S&P Global Inc. (SPGI) Earnings Call Transcript & Summary

All right. Good afternoon, everybody. Thank you for joining us here after lunch. My name is Manav Patnaik. I'm Barclays' business and information services analyst. And we're pleased to kick off the afternoon session here with Doug Peterson, who is the CEO of S&P Global. So Doug, thank you for being here.

Thank you for having us. Great conference.

Okay. Appreciate that. So first, Doug, maybe just a broad question. We're about 2-plus years since the close of the IHS Markit deal, almost 3 since you first announced it. So maybe just -- we've gone through a lot of cost synergies. You've had that much time to look through it. So maybe your assessment today on what your portfolio looks like and your level of happiness with that.

Yes. So going back 3 or 4 years ago when we first conceived the merger with IHS Markit, we were -- had identified some long-term trends, secular trends that we thought at that time were going to be driving the growth of the financial markets in the future, data analytics, in particular, things like sustainability. We didn't call it private markets at the time, but it's the same thing as private markets. We thought that core data and analytics would also go into the benchmark business. It might change the rating space, and that we needed a partner who would allow us to be -- get a lot more scale. We figured we couldn't do it through organic growth, and we couldn't do it through bolt-on actions, so we wanted something much bigger and bolder that we can move faster with. And thus, the merger with IHS Markit. We started after the merger before it was closed to talk about our organization structure, our management, how we're going to approach towards integration, our approach towards the culture. And I think we've done really well. We've exceeded our approach to synergies earlier than we thought. We had an original target of $480 million. We achieved $620 million. And our revenue synergies, our target is $350 million, 2.5 years from now, and we're already at $180 million of realized synergies. A lot of that is through cross-sell. And the power of the brand of S&P Global has been even stronger than we originally thought. A couple of surprises. There were a lot of upside surprises, the quality of the customer relationships on both sides, the quality of the technology, the ability to move with speed was an upside surprise. Downside surprise, there are a couple of businesses in IHS Markit that were more market-driven than I had probably originally thought. I kind of knew it, but then there's a business within Market Intelligence, which was -- had the Enterprise Solutions business that was impacted by a very low, very slow M&A market, et cetera. But net-net, very successful, strong team, strong brand, really good execution. And as a portfolio, the second part of your question, we now have a portfolio. And you think about it, we've got the Ratings franchise. We've got the Market Intelligence. We've got an incredibly powerful Commodity Insights franchise, the Mobility business and then, finally, the Index business. We're seeing a lot of synergies across the portfolio. That's some of the upside, especially in things like energy transition and sustainability. We're finding ways to move as fast as possible. And one of the interesting areas right now where we're getting a lot of synergies in terms of learning is artificial intelligence.

Okay. So well -- since you ended with that, maybe let's jump to that right away. Just -- we've asked a couple of companies before you as well, but your general view on gen AI and how you think S&P can benefit, either on the cost side or top line as well.

We think we can benefit in both ways, but we're not ready to call it yet what will be that top line growth or what will be the margin improvement we've got. As you know, we bought Kensho 6 years ago, which was an artificial intelligence team based in Cambridge in Boston, Massachusetts that we brought in-house. And over the last 6 years, we've used the traditional AI to build out a whole set of products and services that are really around data, data linking, data management. The reason I mentioned that is the secret to gen AI is to have data that's clean, it's easy to use, you can tokenize it, you can feed it into models. And it's very simply put, what we've done is we've put in place a vision for artificial intelligence for gen AI, which is that it's going to supplement what all of us do but not replace us, that we're going to put in place a set of governance. We have a Chief Artificial Intelligence Officer, who is also the CEO of Kensho. And we put in place a whole learning approach across the entire company. And then we now have something called S&P Spark Assist. It's a copilot that we built internally that allows us to have a firewall around all of the data. So we're protecting all of our data, our intellectual property by bringing the models inside of our firewall, so we can bring in models from anybody, OpenAI, from Microsoft, from AWS, from Anthropic. And we can bring them in into our firewall into S&P Spark Assist. And then all of our analysts, our developers, they can use the tools, download data sets inside of them, and we're seeing an incredible amount of creativity. If something is big enough that we think it's going to make a big difference to either productivity or to a product, we move it into something we call an accelerator, and we've had some successes of getting things out of the accelerator. Nothing that is fully, fully in the market yet. ChatIQ is one we talk about. It's a conversational search on top of Cap IQ Pro that we now have in the market as test gates as a pilot. But we're very excited about it, but no numbers yet.

Got it. And so maybe just a few follow-ups on that. So first, in terms of the models, I know you said you could bring in anything with the Spark Assist, but I believe you guys have your own large language model. So can you talk about the split of usage, if that's the right phrase? How much do you use your own versus all these others?

Yes. We have a model called [ Fin LM ]. It's a financial model. It's -- maybe I'll call it an SLM, a small language model. It's not a large language model. But what we found is that large language models are built off of the word language. They're not financial models. One of the things that large language models do very poorly is financial calculations. If you want to look at things like CAGRs and comparison over times or stock prices or multiples or corporate actions, it can't do the calculations. It can find the language, but then it has a hard time. So we're building tools that allow you to use the large language model to extract the information and then layer on top thing like a [ Fin LM ] that can turn it into usable information. We're also building tools that allow you to do visualization, extraction, et cetera, to build it into workflow.

Okay. And is Spark Assist, just so I'm curious, is it like a sandbox, which any of your divisions can jump in and...

Any person can use it.

I guess, how do you control the cost of using those models? Because everyone talks about how it's very expensive.

Yes. Well, it is expensive, but we are controlling it. We have something we call the S&P Global Kensho benchmark tool. We're now benchmarking all models against 600 standardized questions. You can look at it on our website. And we rank models for the quality of the answer, the precision of the answer, and then we have an overall score as well. And we can then use different models for different use cases. But we're controlling the cost, as an example, by being open architecture. We have -- we're working with everybody. Anybody that's building large language models, we're working with them. We also have the firewall approach, so we bring the models in. Right now, we're running our own, call it, our sandbox, or if you want to call it, this assist tool, we call it S&P Spark Assist. We're running it for -- at a cost of lower than $1 a month per user as opposed to what it would cost to have an exclusive arrangement with somebody else. The tools don't necessarily automatically link into workflow, but we feel like at that cost, it's a really -- it's a great value for us to have. And it allows us to have -- right now, we have 1,800 power users of Spark Assist that are also have built a library of queries. One of the things you learn about AI is that the question you asked determines what the answer is. And so we now have a lot of people who become experts that write very long queries. It's not just write me a poem in Shakespeare style. They are very long queries that will ask for information to be extracted about, let's say, corporate actions or about M&A. And then it not only does ask you to give the answer in this format, in a tabular format with -- et cetera, et cetera. So we're now getting a lot of expertise in how to use gen AI. And so it's a long way to go, but we're really excited about what we're achieving.

Got it. And I understand you have nothing to share now, but just rough time line, what should we think about when you should see some stuff? Just because in your business, you can't be 95%, right? You have to be 99.99%. So just help us appreciate that difference and the time line in which we should be realistically expecting stuff.

Yes. By the fourth quarter into the first quarter of next year, you should get -- please expect us to give you something that's much more complete. Give us the rest of the year to target applications, understand what the use cases are, test the veracity and the quality of the output on the models. But we cannot be -- as you said, we cannot be wrong. It can't be 95%. It has to be 6 9s. We do have one of our systems right now. We are like the old SNL, which we built into Cap IQ Pro, where people that find a mistake, we pay them $50. We're starting to do that with one of our pilots. So if an answer is wrong, we want people to give us the right answer because it actually trains our models. And so we've got a loop now to pay people to find wrong things with some of our AI models because it's just -- it's a very inexpensive way to train the models.

Yes. But be careful with that one because I heard SNL used to send out W-2s to people because they found so many errors.

Yes, they did. There were some people who were good at that.

So let's go back to just the deal. It was a transformational deal, but the one segment that obviously benefited a lot from this deal and that you've been transforming with yours is Market Intelligence. We went -- it was just Cap IQ. Then you bought SNL. Now, Markit. You've got Visible Alpha. So just maybe a big picture question is just help us appreciate your vision for Market Intelligence. And is there a competitor you're specifically going after? Or how should we think about what that solution set ultimately will be?

Yes. So first of all, we think of -- it's a broad set of solutions that are going to meet the needs of multiple users. We're broader than just thinking about a trading floor or a sell side or buy side. We have tools that will be diversified. They're going to be -- there are tools for buy side, sell side and within the sell-side insurance, banking, investment banking, et cetera. So we have a very broad set there. On the corporate side, we have large corporate access to treasurers, the heads of strategy, CFOs, planning offices, heads of marketing, et cetera. So we have a large customer base there. We also have governments, regulators, academics. So one of the beauties is that we don't depend just on any single base. And if there's a -- if we have a weak sales performance or somebody who cancels a contract, it's not going to have a big impact on us because of our diversification. When it comes to the tools, we have different segments. We've got -- one segment, obviously, is the risk services, which is the distribution of Ratings intellectual property. That service is also being enhanced, including a lot of other risk and credit information models, et cetera. We have another business, which is the Desktop. The Desktop has benefited tremendously from the last few years of the upgrades that we've made to the quality of the data, the data sets. We benefited tremendously from bringing in IHS Markit data sets into the Desktop. Last June, we had a major upgrade, which included an upgrade also of graphics of the design of the interface. This is where Visible Alpha is going to sit. So this is a traditional, think of a Desktop, the users are risk managers, investment bankers, bankers, strategic planning heads. The media actually is a big user of this. We have a lot of media clients that use it to do their research. The -- in Data & Advisory, we have things like PMI, and we've got information about supply chain. So it's another type of data set we have. And then finally, we've got the Enterprise Solutions, which is where we have installed software products. This is where we're benefiting something like iLEVEL or Wall Street Office from business activity related to IPOs to financial markets and then private markets. So it's across the group. It's a broad set of capabilities. We also think of it as our data engine for the company as well as our technology engine. There's a lot going on here. We have a large set of data experts there. And it ends up being that also we benefit. As an example, we launched last year in the Commodity Insights business something that we call Platts Connect. And Platts Connect was the convergence of the Platts Dimensions Pro and Connect from IHS Markit. But we actually build it off of the framework of Cap IQ Pro. So this is where we get these kind of technology benefits across the company.

Got it. And in terms of the revenue synergies you talked about, I mean, I think Market Intelligence and then Commodity Insights were the 2 big areas. So in Market Intelligence, stick there, where are these synergies coming from? $180 million I know was the number for the company, but guessing a lot from Market Intelligence. So what are some of the examples of...

Yes, examples of that. Interestingly, we probably got off to a much faster start and have done much better than we thought in terms of cross-sell. A typical cross-sell is where the financial services business at IHS Markit was pretty much focused on financial services. But they had products there, whether it was Investor Relations products, reference pricing, dividend analytics, et cetera, that it turns out that they're really valuable also for commercial banks, for treasurers, for corporates. And the brand opens the door. So we were able to do a lot of cross-sell into new segments that were already clients of Market Intelligence and S&P Global. And so there's been a lot of cross-sell, so cross-sell across the board. That's actually -- we got a lot faster on that. And then the new products are starting to come out. And I'll give you an example of one, which is not only an IHS Markit, Market Intelligence, but it also includes the information from ENR, which was in IHS Markit. So this is something called a Power Evaluator. Power Evaluator is a tool. It's in Market Intelligence, and it's used by power companies to evaluate their entire market, their competitive analytics, what's happening with alternative energies, what are the price of energy. It's a planning tool for power companies and anybody that's distributing power. And we're able to build it with data from Market Intelligence, from financial services and from the ENR business. So we've got this Power Evaluator business. It's off to a start. As soon as we launched it, we had clients because it's one of those huge needs. The energy transition is moving fast, especially in the power industry. And so that's an example of that kind of a product. We have a lot of other products, especially pricing products. We've been able to enhance products like iLEVEL with some of the information that comes from Market Intelligence. So we have 15 different products in Market Intelligence that have been launched. And in the Commodity Insights business, there's also been a whole set of products, especially marrying prices with research. And so there's a lot of new products there. And one final example I'll give you is CERAWeek. CERAWeek, as you know, is the premier energy conference that takes place in Houston every year. We're -- we've now grown it the last couple of years at about 20% per year. This year's up was not quite 20%, more double digit, but not quite as high. But now the entire S&P Global, the entire company is benefiting from it. We've got clients there from every single division, and we've got clients that went way beyond energy and especially way Beyond traditional oil and gas. So that's another area that type of business as well that's benefited.

Got it. And then within MI, I think 1/4 of your business is Desktop. And the question is more around the market environment there. We had FactSet before you. And MSCI, we'll have tomorrow. But -- and you guys reported it, too. But can you -- in your view, the market environment of the customers, where do you think we are? Would you call it cyclical, structural? Just your thoughts there.

Yes. We've seen -- we've been actually talking about for the last 4, 5 quarters, I can't remember if it goes back longer than that, that we've been seeing some headwinds in the Desktop business from banks, from investment banks, not necessarily commercial banks, from investment banks. We haven't seen it so much from the buy side, and we haven't seen it all from the corporates. In fact, corporates has been a growing area for us. So we think that when there has been -- there have been some surprises. Growth hasn't been as fast as we could have expected. We've already built those surprises into our forecast. What we saw earlier this quarter -- or from last quarter, we saw that there was some discussions in the markets about some large banks in Europe that had stopped using the product. We have that in our numbers as well, but we don't have to talk about it because it's already in our guidance. We had already planned for it. We've been talking about it for a while. We've seen a lot of investment bank seats, we've seen a decline there across the board from many of the different banks. But because we have enterprise contracts, we've been thinking about this for the long run, we've already been experiencing this slowdown for a while. It didn't go -- we didn't need to change any of our guidance related to that because it was already built into our forecast.

And are you seeing any kind of pain from the asset manager side or the buy side? I don't know if that's a big mix of your clients.

It's an important part of our mix, but we're actually seeing -- we're not seeing a lot of pressure there. We are seeing consolidation take place. But we're actually seeing new products that we're bringing to them. So we see more -- it hasn't been one of our strengths. We've always had a lot of Desktops inside of the asset management. But we think we're bringing more products, more pricing. We think the fixed income indices that we can bring in through. Our Index business supports that, and then we're including improving some of our tools. So for us, we see that more as an upside.

Got it. And then how about our hedge fund trends? I know your exposure is about to go up now that you've closed the Visible Alpha deal. So just what's the environment there look like? And I guess, what strategy with Visible Alpha is going to be fully integrated into Cap IQ? Or...

Yes, we haven't seen anything yet that I would say is a major trend related to the hedge fund. In fact, the hedge funds are moving faster. They're getting -- seem to be getting larger all the time, a lot more capital allocated there. When it comes to Visible Alpha, this is a deal that we think we're the only organization that could have done it. It was a complicated deal. 12 banks were owners. All 12 banks, we have excellent relationships with, and we negotiated with them to be able to benefit from the 180 analysts and organizations that provide estimates into Visible Alpha. So we can keep Visible Alpha going as a tool on its own and then put it inside of Cap IQ Pro. It's already -- you can already link to it through Cap IQ Pro on day 1 on May 1. We're able to set that up immediately. But in terms of being integrated, it's not integrated yet. But you can link to it, you can use it from Cap IQ Pro, and you can use it as is. We see that this is the type of differentiated data that's very beneficial to add to S&P Global. We think that there probably are a lot of other similar companies that, at some point, will be on the market. And the question is, will they get bought by a strategic, will they get bought by a private equity fund or roll up or something. But this is one of the ones that -- we knew it incredibly well. We knew the quality of the data. We knew who the users are, and we think there's a real stickiness of that user base with Cap IQ Pro, and then there should be a real positive synergy there as well.