. Satya Nadella, the CEO of Microsoft is one of the biggest advocates of this growth mindset or this no regret mindset, how he calls it. And I think the Microsoft success story in itself is a reflection of that.

Meb:

That’s easy to say, so give us a little more depth on that, “All my friends have an open mind” quote. Then you start talking about religion, politics, COVID vaccines, whatever it is, and then it’s just forget it. Our own personal blinders of our own personal experiences are very huge inputs on how we think about the world. So how do you actually try to put that into practice? Because it’s hard. It’s really hard to not get the emotions creep in on what we think.

Ulrike:

Yeah, maybe one way at least to try to keep your emotions in check is to list all the potential risk factors and then assess them as time goes by. And there are certainly a lot of them to keep track of right now. I would not be surprised if any one of them or a combination could lead to an equity market correction in the next three to six months.

First off, looking at AI, we spoke about it. There’s a potential for a reset in expectations on the speed of adoption, the speed of enterprise adoption of large language models. And this is important as seven AI stocks have been responsible for two thirds of the S&P gains this year.

And then on the macro side, there’s less potential for positive earnings surprises with more muted GDP growth. But then there are also plenty of other risk factors. We have the budget negotiations, the possible government shutdown, and also we’ve seen higher energy prices over the last few weeks that again could lead to a rise in inflation. And those are all things that cloud the macro picture a little bit more than in the first part of the year.

And then there’s still a ton of excess to work through from the post COVID period. It was a pretty crazy environment. I mean, of course crazy things happen when you try to divide by zero, and that’s exactly what happened in 2020 and 2021. The opportunity cost of capital was zero and risk looked extremely attractive. So in 2021, I believe we had a thousand IPOs, which was five times the average volume, and it was very similar on the private side. I think we had something like 20,000 private deals. And I think a lot of these investments are likely not going to be profitable in this new interest rate environment. So we have this lost generation of companies that were funded in 2020 and 2021 that will likely struggle to raise new capital. And many of these companies, especially zombie companies with little cash, but a high cash burn are now starting to go out of business or they’re sold at meaningfully lower valuations. Actually, your colleague Colby and I were just talking about one company that is a virtual events’ platform that was valued at something like $7.8 billion in July 2021 and just sold for $15 million a few weeks ago. That’s a 99.9% write down. And I think we’ll see more of these companies going this way. And this will not only have a wealth effect, but also impact employment.

And then lastly, I think there could be more accidents in the shadow banking system. If you wanted to outperform in a zero-rate environment, you had to go all in. And that was either with investments in illiquids or long duration investments. Silicon Valley Bank, First Republic, Signature Bank, they all had very similar asset liability mismatches. So there is a risk that we’ll see other accidents in the less regulated part of banking. I don’t think we’ll see anything like what we’ve seen in the great financial crisis because banks are so regulated right now. There’s no systemic risk. But it could be in the shadow banking system and it could be related to underperforming investments into office real estate, into private credit or private equity.

So I think the excitement around generative AI and also low earnings expectations have sprinkled this fairy dust on an underlying challenging economic backdrop. And so I think it’s important to remain vigilant about what could change this shiny picture.

Meb:

What’s been your most memorable investment back over the years? I imagine there’s thousands. This could be personally, it could be professionally, it could be good, it could be bad, it could just be whatever’s seared into your frontal lobe. Anything come to mind?

Ulrike:

Yeah. Let me talk about the most memorable investing opportunity for me, and that was Nvidia in 2015.

Meb:

And a long time ago.

Ulrike:

Yeah, a long time ago, eight years ago. Actually a little over eight years ago, and I remember it was June 2015 and I got invited by Delphi Automotive, which at the time was the largest automotive supplier to a self-driving event on the West Coast. After reverse commuting from New York to Connecticut for close to 10 years as a not very proficient driver, autonomous driving sounded just like utter bliss to me. And, in fact, I could not have been more excited than after this autonomous drive with an Audi Q5. It carried the full stack of self-driving equipment, camera, lidar, radar. And it quickly became clear to me that even back then, when we were driving both through downtown Palo Alto and also on Highway 101, that autonomous was clearly way better than my own driving had ever been.

I’m just mentioning this particular point in time because we at a very similar point with large language models, ChatGPT is a little bit like the Audi Q5, the self-driving prototype in 2015. We can clearly see where the journey is going, but the question is who are going to be the winners and losers along the way?

And so after the drive, there was this panel on autonomous driving with folks from three companies. I remember it was VW, it was Delphi, and it was Nvidia. And as you may remember, up to that point, Nvidia was mainly known for graphic cards for video games, and it had just started to be used for AI workloads, especially for deep learning and image recognition.

In a way, it’s a neat way to think about investing innovation more broadly because you have these three companies, VW, the producer of cars, the application layer, then you have Delphi, the automotive supplier, sort of middleware layer, and then Nvidia again, the picks and shovels. You need, of course GPUs for computer vision to process all the petabytes of video data that these cameras are capturing. So they represented different ways of investing in innovation. And just wondering, Meb, who do you think was the clear winner?

Meb:

I mean, if you had to wait till today, I’ll take Nvidia, but if I don’t know what the inner period would’ve been, that’s a long time. What’s the answer?

Ulrike:

Yes, you’re right. The clear standout is Nvidia. It’s up more than 80 times since June 2015. VW is actually down since then. In that category it’s been Tesla who has been the clear winner actually, somebody more in the periphery back then. But of course Tesla is now up 15 times since then and Delphi has morphed into different entities, probably slightly up if you adjust for the different transitions. So I think it shows that often the best risk reward investments are the enablers that are needed to innovate no matter what. They’re needed both by the incumbents but also by the new entrants. And that’s especially true when you’re early in the innovation curve.

Meb:

As you look out to the horizon, it’s hard to say 2024, 2025, anything you’re particularly excited or worried about that we skipped over.

Ulrike:

Yeah. Something that we maybe didn’t touch on is that something as powerful as GenAI clearly also bears existential risks, but equally its power may be key to solving another existential risk, which is climate. And there we need non the nonlinear breakthroughs, and we need them soon, whether it’s with nuclear fusion or with carbon capture.

Meb:

Now, I got a really hard question. How does the Odyssey end? Do you remember that you’ve been through paralleling your career with the book? Do you recall from a high school college level, financial lit 101? How does it end?

Ulrike:

Does it ever end?

Meb:

Thanks so much for joining us today.

Ulrike:

Thank you, Meb. I really appreciate it. It’s probably a good time for our disclaimer that Tudor may hold positions in the companies that we mentioned during our conversation.

Meb:

Podcast listeners will post show notes to today’s conversation at mebfaber.com/podcast. If you love the show, if you hate it, shoot us feedback at feedback@themebfabershow.com. We love to read the reviews. Please review us on iTunes and subscribe the show anywhere good podcasts are found. Thanks for listening, friends, and good investing.