How Moody's can be an AI-enabler, but remain resilient to AI disruption itself. CEO Robert Fauber lays out the data

One positive reality check to emerge unscathed from the ongoing hype cycle is the criticality of having a solid data foundation underpinning AI. Garbage in, garbage out has always been an enterprise tech maxim and it’s never been more true than today.

So that should leave companies specializing in data in a prime position to benefit from the AI revolution? That’s certainly the view of Robert Fauber, CEO of Moody’s, the US financial services and credit rating giant, who states:

I think we all understand that data and trusted data is going to be the fuel for AI, especially for the big regulated institutions that are big customers of ours, and so we feel very good about having built out this massive data estate.

What Moody’s has at its corporate fingertips is not only massive, it’s proprietary and that’s going to be a major competitive benefit, argues Fauber, pitching that proprietary data sets will be at a premium in the AI age:

We have a massive proprietary data estate and we're in the process of unifying all of that, all the data, the models, the ratings, the research, the risk assessments into really a single normalized record for each entity. That is going to be able to give us the ability to create a very, very powerful knowledge graph, and then we're going to keep adding to that. That is going to enable agents to be able to access a comprehensive inter-connected view of any entity, give unique insights, and allow for richer decision-making.

Trust is going to be a critical success factor here, adds Fauber:

We're assembling all of that into what we call a trusted context layer that sits between the raw data assets and the AI reasoning engines. So it makes the data usable for reasoning. What that is, is a structured, governed representation of what the data means, how it relates across entities and time and scenarios, when and why the data should be applied and much, much more. It is a deep contextual understanding of the data.

Orbis 

One of Moody’s prime assets is the Orbis database, containing financial data on over 600 million companies with a ten year rolling window. This includes ‘gold dust’, such as corporate ownership information, including shareholders, subsidiaries, ultimate owners, company news, deals, & royalties information and so on. Fauber says:

It's not just company data. It's years of entity resolution, ownership mapping, expert judgment, and of course, a complex ecosystem of licenses and IP rights. We've built all of that context directly into our analytics, our methodologies and our models so that then the outputs are accurate, they're explainable and they're defensible, and they're ‘decision-grade’.

As such, Orbis is one of the biggest parts of Moody’s data estate, he adds, and it has the added benefit of being an asset that would be very hard to replicate by competitors:

First of all, a lot of the data just simply isn't available to the public. We have a complex ecosystem of commercial agreements and IP rights that has taken us decades to build and we're constantly curating that. Second, there's legal and regulatory issues, privacy laws and export controls, and all sorts of things that our customers need to know that we're abiding by if they're going to use the data. There's semantic complexity. This gets into things in different jurisdictions mean different things.  [AI]  models have a lot of challenges with semantic drift. We've been curating all this over decades and our local experts understand what different things mean in different locations. [so] then they're cleansing and normalizing that data to make it valuable.

And there’s more:

There's entity resolution and ownership inference. The models are not simply doing entity resolution. It is a really important thing to be able to resolve against the right entity and we've combined probabilistic models, human-in-the-loop validation, and proprietary logic, and we've been doing this over years and years and years.

And then we've got all this historical depth. In some cases, the data has either been archived or it doesn't exist in digital forms. It's not easy to get some of that history.

And of course, given the regulatory regimes under which so many Moody’s clients operate, governance is crucial. This makes them demanding customers with high standards, Fauber explains:

Every bank I talk to tells me, ‘Good enough is not good enough for our institution’. What they want from us is they want to move, in many cases, to fewer trusted providers, so they want us to be able to meet their needs.

And all of that is not something that AI tech is going to replace wholesale, he argues:

I'll acknowledge that things like automated data ingestion and things like that will be done by AI, but it's those things that I talked about [that make the difference]. And it's not just Orbis. You could go across a number of other data sets that we have, and the same is true.

First principles in action 

There are some basic first principles to which Moody’s adheres, he adds:

Our accuracy, providence and the audibility are non-negotiable. Our data can't be synthesized from public sources. It reflects how ownership and control actually work in the real world, cutting through complex multi-layered structures across jurisdictions, and reflecting years of proprietary data curation, entity resolution and relationship mapping.

It's that breadth and depth that makes our data both AI-enabling and AI-resilient.

Theres a common thread running through all of this Fauber pitches namely that as AI proliferates, value will accrue to providers of trusted context, decision grade data and analytics that are embedded, auditable and difficult to replicate. That suits Moody’s, he says:

A broader truth that as AI becomes a new interface for decision-making, the need for trusted context increases, not decreases. AI systems require verifiable permission, domain-specific data and analytics to produce outputs that are accurate, explainable and defensible. That's exactly what Moody's provides, and it gives us the opportunity to become even more deeply embedded in customer workflows.

This is a reality reflected in client behaviors of late, he attests:

Customers who have purchased or upgraded into at least one stand-alone gen AI or agentic solution are retained at a rate of 97% and are growing at roughly twice the rate of the rest of the customer base...AI adoption is driving greater consumption of our proprietary data, expanding our share of wallet, and re-enforcing long-term customer economics, particularly amongst our largest strategic accounts.

A key reason for adoption accelerating is the experience offer to enable customers to consume Moody’s data, he suggests:

Moody's solutions are delivered through our own applications, and increasingly, they're embedded directly into customers' existing technology stacks and third-party workflow platforms, including systems like Salesforce, ServiceNow, Coupa, Intapp, Databricks. And we've made our content available through smart APIs and MCPs and specialized agents for consumption through our customers' own AI platforms and going forward through AI portals like Claude and OpenAI. 

This is enabling us to serve our customers on a different level and in different ways than ever before. So for our banking customers, AI-enabled workflows such as automated credit memos and early warning systems are delivering some material efficiency gains, reducing cycle times while improving consistency and regulatory compliance.

This is delivering results he argues, citing a stat that roughly 2/3 of eligible renewals converted to Moody’s AI-enabled lending suite in 2025 with an average uplift of about 67%. He also points to “a large globally systemic important bank” that consumes the firm’s gen AI-ready data and smart APIs to embed into its digital credit platform in order to automate financial analysis and accelerate wholesale lending decisions.

Elsewhere, a  “Tier 1 US bank” has deployed Moody's agentic solutions to automate credit memo creation he adds:

They've told us that it can generate roughly 35% to 40% of each memo and saves analyst hundreds and hundreds and thousands of hours of time, equating, in some cases, to millions of dollars saved. And that work is expanding into enabling real-time commercial real estate risk monitoring, API-based screening, and KYC (Know Your Customer).

But does it pay? 

But in this age of Wall Street short termists demanding that everyone with an AI tech offering has to ‘show us the money’, how much is Moody’s adding to its own bottom line line through its investment in growing out its capabilities here? Fauber is careful in his answer to such questions:

Everybody wants to understand how much revenue is being generated by AI. There were two stats that I want to come back to because I do think they are leading indicators for us.

One, is the fact that those largest accounts for us are growing at about twice as fast as the rest of the portfolio. That's really important because that's where we have the deepest engagement with the most sophisticated institutions on the planet, and that's where they all want to be able to consume our content and bring it into their own AI workflow orchestration platforms and consume it through AI portals. So there is a lot of AI-oriented engagement with those big institutions. That's what's driving, and importantly, driving that growth.

And then second, we have that stat about the cohort of customers who have bought at least one stand-alone packet or upgraded into an AI solution, that's growing twice as fast, again, because of the level of engagement.

So I feel good that the most sophisticated institutions are where we've got the most growth and the most engagement around AI. And our view is that that's going to then trickle through the rest of the customer base over time.

My take

It’s a compelling thesis and pitch and Moody’s is undoubtedly sitting on top of a data gold mine. As the message about the importance of having solid, clean, actionable data to support AI systems scales up across enterprises of all kinds, that’s an asset that is going to serve the organization extremely well.