Moody's Corporation ($MCO)

Thank you for being here. For those of you who don't know me, my name is Manav Patnaik. I cover business and information services of Barclays. We're pleased to kick off day 2 for us, at least here with Moody's CFO, Noemie Heuland. Thank you for being here, Noemie.

Thank you.

Maybe, Noemie, I figured the best place to start would be, I think, last year when you came here, it was your first time in London and you just started, it's been about 2 years now. So maybe just some of your reflections and thoughts over the last years. I know maybe when you first added things where you're in a nice stable Moody's company. Now things have completely changed, but just your thoughts.

Yes. It's always been pretty moving environment for the past 5, 6 years. So we're getting used to our first quarter being a little bit hectic. But it's been 2 years. I think we're fortunate to have joined at a very exciting time for Moody's. We have very strong momentum and deep currents in our debt capital markets across both the U.S. and Europe and Asia Pac. So Moody's has a big rule about the different dynamics between public and private markets. A lot of very strong funding needs that underpin the demand for credit and ratings, AI-related infrastructure, maturity walls are very strong. So a lot of exciting signs. And we've invested a lot in our Ratings business to support that demand. And on Moody's Analytics, it's been growing fast over the past 10, 15 years. And we have now completed the integration of most of the acquired entities, and we're excited to join -- to have a new president joining to help us scale further and a lot of opportunities with our proprietary data sets. And I'm sure we'll talk about AI and what that means for us, but it's been definitely a very exciting time. And on the culture side, as you said, I think in the recent note, Rob has been really imposing a lot of change and evolution and automation of a lot of things we do, and that's been just a great experience so far.

Got it. If I could just spend 2 minutes on the culture points. So firstly, on Rob, I mean, I think one of the things that have stood out, like you said, is the is the tech forward change, almost it feels like Moody's has gone through from what was traditionally a rating agency. So can you just elaborate a bit more on that on some internal anecdote perhaps and how that changes happened?

Yes. We -- I was surprised coming in 2 years ago, I spent most of my career in technology companies, and Moody's was actually very advanced in the development of automation tools in both sides of our businesses, are really leveraging technology to improve what we do internally, but also how we serve customers and issuers. We had rolled out, at the time, Copilot across the entire enterprise. And we wanted to really have everybody experiment with the tool, make sure they were familiar with it, how to get their workflow better and serve their customers better. And now we're at the phase where we're scaling a lot of those exciting projects. So that was a pretty good surprise. Everybody was coding and using those tools, which I wasn't quite frankly expecting coming at Moody's. A lot of focus on continuous improvement. And I think that's been really interesting.

Got it. And you mentioned it here, but on your call, you introduced your new Moody's Analytics Head that you're bringing in. Maybe just a little bit more about here because I know when you first took over, you were also a little bit more focused on Analytics. So how is that going?

Yes, Christina Kosmowski, the new CEO for Moody's Analytics is joining in June. She has a very strong pedigree in a technology company. She is very strong in customer success and customer experience, she is what we call customer obsessed, which I think is going to be really exciting for us as we are scaling. And as we move into new territories with the monetization of AI, evolution in go-to-market, evolution in our partnerships, so that we're very excited about her joining. And we also had very stable leadership, very strong foundation, too, that she can build on. So quite exciting.

Got it. Okay, let's move to the business. Usually, with Moody's, you would think we'll start with the Ratings, but I'm going to start with the Analytics first. We'll start with yesterday's headlines around Claude putting out a bunch of new financial tools. Initially, the whole sector, including yourself, sold off a bit, but then you guys rebounded because news broke that you guys were providing some of the data. So let's just start there specifically, if you can just give us some details on with that.

Yes. We have a partnership strategy with a lot of different providers, including Anthropic. About a couple of weeks ago, we announced that we have an MCP app that is available on the Claude desktop environment. And what an MCP app is if you think about our content, our proprietary data, our insights and models, we typically have distributed those in the past through data dumps, SaaS platform, APIs, too, for customers to be able to consume those data sets directly in their environment. It's just a natural evolution of that where we now have enabled customers who are using Claude or OpenAI or other large language model to consume that data directly in their environment to remove some of the friction associated with having to log into a separate website and get the data down and then replug into the Claude or OpenAI or other environment. The MCP app with Anthropic is a step further. It's actually our -- and you can also have our own data separately, but you can consume in your desktop environment in the Claude environment, the automated credit memo agent as well as the KYC agent, which are preconfigured workflows. And that, what this does is it really synthesize, organize the data that you're consuming in those environments and also enables you to be more efficient in your token consumption, which I think is a very strong advantage for our customers. You will see us including more agents in this MCP app, but that's the first step, and I think that was well received by customers. We have our customers in pilot phase. We're actually paying to use those tools and getting some significant efficiency gains as a result of that.

Got it. And in terms of the content that you're providing, including yesterday, like what all content is available through these MCPs or MCP app?

So the MCP, if you think about a traditional either an API feed or an MCP, most of our content is available. If you think about just the MCP app, which is the agentic workflow that we have built over the past 2 years, the automated credit memo agent and the KYC agents are available today on the MCP app that is fed through the Claude environment, and we expect to release more of that.

Got it. And talking about efficiency, token usage, et cetera. Maybe just taking a step back, how should we think about the revenue model or the revenue sharing, whatever it might be today when you signed these agreements?

Yes. A couple of things I'd say, and it's early days. I think we are all thinking with our partners about how best to monetize the opportunity. We keep the direct relationship with the end customers. So there is no revenue share agreement yet. We're not a subcontractor of those large language model providers. Our customers have a direct licensing agreement with Moody's. And as a matter of fact, when they call out the agents or the data, you could see the Moody's, our branding and interface and that directs you to our environment. So that's the first thing I'd say. In terms of monetization, we have -- customers are paying a premium to access those MCP apps and content right now. What is likely to happen over time as we're getting the learnings from those early experimentation and early pilots is we'll likely have a flavor of a consumption pricing with our customers. I want to be careful because our customers have told us, and we also like the recurring subscription-based revenue that is very stable and predictable. But we also want to account and benefit and make sure we monetize the peaks of data consumptions that are going to be resulting from increased usage and consumption of our content through those applications. So you'll have a likely a base fee with consumption-based pricing on top, but we're still experimenting and getting some learnings from those pilots.

Got it. And obviously, we talked about Claude and Anthropic, but are you LLM agnostic? Just can you talk about some of your other partnerships and strategies?

Yes. I talked a bit earlier about our historical distribution channels where we had either a direct API feed to our customers' environment. If you think about large banks, they've always consumed our content through their own applications. If you go move up, down the tier a little bit, you have our customers in Tier 2 and Tier 3 banks who are consuming our content together with prepackaged workflows because they don't have the IT firepower to build their own things themselves. So we're -- we've already had a pretty agnostic distribution channel depending on where the customer is in their journey and how they want to consume our content, and that will continue to be true with the large language models. We have partnership with Anthropic, OpenAI. We have a partnership with Microsoft and also AWS and so on.

Got it. Just to dig deeper on some of the moats that are either debated or that need some clarification. Maybe you could just help start us by setting up Moody's Analytics and the mix, the 3 different segments and what's in there?

Yes. So you have Decision Solutions, which is about 45% of ARR, so annualized recurring revenue. Now we -- 98% of Moody's Analytics revenue is recurring. So that's why I like to talk about the annualized recurring revenue is a good proxy for how we think about the business. Decision Solutions has been growing the fastest. This is where you have the workflows for banks, insurance as well as the KYC. So for banks, the flagship product is our CreditLens offering, which has grown in the high teens in the recent past. That is what I referred to earlier is for Tier 2 and Tier 3 banks to perform loan origination workflows all the way from underwriting to portfolio monitoring and so on. We have our insurance models, catastrophic models that are fed on industry claims data over decades. So it's really proprietary models based on data that we have curated over time. And then KYC workflows that leverage our Orbis data estate as well as politically exposed people database. So that's in Decision Solution. In Research & Insights, we have -- which is about 29% of the ARR. This is where you have CreditView, well, now called Moody's View, the models, ratings, probability of default model, economic research that is used by banks to do their stress testing. And the models have been calibrated over a decade by -- it gets data consortium, so very proprietary as well. And then data and information for the rest of the ARR, which is where we have the pure data feeds from the rating agency as well as the rating -- the data feeds from Orbis that we feed into our clients' environment directly.

Okay. Let's start where you ended. And on the Orbis data set, there is some debate there because a lot of the raw data is available publicly, but then there's some partnerships, some licenses. So can you just help us sort through all that nuances and why you think that's still moated?

Yes. If I think about Orbis, Orbis is a private company database that has over 600 million records now. That's gone from 300 million when I joined. So we continue to invest and expand the coverage and the breadth of the data set. I'll start with the first layer, which is the firmographic data, which is one could argue available more broadly on public sources. But we take that and we curate it and we make it relevant for you, a specific use case. And let me give you a few examples as well. But if you want to know what is relevant for a particular entity that you're doing business with, you can use some of that firmographic data to be connected with the other data sets that are proprietary that you're using to make that assessment. Then when you pass the firmographic. And by the way, if you think about what customers are tapping into when they go into Orbis database, we have a lot of data into that. This is not so much for the firmographic, that's a small portion of it. The majority is for what I'm going to talk about next, which is the curated -- corporate [indiscernible] mapping that is we spent decades curating those, and we have license rights and IP agreement with registries in about 800 jurisdictions. So they have -- we have license rights, we have IP. And then we take that raw data from those registries and we contextualize it. Semantic definition is also important. A dissolution doesn't mean the same thing in Germany as in other jurisdictions. So we want to make sure our customer ingests and have context around those data sets.

And some of these data licenses that you talked about, so one of the debates, obviously, is LLMs can get the data and do it. But are your license partners opening the data app more?

No. Actually, that's the interesting trends that we've observed over the past couple of years. These registries are reducing the number of partners that they're monetizing those data sets with. And again, it's not just about the raw data. It's about all the context that we have built on top of it. The entity mapping, the corporate resolution. That's very important. And if you think about the use cases for those data sets. If you, Barclays, wants to do a KYC, you don't go with good enough data that you scrape on the web, right? It has to be auditable, traceable, documented. And then we have auditors of our customers who come in and audit our data sets. And so that's, I think, where the value is beyond just the context that it provides. It's also a trusted content that's been trusted for the case.

And the corporate hierarchy aspect that you pointed out, that's why people come to Orbis more. My understanding is right now, LLMs are not good at doing that. But is the moat there more the context that you just provided?

It's the context layer that we bring on to give you all the associated risk aspects or corporate links for a particular given entity. I think Shivani, my IR lead, has a great example that I think is very powerful. A few years ago, there was a school district in New Jersey who was using a bus company registered in California. And that company had a linkage a few layers above by a Russian oligarch. So they won't allow to do business with that particular party. And that, the Orbis was the only way for that school district to find out. And that was actually a very strong argument for why large language models or other public sources cannot go into that level of detail.

Got it. Just one layer back that I thought about was I think whenever Claude or one of you guys put out a tool they think, okay, they're coming after our space. But when you guys partner with them, I mean, they are partnering with you, right? So what does that discussion look like? I mean they need you to help.

Yes. They don't have -- so we don't train those large language models with our IP. That's a very important part. But they -- if you think about what a large language model does is it provides some workflow support on actual context and data. Those large language models do not have that contextualized information and data that we have. And so I think that's a natural partnership for us. We're not also trying to compete in the UI or front office. That's not what we do. We pride ourselves in having trusted what Mike and Rob like to call decision grade data sets and insights that then get used into different applications depending on where you are in your journey. So that's where our strength is. And again, we're not trying to compete in the front office or UI space.

Got it. If we can move to Decision Solutions, when you were describing the subsegments, you mentioned workflows a lot. Nowadays, workflows is a risky word, I guess. So can you just help how moated are those workflows? Or is there something more to it than that?

So I'm going to go through the main ones. Lending in banking. We typically cater to Tier 2 and Tier 3 banks who are looking to automate their loan origination systems. We acquired a company called Numerated about 2 years ago that had some AI native embedded feature like spreading financials, automated credit memo and things along those lines. So we've embedded now all the Numerated capabilities in our lending package, and we gave some stats about customers upgrading from the legacy package into the new one with a pretty significant price uplift. That's a product that's grown in the high teens. It has a very strong retention rate. So we feel pretty good about lending. If you think about lending and credit underwriting, you don't want to be the one who are using a third-party large language model or unverified source to come up with your assessment and have this come back and bite you later on. So I think we -- and again, we are audited by our customers. Our solutions go through a very rigorous regulatory and audit process, and I think that's very important. Then I move to insurance, which we have -- the main one here is the climate risk assessment models. And again, this model or these models are -- could the AI come up with a model, sure. But the model is fed on data that comes from the industry. Industry claims data that we get to train and inform our risk output. And again, that's used by insurance to do their risk underwriting. So very core to the business of the insurance company, very core to the front office and growth of those insurance companies. And then KYC, this feels more like a preconfigured workflow to perform KYC checks, leveraging our Orbis database, leveraging our politically exposed people database as well. And again, if you operate in a regulated industry, you want to make sure you have the right tools and workflows and auditability, traceability. How did you come up with the decision? Where is it recorded? What is your justification for making that decision? And the regulator comes in and audits that. So I think that's also another area where we feel quite comfortable.

Got it. On the KYC, we get a lot of questions, too. So maybe first on the data side. You already talked about Orbis, but can you talk about the politically exposed PEP data set, like how important that is, how differentiated that is?

Yes. This comes from an acquisition we did a few years ago. We've continuously enriched that politically exposed database. I gave an example earlier about how powerful that was in either defying sanctions and sanctioned individuals. After the Russia invasion of Ukraine, we had a lot of interest and demand for that politically exposed people database because again, if you operate in a regulated business or even you're looking at supply chain or vendor risk management, especially with DORA and things like that, you want to make sure you have a clear understanding of who you're doing business with from a third-party risk standpoint.

Got it. And then in KYC as well, and maybe it's a broader question, right? But the view is AI can scour the web, find out who's exposed, whatever it might be. But those tools that can be more efficient, are you guys using those tools to make yourself better yourself?

Yes, we are actually using it in our procurement department now. So for our own customer and front office, we're using it. But we're starting using it now also in vendor risk management. As you know, we operate in Europe. We are regulated by DORA, so we have to comply with those regulations. And our tools actually help us do that more effectively.

And just to round up the segment then, the Research & Insights part of it, I don't think there's any debate, it's your ratings research proprietary, but anything else you can say?

Yes, credit research. The other thing I would say, we hear a lot about economic research, which is a part of it, but hey, economic data, you can find this on the web and other publicly available sources. The one thing to note though is those economic forecasts and research are used by banks to do stress testing. So if you listen to Mark Zandi, who's our Economist at Moody's Analytics, he's very careful about when he talks about recession odds and other risks from an economic standpoint because he knows that the minute he passes a certain threshold, this is going to be used by banks to adjust their stress test. So it's pretty serious, again, things that we think is very valuable.

Got it. Before we leave the segment, since you are a CFO, I got to ask about margins. Can you just help us with -- since you've come, of course, the trajectory has improved. And just remind us of your goals and how you get there and some of the moving pieces.

Yes. So Moody's Analytics margin, we have a medium-term target that's by the end of 2027 to be in the mid- to high 30s. We're well underway. We have made some significant improvements again in the first quarter. We started at about 30-ish percent when I joined. That was after a lot of the acquisitions that we had made, investments to be ready for AI and make sure we had the right, again, context layer in our data. We've improved significantly. We're now in the 33% range, and we're again aiming to be in the mid- to high 30s by '27. We're about 150 basis points improvement again this year. And we're doing that by -- there's a few things that are going on in MA, and not a lot of it yet has to do with deploying AI at scale in that segment. So there's a significant opportunity beyond that. It's really about resource allocation. We came to the realization, we had acquired, ingested and integrated most of the entities we acquired. We simplified the operating model. We looked at having one product organization, a go-to-market and make sure we have the thoughtful allocation of resources to drive areas of growth in lending, KYC and data. And so that's really what's been driving the growth. We also have deploy AI and customer support, as you would expect, a pretty obvious use case at first that has allowed us to get some efficiencies. And we're looking at product development life cycle, all the engineering and product development with AI, and that's very promising.

Got it. Maybe let's use margins and shift into Ratings. By any measure, Ratings has very impressive margins. But it sounds like AI could help improve that even. Can you just talk about Ratings?

Yes. Ratings is operating right now at about 65% margin. I mean we, what we say, like to say is being volume agnostic. We have a medium-term target in that ballpark. But we, obviously, if you have a year like 2022, where revenue went down 33%, you're not going to make it up through the efficiencies and the automation, but we're trying to be volume agnostic within a certain band of issuance, and we do that by -- and they started way before I joined, actually, we have automated a lot of their workflow that an analyst go through before the actual human rating committee and assessment happens, things about like spreading financial statements, getting data from different sources, putting those into the methodology template. Running all those workflows has been really automated such that the analysts can spend a lot more time talking with issuers around what's happening in the sector. And last year during around Liberation Day, we had a peak of demand to our analysts to really make sense of the noise beyond the headline. So that's really what's been driving a lot of the margin expansion. We are continuing to invest, though, in Ratings. There's -- I talked about all the funding needs and the currents in the market. We want to make sure we have the analytical capacity and skill set to address those demands. And as you can imagine, it's a long-term workflows planning. You don't switch on and off and adjust the analytical skill set overnight, but we've been able to increase the load of our analysts by automating a lot of the processes that precede the time where they sit down together and think about the rating.

Got it. If you can just touch on issuance and trends and so forth, maybe first on a high level before we touch on a few specifics. But if you could just help us appreciate what happened last quarter and what the current trends look like?

Yes. We had a very strong quarter in Moody's Ratings for -- and I'm not going to repeat, the data from the first quarter. But hyperscaler issuance was very strong. The 5 hyperscalers have issued as much debt as they did in 2025, and we expect this trend to continue. So that was a strong driver for investment-grade issuance in the first quarter. We had, interestingly, in terms of if you think about the macro environment and the geopolitical disruption, in March alone, the 80% of the activity was concentrated in 6 days. So what this tells you is when the markets are open and when it's a risk on day, there's demand and transactions happen and there is no constraint. We'll have to see how that evolves, but that was a very strong driver of growth. In the first quarter, M&A activity was also very sustained in bank loan. We talked to our banking partners again recently, and they see a very strong pipeline of M&A transaction. So that's -- we're pretty pleased with the growth in our Ratings business in the first quarter.

Got it. And on M&A, I know in the prior years as well, there's always been fits and starts, clearly. But how have you -- what are your assumptions for M&A being for the year? It clearly seems like AI, there's upside, but on M&A, how do we think about it?

Yes. We've guided to about 40% -- we've considered about 40% increase in announced M&A. So that's our assumption. Before we came to the market in April, we looked at -- we talked to our banking advisers, and they haven't seen a slowdown in the pipeline. So we've hold on to that, and we saw a very strong first quarter in that regard. I think the -- we always talk about pent-up demand in M&A. We've talked about this for a while. Our rating assessment services, which is a leading indicator for M&A. So an issuer comes to us ahead of a transaction and says, if I do this types of financing or if I structured the deal that way, what would that impact be on my ratings? And that business has been performing very strongly in the second half of '25 and again, first quarter. So that's again a good leading indicator.

Got it. Private credit, obviously, another big topic out there. Maybe just a high level from all your insights at the rating agency. Is this just a headline issue? Is this a potential systemic restructural risk? How do you guys look at that?

So we -- you're referring to a few idiosyncratic events that have a lot to do with fraud or other types of considerations. Private credit has been a strong growth driver for us of a very small base. We raised about $80 trillion of stock of public debt. So private credit is at much lower than that. As you can imagine, it's grown 80% in the first quarter. I think what those headlines bring is a flight to quality, requests for transparency, signpost, indicators and benchmarks, which we are very well positioned to provide. We have a lot of interaction with the private credit players. As a matter of fact, we had 2 credit conferences in the past 2 weeks, 1 in New York and 1 in London, where we had prominent leaders in private credit speak and interact with Marc Pinto, our Head of Private Credit for the Ratings franchise. And you could see really, there is demand. We start to see insurance companies for those guys disclosing how much of their portfolio is actually rated. So there's a demand from investors about signpost and quality, which I think we're well positioned to serve.

Got it. And I know you've talked about as a company focusing on private credit across the company. On the rating side, it's pretty obvious. It's the different categories. How do we think about private credit and the opportunity on the Moody's Analytics side?

Yes. We provide credit. I talked about the credit assessments and the credit models that we provide. So we provide that as well if a joint investor who want to know the quality of the loans you're invested in or the fund you're invested in. We provide that for Moody's Analytics as well. And that gives us the opportunity as those players scale to come and potentially rate those transactions down the road if there's a need for a rating.

Got it. And I think one of the initiatives, and there has been a partnership with MSCI. Can you talk about overall, what that entails and how that's going?

Yes. The partnership with MSCI that we concluded last year, we come in with our probability of default models, our credit estimates and then they have the best, most comprehensive database of loans. So we combine those 2 to provide a probability of defaults and credit assessments on those portfolios. And it's a revenue share agreement, and that's going pretty well.

Got it. And is there another leg where they could do private indexation on the Orbis data set?

That could be -- we're exploring different avenues on the partnership, and that could be one.

Okay. Got it. Maybe in the last 5 minutes or so, just let's talk about capital allocation. Actually, one step back on the Ratings, just to wrap that up, there's always a lot of different moving pieces. There's a lot of focus in the forest for the trees. Maybe you can just remind everyone of the long-term model for Ratings to wrap it up?

Yes. So it's a business that if you look over the cycles, that's grown in the high single digit -- mid- to high single-digit over time. Obviously, there is years like last year, it was significantly higher and this year as well. But the algorithm, first and foremost, GDP is the first pillar of that -- first block of that algorithm. It fuels asset formation. So you take GDP growth, then you have the value that we provide to issuers, and we've recently updated our study to show the savings on the coupon for an issuer that has a ratings versus a bond that doesn't have a rating. So that's 2 or 3 percentage points. And then the last 1 to 2 percentage points of emerging market trends. So here you have a contribution of private credit, domestic market. We have investments in Latin America, Asia Pac and affiliates to serve those domestic markets that are going to be very important in the next decade. And then digital finance as well and other emerging trends to get to the high single digit.

Got it. Okay, moving to cap allocation, I don't think you've changed a whole lot, but just your approach and how you think about how capital allocation priority is set up?

Yes. First and foremost, growth. There's a lot of opportunities to grow both businesses, and I've talked about what those underlying demand drivers are, funding needs, the maturity walls are very healthy, understanding who you're doing business with, understanding the risk of climate-related events on new businesses. So we want to make sure we have investments for that, and we've been investing in acquiring companies in those different spaces. In terms of M&A, obviously, very thoughtful. We have a pretty good record of strong return on those transactions. We'll continue to be very focused on that. I think the bar is raised. Obviously, if you have asked me a few month ago, whether we would be contemplating their specific workflow for one area, I would have said probably yes. Now it's maybe a bit of a different answer, focus on data assets and data coverage. And then the rest is capital return to our shareholders, and we just did $1.5 billion of buybacks this quarter in Q1.

Got it. Maybe just on the buyback. You, obviously, you did a big number in Q1. Therefore, you upped the amount for the year. Just the thought process around there.

Yes, so we were very pleased with what we did in the first quarter to take advantage of the price to increase our buyback. We just divested our regulatory business, and we just closed that transaction on April 30. So we'll use the proceeds to continue on our buyback program, and we'll continue to be opportunistic. I think the good news is we have a lot of flexibility in our capital allocation program because of our operating leverage and our profile. So first and foremost, growth and then making sure we return capital to our shareholders as well.

Got it. I got one more. We do have a few minutes. If anybody has any questions, you can put your hands up. But just on M&A, I want to touch real quickly. You did mention while we were going to Moody's Analytics, a lot of what you built up has been through M&A. And so in the next, call it, 5 years, should we be expecting a lot more even if the bar is higher?

I think again, we'll be very thoughtful. We have high hurdle rates in our M&A. There's a whole debate right now on whether software workflow makes sense. I would argue if it's deeply embedded in customer workflows, if it's a vertical that has a lot of value and efficiency gains for the customer with a lot of proprietary data associated with it, I think it's something that definitely should be looked at. Data coverage, we just acquired a year ago, 1.5 years ago, CAPE Analytics, which is a geospatial data to be used by insurance companies to underwrite their risk on properties and insurance actually wasn't [indiscernible] an insurance company in the Midwest last year, and they loved it. They really said that was something we were really expecting you to do. So we'll listen to our customers. Some of those ideas come directly from customers in terms of data integration on tuck-ins. And so you would expect us to continue to do that.

Got it. I think you had a question. Just wait for the mic, if you don't mind.

I'd love to hear a bit more, so you get a lot of questions on MA and the moats there. But on MIS, I guess you got a lot fewer questions. I guess when we hear about you putting a lot of financial data and then the credit guys sit in a room, the obvious question is why can't Anthropic hire 300 credit analysis and do that? Can you just make sure we understand, is it the NRSRO certification? How do you think about the moats in MIS, and why is that so impenetrable ?

Yes, you're right to say it really comes up as a concern of question. I think people understand and appreciate the value of having the benchmark and the signpost for appreciating the -- assessing the credit quality of underlying loan. I think we have experts who are called in and have a dialogue. So you talk about large language model and having the analyst crunching data. I mean that's the, I would say, the relatively easy part of the job. Where it gets really interesting and where the secret sauce comes in and what our issuers are telling us, and I was an issuer myself, so I interacted with Moody's analysts as well, it's really the depth of understanding of the historical trends of the sector. We have access to MNPI and things that obviously, customers share with us. We do -- our analysts go on site visits. You think about all those infrastructure projects, AI, data center construction, they actually go on site and look at the construction site and talk to the engineers, they have a really deep understanding of the business and not only what's happening in a given quarter or a given trend, but like long term, like what does that mean in the long term and the ratings, obviously, are not expected to move for one quarter to the other, it's really that long-term view, and I think that's what our issuers value today.

We got time for one more, and then we can wrap it up, too. All right. Let's just leave it there then. Thank you.

Thank you. Thanks a lot.

Appreciate everybody being here.