The AI Productivity Boom with David Solomon

Well, first of all, I just say that I tend as a CEO and somebody that's running a big business, I tend to think in five- or ten-year increments. And so I'm spending a lot of time thinking about how the world's going to unfold over the next 5 to 10 years and how Goldman Sachs is going to support our clients and participate in that and how the evolution of the world shifts what we need to do to be successful as an enterprise. And so, I think over 5 to 10 year increments. And I would just say I'm extraordinarily excited about the opportunity, particularly for our country, to really grow productivity and grow the economy as this technology is deployed and gets wrapped into enterprises and helps people be more productive in the context of what they can do. That doesn't mean that it's going to be a straight line and going to be simple. In fact, I'm sure and I stated in the op-ed that there's going to be dislocation as there always has been as technology has scaled and it's deployed into our economy and as there are economic shifts over time. I decided to write an op-ed, you're putting yourself out there. There are a number of people that are talking about this and I felt they weren't creating a constructive discussion about the gives and the gets and the balance and I decided that it's an important discussion. There's a lot of change in the world that's happening very very quickly and I thought it was important for Goldman Sachs to weigh in on some of that evolution. And so I don't believe that we're going to have massive structural unemployment. But I do believe that AI is going to interrupt jobs and dislocate. It's moving at a very quick pace. We're going to have to upskill people. The labor force is going to have to adjust. But I think it's important to really step back and understand the labor force, understand the construction of the labor force in the United States when you think about these things because then we can make good decisions as we go forward. And one of the things that I stated very clearly in the op-ed is if the disruption gets to a place where it's very dislocating to certain groups of people then government and business has a responsibility to work on policy to soften that journey. But I really do believe as we get 10 years out, just as other technologies have disrupted, we will have reasonably full employment. We will have a very productive economy and in fact I believe this is going to unleash a growth boom and a productivity boom that could be quite constructive and I think it's really a generational thing. We've seen this before with other technologies. I think one of the different things here is the pace of change, the pace with which it's moving, which can make it more dislocating. But I'm incredibly encouraged, but it's going to be a journey and we want to be engaged and we want to have thoughtful discussions, we want them to be fact-based. People look at the labor force, they talk about technology jobs, for example, technology employment in the United States is 2.3% of the labor force. These are important things to think about. And so, I'm glad I did it. The feedback I've gotten first and foremost, I was at a big meeting, private meeting of a large group of CEOs, people want to engage. They want to think about it. They want to talk about it. They don't like the doomsday narrative. They want a productive narrative about the goods and the bads and how we balance and how we move our economy forward, how we move our communities forward, how we move our society forward. And that's important.

Well, I think, Leslie, the way you frame that, it's just a little bit overly simplistic. Let's start with the fact that we live in a service economy and 65% of all the jobs that are created in this country. This country is a very dynamic economy. Over the last 25 years on average every year we've created and destroyed 30 million jobs. 65% of all the job creation comes from small and medium-size enterprises, a lot of which are service-based. There's an enormous amount of the economy that has to do with people serving people. Okay? And that's not shifting anytime very quickly. So we're talking about what's different is we're talking about certain white collar jobs, right? Okay. And that's a different disruption, but we've seen disruptions to other jobs that people strove for. If you look back 25 to 30 years ago as we opened up the global manufacturing economy, we dislocated a number of people. And I think there are a lot of things that we can learn from those dislocations. And let's hope as there's change here, we find better ways to manage through some of the dislocation that will come. But the economy is very nimble, new jobs are created, new skills develop, and people will evolve appropriately. And it's just not black and white. People want black and white answers. We were talking in the back room there about writing as a career. And it's not going away. It's not going away. And so it's just not simple. It's much more nuanced. And I think the economy is dynamic. I think people are dynamic. And I think people will adapt, but they need time. They need help and it's our job to figure out, as that comes, how to operate and we shouldn't be so sure about how everything plays out. There are a lot of people that are very sure. I'm very unsure. But I think the direction of travel is going to be quite exciting and quite positive when you look at it holistically over a moderate to longer term period of time.

...doesn't mean it will be, doesn't mean it will be, doesn't mean it will be.

Well, like any, I mean, there are different issues tucked in that. First of all, communities will wrestle with the parameters of benefits and drawbacks to having data centers in their communities and it's different in different communities. And that's been the case with all sorts of infrastructure over time and the way it evolves. But the thing that I'd highlight when you talk about all this that I know is true is whenever you have a technological acceleration, everybody anticipates the journey as though it's a straight line in one direction compounding and just the demand for the compute is just not going to go in the straight line that everybody's now currently projecting. And I think we have to be prepared for that to have and flow. Also the technology will change, the productivity of the chips will change, the cost of manufacturing will change, the cost of distribution will change. And so all of this is in the early stages and there are going to be more data centers for sure. I think that the implications for power infrastructure are real and we're going to have to wrestle with that more because certainly it can't increase the cost of power and the cost of utility services to average Americans. That's not going to work. And so this will continue to evolve, but it's not going to be the demand curve that I think people are expecting. It will not go in a straight line. When you think about other technological evolutions, we don't always get them right. I mean, think back to demand for digital infrastructure back in the late 1990s and undersea cable and things like that. People had enormous expectations of how the information would travel. Nobody imagined wireless. They thought everything would travel on fiber optic cable and the world changed. And so the world will change here too and we'll have to figure out how to evolve that over time and we will.

Well, you've got two things going on. You have the infrastructure being built to supply the compute that ultimately has to be bought by enterprises to be used in their business. The one thing I know for sure is that enterprises broadly will go slower at that. They will be slower to change. They will be slower to adapt than I think some of the current expectations. And one of the things you're seeing that is important is these tokens are super expensive. So you saw the article about Walmart saying yesterday 'whoa, token spend going up, whoa'. I'd also just observe as somebody that's talking to a lot of CEOs that there are a lot of things where this technology is helpful but you don't need a super high-powered expensive model. Just a very base model will give you something that improves your operating processes and you don't have to pay for that. And so if you're running a very high margin business, you probably have more propensity at this point to be experimental and aggressive with your token spend and using the technology in the enterprise. If you're running a low margin business, even if it's a low margin business at enormous scale, you start playing with a half a margin point, it has a big implication. And so I think enterprises are going to be cautious about this and the demand curve might be different than the demand curve that's now imagined. So obviously, there's a lot of capital that's going in to build this. The hyperscalers and other big companies who have other businesses that generate a lot of cash are making what they see as a generational bet. And so, even if the returns aren't as good as they'd like them to be, they want to carve out their position and they're going to make significant generational bets. But then ultimately they have to get demand from other enterprises and by the way potentially consumers, and the consumer model is a much more complicated model than the enterprise model. They've got to get demand to pay for all of that. And the one thing that I know is I don't know what that curve looks like. I don't know how that's going to play out. And I think that there's going to be a lot of volatility around that. But ultimately we'll find equilibrium and balance in both the amount of compute, the demand for that compute and how it gets deployed in enterprises and how individuals pick it up in their daily lives.

Well, global equity markets are about... these are rough numbers, don't hold me to them exactly. Global equity markets are roughly 150 trillion. US equity markets are 100 trillion. US money market funds are about 8 trillion. There's enough capital for what we're talking about at this flow in this point. And while I'm not going to talk specifically about any of the deals, I just make an observation that any of you can make by looking at the screen today as Alphabet is going through the process of raising this $80 billion and we are in the middle of it. The stock is trading quite well. This is the largest equity deal, largest follow-on equity deal that's ever been done. The stock's trading very well. That's it. This is the first actual concrete data point for bringing something of this scale and it's encouraging. Okay. And so I think there's plenty of liquidity in the system if the world continues to remain as optimistic and I do think we're in a period... And I know when I say this, I'm pausing for a second to say it, but I know when I say it, it will get quoted, but I think it's definitely true and something for us to reflect on. We are definitely in a moment where there's more greed than there is fear. Okay? And that's one of the reasons why people that need this capital are coming to the markets because the capital's available. And so if you're advising companies that need capital, one of the big pieces of basic advice for 42 years of doing this: when capital's available, if you're capital consumptive and it's available, take the capital, if you know that you're going to need it. And so given the way equity markets are robust, debt markets are robust at the moment, I think you're seeing a lot of activity and I think you're going to continue to see activity. But I just say in the context, these are unprecedented in terms of the size. There's also unprecedented liquidity and wealth in the markets to absorb some of this. And these things also create a virtuous flywheel because you've got a lot of people that have made a lot of money in a bunch of these companies and they're going to be monetizing and reinvesting that into the system into other things and paying taxes on those gains. And so there is a virtuous ecosystem in the context of this too that should support it. But it is unprecedented. We'll see how it goes. But I'm encouraged by what I see so far. But that doesn't mean that something couldn't change that creates more volatility around this just given the size of the scale, the amount of capital that's being raised.

It can turn into fear very quickly. That doesn't mean it will. History is always a great lens and it's never the same, but it's never really different. Alan Greenspan started talking about irrational exuberance in markets in the fall of 1996 when the Nasdaq was like 12 or 1300. Three and a half years later the NASDAQ was 5200. But then ultimately the NASDAQ retraced by over 80%. So he was right, he was just three and a half years too early. So exuberance can go on for big periods of time. At the moment there is enormous opportunity to invest in these technologies. It's one of those generational kind of technological shifts. And I think there's a good chance that we're earlier in the cycle than later, but I don't know. And we could wake up in a couple of months and the lens would be different. Something will happen that will change the lens. But at the moment, that seems less likely in the short term. You know, you haven't asked me about just kind of the macro economy at all. One of the things I'm watching very closely: I in January or February was really quite supportive of higher nominal growth and therefore higher real growth with lower inflation than the market was anticipating at the beginning of the year. After the war started, I think we now have higher and better embedded inflation. We obviously have energy pressure. It's filtering through to supply chains that affect a whole variety of other things. It hasn't yet in ways that are tangible affected consumer behavior that we're seeing, but I'm starting to hear whispers of it. And my expectation is you're going to see more of it in the second half of the year. You're going to see more shifts in consumer behavior because at the end of the day, consumers 14% of their wallet approximately average American is gas and they're spending more and that will force choices and it will filter through in the economy. Also the supply chain pressures because energy gets into supply chains, increases prices on things that are made, and that has to go through. My own view the capacity to keep passing that through is limited and so you'll see that in that context. So I'm watching that. You could see some economic data in the next six months that shifts the sentiment and that changes the balance in all this, but for the moment that's not coming through.

I think the war is a very significant issue. I think the market believes - I'm not saying this is right - the market believes that over some period of time there'll be resolution to that. And with resolution there'll be an easing on energy prices. The market may be overvaluing the pace of the speed of that resolution at the moment. If you did see a material shift in consumer sentiment, I think it would affect the market. I think you've also got to remember you have a small group of stocks that have run very quickly. We could wake up any day and have those stocks retreat meaningfully. One of the things that's interesting is if you look at the S&P 500, you've got 10 stocks that have driven the returns and are 30 some percent of the market cap weighting of the S&P. You've got 490 stocks, this is a generalization, that have kind of gone sideways. And so you look at those earnings multiples, they actually look pretty reasonable. So complicated formula. I do think slower economic growth if we saw it because of the pressure from energy could have an impact on equity prices. But for the moment the medium-term opportunity set of the technology is certainly driving a lot of capital allocation. If things got more complicated in the Middle East, of course that could play through. One of the big risks that I have talked about for years and continue to worry about, it obviously changes with these models and it's getting more attention, is cyber risk. And when you look at the largest companies and the industries that are most vulnerable to this, the largest companies have been investing a lot in cyber. Think about financial services. The large banks have been investing a lot in cyber protection. That doesn't mean that we don't have vulnerabilities. But when you get into the medium-sized banks and the smaller banks, they don't have the same capacity to invest in protection. And we know that if you had a problem in a medium-sized bank, it would filter or create fear in other institutions. We've seen that at other points in history. So I think cyber risk is a big risk. The Conference Board actually did a survey in the last couple weeks where they asked a broad group of CEOs, what is the number one thing that you worry about from a risk factor? and cyber 60-some percent put cyber as a significant risk. So that's something that's getting a lot of attention. You could have a cyber incident that could change confidence and could shift the way we look at markets or the growth trajectory of the economy. There are a lot of things out there. But you also have to ask instead of what can possibly go wrong, what could possibly go right at the moment. There's a lot from a market perspective that's possibly going right. And in the distribution of outcomes, I think there are more distributions that given the technology investment, this continues to go for a period of time. But we at Goldman Sachs, on guard, humbly focused on all the different distributions, trying to talk to our clients as much as we can about things we think they have to think about in what's become an ever more complicated landscape.

They're very, very powerful models. What I think I can say about it is there are a large group of companies that are doing a lot of testing and I think it's important that that testing is done. I think it's important that the information gets shared more broadly so that other companies in the economy can learn. It's one of the things a number of us have been talking about in Washington. How do we get the information that the 40 or so companies have gleaned broadly disseminated to other companies and other enterprises to create more protection? But a lot of this is going to be correlated to consistent long-term investment in cyber protection. And obviously bigger companies do better with that. And as you move down the chain, people are more vulnerable. And so I do think I heard that the president signed an executive order that was a little bit watered down. It talked about voluntary 30-day process for testing these models. I think we want transparency and information and protection before this stuff goes out. But we also have to understand they're open source models that are all over the place and they have the same capacity very quickly to do the same things. And so this is a place where I do think thinking thoughtfully about guard rails and how these companies that are producing this can work with the government and the government can help create some guard rails is something we should be thinking a lot about. But to your point about open source, you went to China with the president. We can set guard rails here in the US, but there are other countries, China working on these models as well. How concerned are you about the ability to set the guardrails and really have them protect the system? You know, everything's just everything's level of step. At the end of the day, the models exist and the models in the wrong hands can do bad things. There are bad actors in the world. There have been bad actors in the world. There going to continue to be bad actors in the world. What we have to do is try to do everything we can to protect against that. And it's not a straight line and it's not perfect and it's not easy.

It's always something. I mean, I'm eight years in. I woke up at the beginning of this year and I said this is going to be eight years in. This is going to be the first normal year where we're just going to run the business. Well, so much for that. War pops in. It's such a dynamic world. We run a big complicated global business. We have extraordinary people all over the world. There are always things going on. It's always changing. But my job as the CEO of the firm is to set a strategic plan to grow the firm. At the core, we serve clients. We want to do it at the highest level of excellence that we possibly can. We want to do it in a differentiated way compared to people that we compete with. And we have a responsibility because we have $110 billion of other people's capital. We have a responsibility to drive good returns on that capital. And it's very hard to be a public company and drive reasonable returns on the capital that you control if you don't grow your business. If you don't grow your business, it's very, very hard. And so I think one of the things we're most proud of is that we've really invested in growing the firm and growing our client franchise and growing our business and we've successfully been executing on it. It has not been perfect, has not been a straight line. I think one of the most powerful things we did is that there were things that were really working and there were things that weren't and we were willing to say not working, we're going to pivot, we're going to change. And I think when you do that, you gain more credibility with your people, with investors, etc. And so we're making good progress. But it's an incredible firm. I feel incredibly lucky to have a great leadership team around me where we're stewarding it together. And I'm excited about the world. There are lots of problems and I wring my hands over problems and I'm concerned about a lot of things going on in the world, but I think the glass is half full. And we have an incredible resilience and ability to grow, to do more, to do better when you start thinking about 10, 20, 30 year periods. And I'm hugely optimistic.