管理層發言
Good day, everyone, and welcome to the Verisk Second Quarter 2026 Earnings Results Conference Call. This call is being recorded. We will have further instructions for you at that time. For opening remarks and introductions, I would like to turn the call over to Verisk's Senior Vice President of Finance and Investor Relations, Ms. Stacey Brodbar. Ms. Brodbar, please go ahead.
Thank you, operator, and good day, everyone. We appreciate you joining us today for a discussion of our second quarter 2026 financial results. On the call today are Lee Shavel, Verisk's President and Chief Executive Officer; and Elizabeth Mann, Chief Financial Officer. The earnings release referenced on this call as well as our traditional quarterly earnings presentation and the associated 10-Q can be found in the Investors Section of our website, verisk.com. The earnings release has also been attached to an 8-K that we have furnished to the SEC. A replay of this call will be available for 30 days on our website and by dial-in. As set forth in more detail in today's earnings release, I will remind everyone that today's call may include forward-looking statements about Verisk's future performance, including those related to our financial guidance. Actual performance could differ materially from what is suggested by our comments today. Information about the factors that could affect future performance is contained in our recent SEC filings. A reconciliation of reported and historic non-GAAP financial measures discussed on this call is provided in our 8-K and today's earnings presentation posted on the Investors section of our website, verisk.com. However, we are not able to provide a reconciliation of projected adjusted EBITDA, adjusted EBITDA margin and adjusted EPS to the most directly comparable expected GAAP results because of the unreasonable effort and high unpredictability of estimating certain items that are excluded from projected non-GAAP adjusted EBITDA, adjusted EBITDA margin and adjusted EPS, including, for example, tax consequences, acquisition-related costs, gains and loss from dispositions and other nonrecurring expenses, the effect of which may be significant. And now I'd like to turn the call over to Lee Shavel.
Thanks, Stacey. Good morning, everyone, and thank you for joining us. Today, I will provide a broad overview of our second quarter financial results. We will also offer a perspective on our industry engagement including client discussions focused on the use of advanced technologies, including the evolution of AI and the current operating environment. Finally, I will wrap up with some updates on recent groundbreaking innovations we have introduced into the market, including the release of Verisk Synergy Studio for the Catastrophe Risk market. I will then hand the call over to Elizabeth for more detail in her financial review. Turning to our second quarter results. Verisk delivered organic constant currency revenue growth of 5.8% with balanced growth across underwriting and claims and demonstrating the sequential acceleration that we previously anticipated. Revenue growth was underpinned by an 8% increase in subscription revenues, demonstrating the stability and value of the solutions we provide to the insurance industry, helping them make better decisions. These are supported by our broad connectivity and deep client relationships we have within the industry. Our focus on efficiency and cost discipline drove organic constant currency adjusted EBITDA growth of 7.4%, delivering another quarter of very healthy margins. Growth was modestly ahead of our expectations and reflected the short-term factors we previously described, namely the carryover impact of the very low weather activity, tough compares from strong renewals last year and a work stoppage in a federal government contract. We continue to have confidence that the resolution of these short-term factors and core growth momentum will result in continued sequential improvement in revenue growth as we move through the second half of the year. Moreover, we expect 2026 to be another year of performance in line with our Investor Day targets and reaffirm our 2026 financial guidance. Throughout the quarter, our client engagement was very active. We held several executive sessions across underwriting with top national and regional carriers discussing how we can support clients in the current operating environment, as well as our broader strategic plans, including our AI strategy. Additionally, we hosted client-specific emerging issues workshops within underwriting focusing on topics including data centers, AI risks and regulations and quantum computing, where we discuss company-specific exposures. For one top 20 carrier, we hosted three different sessions on artificial intelligence that were attended by almost 400 underwriting professionals across their enterprise. These engagements are a further demonstration of the critical role we play in the industry through our deep industry expertise, data analytics and thought leadership. That same combination is why we continue to win new data contributions. So far this year, we have onboarded eight new Core Lines contributors and ten contributors to our new excess and surplus dataset. And with those excess and surplus contributors now online, our dataset covers over $18 billion in historical and current premium. Specific to the operating environment, the first half of 2026 has continued on the trends experienced in 2025, namely improved combined ratios and robust industry profitability, helped by unusually low Catastrophe losses. In fact, 2026 is currently tracking for global Catastrophe losses to be below the $100 billion mark for the first time since 2020. With profitability improvements and rate adequacy satisfied across many lines, carriers have turned their attention to driving growth, resulting in price competition and softening markets. As we mentioned last quarter, this dynamic is most pronounced in property and commercial property in particular, where pricing is soft and volumes are lighter. We continue to watch how these dynamics shape client activity particularly as it relates to transactional usage. It is in these types of markets that carriers are increasingly focused on underwriting discipline and risk selection rather than relying on broad-based rate increases. Our comprehensive datasets and analytic insights underpinned by our continued investments in data currency and data coverage can best help segment and understand risk, helping our clients navigate through these dynamic times. In our conversations with clients, AI remains one of the most important topics, and Verisk has partnered with the industry to help them move from experimentation into production deployments and ultimately to scaled adoption when the industry is ready. As carriers expand their use of AI, they continue to turn to Verisk as the trusted source of data in the industry. Currently, those conversations are increasingly centered on accuracy, efficiency and repeatability as well as driving return on investment from their investments. To meet these needs, we are investing in techniques such as Retrieval Augmented Generation or RAG and broader context engineering capabilities that help improve the quality of AI-driven outputs by better managing how large language models access, interpret and apply information. This is all further enhanced by our highly cleansed and structured datasets and deep industry expertise, which is leveraged through the semantic layer of AI processes where we provide data structure, context and rules that make large language models work more effectively to deliver consistent and predictable outcomes. More importantly, our goal is not simply to adopt the latest AI tools, but to apply them in ways that deliver measurable return on investment and value for clients by combining advanced AI capabilities with Verisk's high-quality and proprietary data assets, deep industry expertise and established industry standard workflows while also creating connections across the broader ecosystem. We believe this positions us well to help clients navigate the next phase of AI adoption while strengthening Verisk's role as a trusted strategic partner across the insurance ecosystem. As a specific example, in May, we announced our first collaboration with Anthropic, launching two Verisk MCP connectors for Claude, which bring our trusted regulatory-grade data and analytics directly into conversational AI workflows in an underwriting and claims use case. These connectors enable clients to interact with Verisk's proprietary underwriting intelligence and claims solutions using natural language, helping them access insights when, where and how they need them. These connectors were developed in partnership with Anthropic on an expedited timeline from initial discussion to production and public launch and we are the first insurance data provider with MCPs available with Claude. This was made possible by the earlier foundational investments we made in data organization and infrastructure making our data entirely AI-ready as well as experimentation and development of independent large language model applications. Importantly, the architecture is designed to maintain the governance, security and trust our clients expect, with data retrieved only within the client's Claude session and not used for model training. Interest levels across our client engagement are building along with usage with a top 10 carrier already using the MCP. We're excited by the early client feedback we have received, and we are actively developing additional connectors both with Anthropic and with other frontier model companies that can deliver value to our clients. Additionally, we are excited by the opportunities to develop agentic capabilities in coordination with and on behalf of our clients. Last quarter, we disclosed one of those initiatives to develop a next-generation digitally native agentic underwriting platform with a global insurance firm and we are pleased with the progress that we are making. We are also in the development stage of agentic capabilities in our life solutions that will build on our conversational Ask Max feature to embed autonomous capabilities within the platform. We're also encouraged by the early traction we're seeing with our generative AI solutions that have had scaled engagement and Premium Audit AI is a good example. Premium Audit AI brings natural language access to our premium audit rules and classification content directly into our clients' underwriting and audit workflows. It's still early as we are now in the first renewal cycle post the launch, but we're beginning to see it show up commercially. The incremental value we have added is supporting stronger renewal growth. It's a proof point for how we're embedding AI into our proprietary content to make our offerings more valuable to clients. In our claims business, we continue to see strong momentum in XactAI, which we launched in the fourth quarter of 2025 and have continued to enhance with new capabilities. XactAI embeds AI directly into the workflows of insurers, adjusters and restoration professionals, helping clients automate time-consuming tasks such as summarizing claim information, labeling photos, extracting key data from documents and generating estimating recommendations. The result is faster claims handling, improved productivity and more consistent outcomes for policyholders. Adoption has accelerated significantly. Since March, the number of XactAI users has increased nearly tenfold, and we now have almost 7,000 licensees on the platform. The feedback from clients has been overwhelmingly positive with strong demand for additional functionality. As a result, we are continuing to invest in our innovation roadmap and expand XactAI capabilities across the broader Xactware suite helping clients realize even greater efficiency and value from their claims operations. The rapid adoption we are seeing reinforces our conviction that AI will become an increasingly important driver of efficiency, consistency and better outcomes across the property claims lifecycle. During the quarter, we also reached an important milestone in our Catastrophe and Risk solutions business, with the on-time and on-budget launch of our reengineered U.S. tropical cyclone model and Verisk Synergy Studio, our new cloud-native platform for integrated Catastrophe modeling and risk analytics. Our updated U.S. tropical cyclone model redefines the U.S. hurricane risk modeling by integrating significant advances in climate science, hazard modeling and vulnerability analytics to provide a more accurate, transparent and realistic view of individual risk and portfolio exposure across insurance, reinsurance and capital markets. Specifically, the model includes key scientific advancements, which provide a more physically realistic view of how tropical cyclones evolve, including the impacts of wind, storm surge and inland flooding. Additionally, as Catastrophe models are used to increasingly inform decisions in industries outside of insurance, including housing, infrastructure, capital markets and climate risk disclosure, this new updated model is designed to support risk evaluations with results that can be used across all segments. Verisk Synergy Studio, our cloud-native platform, is designed to help clients make better decisions in an increasingly complex risk environment. The platform brings together advanced analytics, high-performance computing and modern workflows in a single environment, allowing clients to analyze larger and more complex portfolios, generate insights faster and better understand the drivers of risk and loss across their exposures. The first release of Verisk Synergy Studio includes our complete global model suite in our latest next-generation modeling framework, giving our clients immediate access to our latest views of Catastrophe risk for every model around the world. This allows clients to run more sophisticated analyses, evaluate risk more efficiently and make more informed underwriting and capital allocation decisions without having to navigate a patchwork of model vintages. By combining leading science with modern technology, we are helping clients gain deeper insights into risk while improving the speed, scale and transparency of their workflows. We have already onboarded our first clients onto the system and the initial feedback on performance, scalability and keeping our commitment to deliver on the schedule we promised has been incredibly positive. We have a robust pipeline of additional clients scheduled to migrate to the platform through the remainder of the year. As a further enhancement to our Catastrophe risk solutions business, we closed this week on the strategic tuck-in acquisition of McKenzie Intelligence Services, a geospatial intelligence and event response company specializing in global real-time Catastrophe and conflict event analysis. MIS will become part of Verisk Catastrophe and Risk solutions. We believe that the strategic combination of MIS' real-time geospatial intelligence together with Verisk's Catastrophe models, risk analytics, weather analytics and claims solutions will give clients a more complete view of unfolding events so they can assess impacts, prioritize response and support stakeholders more effectively. Before I close, I want to announce that Nick Daffan, our Chief Information Officer, is leaving Verisk after two decades. I want to thank Nick for his partnership and key contributions to Verisk, which include leading the modernization of our computing platform, successfully migrating from the mainframe to the cloud, strengthening the infrastructure supporting reliable delivery of client solutions and helping position the company well for this next phase of AI innovation. With Nick's departure, Jeff Negrete, our CTO, will step into the role of interim Chief Information Officer. We are confident in the bench strength we have in place within our corporate IS&T organization and will commence a search for a permanent replacement. I'll now turn the call over to Elizabeth.
Thanks, Lee, and good day to everyone on the call. On a consolidated and GAAP basis, second quarter revenue was $806 million, up 4% versus the prior year, reflecting contribution from both underwriting and claims. Net income was $229 million, a 10% decrease versus the prior year, while diluted GAAP earnings per share were $1.75, down 3% versus the prior year. The decrease in net income and diluted GAAP EPS was driven by a divestiture at the end of 2025 as well as the higher year-over-year tax rate, increased interest expense and higher legal fees incurred in connection with ongoing litigation. This was offset in part by higher operating results and a lower average share count. Moving to our organic constant currency results adjusted for nonoperating items, as defined in the non-GAAP financial measures section of our press release, our operating results demonstrate continued solid growth across both underwriting and claims. In the second quarter, OCC revenues grew 5.8% compounding on the 7.9% OCC revenue growth in the prior year period. This was a sequential improvement from the first quarter in both underwriting and claims with reported OCC growth of 5.6% in underwriting and 6.1% in claims. The quarter's performance was modestly better than our expectations primarily due to stronger-than-expected insurance-linked securitization activity. The reported results continue to be impacted by the continued carryover effect of a lower level of weather-related events last year as well as the work stoppage in a federal government contract. Subscription revenues, which comprised 83% of total revenue in the quarter, grew 8% on an OCC basis, compounding the 9.3% growth in the prior year period. The growth in subscription revenue was driven by strength across our largest subscription-based businesses, including forms, rules and loss costs, Catastrophe and Risk solutions and antifraud analytics and reflects strong price realization in renewals, expanded relationships with existing clients and the addition of new logos. The resilience of our subscription revenue underscores the strength of our business model and the critical value our solutions deliver to clients. In Forms, Rules & Loss Costs, our investment in Core Lines Reimagine continues to drive strong price realization through subscription renewal as clients experience the additional value we can provide through the transformation of the platform. During the quarter, we released another seven new client-facing modules and remain on track for 25 releases in 2026, delivering even further platform functionality and value to our clients. Within Catastrophe and Risk solutions, we delivered another quarter of strong growth, driven by the addition of new clients and some notable renewals that provide upside to multiyear growth. As Lee mentioned, we are excited to announce the launch of Verisk Synergy Studio in June with our first clients already onboarded and a pipeline for additional clients to migrate on to the platform. Additionally, we released our reengineered U.S. tropical cyclone model, which delivers comprehensive advances in how hurricane risk is quantified and applied and is exclusively available through Verisk Synergy Studio. In anti-fraud, we delivered another quarter of strong growth, driven by good price realization, early renewals, adoption of newer innovations, including digital media forensics, and healthy growth in adjacent markets. Our transactional revenues, which comprised 17% of our total revenues declined 4.2% on an OCC basis, a sequential improvement from the first quarter. The declines were primarily driven by three factors: one, tougher comparisons from overages in the prior year as well as lower volumes in our commercial property business that could persist in the second half of the year; two, the carryover impact from lower weather events in our property restoration solutions; and three, lower volumes in our international travel business, primarily related to Middle Eastern travel disruption. This was offset by strong ILS issuance within our Catastrophe and Risk Solutions business, which we do not expect to benefit the third quarter. Moving to our adjusted EBITDA results. OCC adjusted EBITDA growth was 7.4% in the quarter, compounding on 9.7% growth in the prior year period. Total adjusted EBITDA margin, which includes both organic and inorganic results were 57.5% down 10 basis points from the prior year. As a reminder, the prior year quarter's reported margins benefited from a foreign currency translation impact, which contributed 120 basis points to margin net. Excluding this nonrecurring benefit, we delivered solid margin expansion driven by revenue growth, disciplined expense management and ongoing global talent optimization initiatives. Moving down the income statement. Net interest expense was $53 million in the quarter compared to $36 million in the prior year period due to an increased debt balance as well as higher interest rates. Our current leverage level is in the middle of our targeted range of 2 to 3x adjusted EBITDA. Our reported effective tax rate was 24.6% compared to 22.7% in the prior year quarter. The year-over-year increase was driven by lower tax benefits from a lower level of employee stock option exercise activity. Adjusted net income decreased 1.9% to $259 million. The year-over-year decline in adjusted net income was the result of a divestiture and higher below-the-line items, including higher interest expense and a higher tax rate. Yet we still grew diluted adjusted EPS by 5.3% to $1.98 per share, reflecting our capital return activity, which reduced our weighted average share count by 6.8%. On a reported basis, net cash from operating activities increased 50% to $366 million, while free cash flow rose 58% to $298 million. The increase was driven primarily by growth in operating profit as well as the timing of certain cash payments. We remain committed to returning capital to shareholders and during the second quarter, we paid a cash dividend of $0.50 per share, an 11% increase from the prior year. Additionally, we initiated a $200 million accelerated share repurchase program which was in addition to the $1.5 billion program we entered into during the first quarter. In total, we retired 8.5 million shares during the first half of 2026. Both programs concluded earlier this week, and we currently have approximately $800 million remaining under our share repurchase authorization. We are pleased to deliver continued momentum in the second quarter and are reaffirming our outlook for 2026. A complete list of all guidance measures can be found in the earnings slide deck, which has been posted to the Investors section of our website, verisk.com. We continue to expect consolidated revenue in the range of $3.19 billion to $3.24 billion. Adjusted EBITDA is expected to be between $1.79 billion and $1.83 billion with adjusted EBITDA margins of 56% to 56.5%. We continue to expect net interest expense of $190 million to $200 million and our effective tax rate to be in the range of 23% to 26%. This results in adjusted earnings per share for the year in the range of $7.45 to $7.75. A few things to note as you update your models and think about pacing for the final two quarters of the year. First, we continue to expect a gradual recovery in OCC growth rate for the remaining two quarters of 2026. Second, the transactional revenues in the second quarter benefited from strong ILS activity that will not repeat in the third quarter as this is a seasonal market. Third, while it is still early, the hurricane season is off to a slower start as June and July have had limited storm activity. These factors could continue to exert pressure on transactional revenues in the second half of the year. Still taking everything together, we remain confident in our ability to deliver results in line with our financial guidance for 2026 and in line going forward with our Investor Day targets. And now let me turn the call back over to Lee for some closing remarks.
Thanks, Elizabeth. In summary, we are experiencing the growth rebound that we anticipated in our financial results. Additionally, the increased value in our products from AI functionality is improving price realization and the increased engagement from our clients and partners as they recognize the value of our data and expertise in generating real returns on their AI investments. This has further enhanced our confidence in the operating momentum we are building in the business. We continue to appreciate all the support and interest in Verisk. Given the large number of analysts we have covering us, we ask that you limit yourself to one question. With that, I'll ask the operator to open the line for questions.
分析師問答
Your first question comes from the line of Toni Kaplan with Morgan Stanley.
Lee, you talked about industry profitability continuing to be strong, and that could lead to maybe a continuation of the soft market. But on the other hand, you're helping insurers with profitability tools. So I was hoping you could talk a little bit about historical periods where you've been in this part of the cycle and how growth typically plays out given perhaps decelerating pricing dynamics, but maybe more cross-sell. Should we be prepared for a few years of growth toward the lower end of your growth algorithm? Or can you more than offset that through some of the new product launches and upselling?
Yes. Thanks, Toni. So I think you touched on a couple of the elements that we've talked about previously, and I want to reaffirm, first for some historical context. Our experience has been in soft markets that we've generally grown at approximately 6.8% organic constant currency growth rate. And again, hard markets 7.3%. So both of those are well within the range that we talk about at Investor Day. I think you also correctly identified that in these periods, the value of our datasets in assessing risk and supporting good pricing decisions become more important. That's something we have certainly heard in our client engagement. The level of engagement of clients wanting to have discussions around AI's impact on pricing and market dynamics is evidence of that. The most important point I would center on is that our growth is driven much more by the value that we are able to deliver to our clients and our ability to participate in that. Over the past six and twelve months, the ability for us to utilize our data and our expertise to support our clients' value journey in using AI has been very clearly demonstrated across our businesses. Despite the softening market, while that will have an effect, I think we're more bullish on our ability to create value for our clients and for the industry by supporting their adoption of AI in workflows and processes. We're beginning to see that in the pricing dynamic described for Premium Audit AI and XactAI. So I think that gives you good context around our growth perspective in a softer market.
Your next question comes from the line of Andrew Steinerman with JPMorgan.
I definitely call out what you just said about June and July hurricane activity. I'll also mention that summertime is not typically the peak of hurricane season. With that, have you assumed a low level or a typical level of hurricane activity when you consider the full 2026 revenue guidance? Also, have you considered the return to an El Nino, and I've read about a possible strong El Nino being a near-term driver for Verisk's climate and Catastrophe Risk analytics and our insurance clients asking Verisk about El Nino preparedness?
Thanks for the question, Andrew. Yes, our guidance, as we typically do, considers a base case and an average year of weather activity and hurricane activity. We're highlighting that if the season plays out lighter, that would be a risk on the transactional side. These differentials are small in the overall context of our business and within the guidance range. As to El Nino and the risk to insurers, it's something our Catastrophe and Risk modeling team is engaging with clients on and discussing the latest weather impacts. I'll highlight the acquisition we announced of McKenzie Intelligence Services as a geospatial intelligence platform and event risk management capability that over time can help provide real-time insights as events unfold.
Your next question comes from the line of Kelsey Zhu with Autonomous.
Lee, how are you thinking about your channel partner strategy today? We've talked previously about how Claude and OpenAI could be helpful as distribution channels. Are you also looking at partnering with some of these AI companies for model co-development? If so, are there specific areas or processes that you're interested in joining forces with them to add that intelligence layer?
Thank you, Kelsey. Certainly, we are engaged in discussions with all of the significant frontier model companies. Tangibly, you saw our announcement of the Claude connectors and MCP solution that had two applications, one to underwriting and one to claims that were starting points in demonstrating what our datasets could do by pairing them with natural language capabilities of the model. The feedback from our clients has been strong. In many ways, those were test cases or proof points that we could roll out quickly. One thing we've heard from clients is they want us to focus on what we can deliver now that has an impact and not just conceptual work. That was a demonstration of it. Coming out of that was a clear appetite for us to go deeper and broader in those types of connectors, which will require further development in a number of our product areas. We believe that engagement with the model companies will be beneficial for our clients. It will take time for clients to test the products and develop them, but there is clear enthusiasm and engagement. We are doing this across the board with the model companies. On distribution, our ability to provide distribution to the insurance industry as the last mile, particularly to workflows, is an important dimension. Our connectivity to claims professionals, underwriting professionals and risk professionals gives us the ability to identify where those models can be best and most effectively applied to generate real returns for our clients. That's been recognized by the model companies we're in discussions with and strengthens our position in helping them and our clients bring model capability together with the datasets.
Your next question comes from the line of Scott Wurtzel with Wolfe Research.
Just wondering if you can talk a little bit more about the lower volumes on the commercial property side and why you would expect those lower volumes to persist in the second half of the year?
Thanks, Scott. I'm going to turn this over to Saurabh Khemka, who runs our underwriting businesses, to give you a perspective on the commercial property trends and what we're hearing from clients.
Absolutely. In the commercial property line and the insurance industry, what we're seeing is soft pricing; it's a cyclical trend. The behavior this is driving is that our clients are looking at their business more selectively; they may not quote as many opportunities. When they do quote, they're evaluating how much they're spending on underwriting data and analytics to analyze that business. Those impacts are being felt in the transactional part of our business. I do want to emphasize the underlying business remains very strong, and the subscription side continues to be healthy.
Your next question comes from the line of Gregory Peters with Raymond James.
One, I know most if not all of the large property casualty companies have walled off general LLM access to their data and pricing infrastructure. So I have a two-part question. Are you seeing any shift among your customers' attitudes toward LLM access? And can you step back and remind us how you're preserving your unique data assets amid the growing proliferation of AI?
Great. Thanks for the question. Data security has been a critical issue that we have managed carefully for decades. It's central to our DNA to make certain we are protecting our clients' data, and any application or new technology we develop is done with careful consideration and discussion with clients about rights to utilize that data and technology. That informs our governance approach to data use. We have had a data policy and an AI policy for many years, and that guides how we protect data. Two additional points: in our work with frontier model companies, it was critical that the use of data be controlled in the client's instance and could not be used to train models. That's an important protection for our clients and for preserving our overall data access. This demonstrates our care in managing access to datasets and protecting them. It is front of mind to ensure our clients' proprietary information is protected. While we are finding ways to utilize that data for valuable applications our clients want, we are doing so in a very safe and carefully thought-out manner.
Your next question comes from the line of Manav Patnaik with Barclays.
Lee, I just wanted to touch on the AI monetization today. You alluded to two examples, which I was hoping you'd elaborate on. The MCP connector: you mentioned one of the top 10 carriers is already using it. Are they using this in addition to legacy systems or instead of them? And second, you mentioned XactAI growing tenfold to about 7,000 licensees; how is that contributing to growth and how should we think about that over time?
Thanks, Manav. We're focused on demonstrating that these technologies create incremental value for clients and help them achieve a return on their AI investments. If we can demonstrate improved efficiency, productivity and accuracy, there will be several channels to monetize that incremental value. One immediate channel, not yet at scale but with potential, is improved pricing renewal capture in subscriptions, as we've seen with Premium Audit AI and with Xactware licenses where we expect to capture the incremental value delivered. Ideally, this migrates into a more stable subscription model over time. A second potential channel could be initial transactional pricing for clients to test and experiment with new AI applications, which could then convert into subscription. Third, licensing fees for platforms where we deliver agentic or platform solutions that clients view as appropriate for monetization. We are realizing some of this now on the pricing front, and as we move from piloting to broader client usage, we could see transactional elements and licensing for platforms. Overall, our experience with Anthropic and other model companies has increased our confidence in the value of our data and our ability to support clients' AI journeys as a clear upside over the next several years.
Your next question comes from the line of Faiza Alwy with Deutsche Bank.
I was hoping for an update on the Auto business because I know you had talked about some new datasets and solutions coming to market this year. Where are you on that and how are you thinking about that business going forward?
I'm going to turn that question over to Saurabh.
Thank you, Faiza. On the Auto side, we continue to make progress. The competitive environment remains robust, but our focus on differentiated platforms like Lightspeed and new analytic objects in our coverage verification business is delivering good results. Lightspeed is being adopted by more customers and we're seeing good engagement on differentiated analytics with other customers. We continue to focus on that strategy as we go forward.
Your next question comes from the line of Henry Hayden with Rothschild & Co.
I had a follow-up on MCP monetization. Is this volumetric? Is there a volumetric component to how you're charging for this or is it covered by subscription? Depending on that, how should we think about incremental margins versus variable usage costs and higher storage costs for vectorized datasets? Also, how should we think about pace of adoption? You mentioned one carrier is on now; how do you see it rolling through the client base?
Thanks, Henry. Specifically on the MCP connections, it is included in the client's subscription; their Claude costs or token costs are covered by them separately, so that is not an expense to us directly. More generally, we have strong governance on AI spending across our enterprise. On the pace of adoption and diffusion across the insurance industry, clients are adopting and we are ready to move with them as they transition into more systematic deployment. Given the focus on governance, data security and protection, monetization opportunities will increase gradually over time, but it will likely be a slow and steady build.
Your next question comes from the line of Andrew Nicholas with William Blair.
This is Tom Roesch on for Andrew Nicholas. I was wondering if you could provide some color on what supports your confidence in the acceleration in second-half organic growth rates. It sounds like transactional, at least from storm activity, is softer through the first two months, but I recognize last year's third quarter was historically low in storm activity. What areas of the business give you the most confidence on the acceleration in the back half of the year?
A couple of things give us confidence. One is the subscription growth rate; that's 83% of our business and provides visibility. Second, the federal government contract work stoppage has been lifted as of the beginning of the third quarter for a one-year term, so that work has resumed and removes that risk. Those elements support steady acceleration. That said, the transactional revenues benefited from strong ILS issuance in the second quarter, which is not likely to repeat in Q3, and weather remains a variable. We expect a steady step-up in year-over-year organic constant currency growth, perhaps less upside in Q3 and more acceleration into Q4, with confidence in the full-year guidance supported by subscription revenues.
Your next question comes from the line of Jeff Meuler with Baird.
Thinking through the financial impact and timing from the Synergy Studio release: is there a platform upgrade fee that you monetize? Or is this mainly upselling and cross-selling more advanced models? Also, one competitor talks about high-definition models as a differentiator. You're calling this a next-gen modeling framework—what's different about next-gen models and how does that change the competitive dynamics, recognizing clients may buy models from multiple vendors?
Thanks, Jeff. I'll start on differentiation. The models are the starting point: we believe we're releasing sophisticated scientific, leading Catastrophe models that are more physics-oriented and realistic. With the U.S. tropical cyclone model, we think it's cutting-edge science. Synergy Studio increases clients' capacity to run larger simulations over longer periods more quickly, improving speed and scale. The platform also makes it easier for clients to receive updates compared with traditional on-premises solutions. A key differentiation is a consistent economic framework so all our models can be used to assess risk on a consistent basis across an entire portfolio, which is valuable for clients who want global portfolio views. Those elements are why we think Synergy Studio delivers more value.
On pricing, all of our models are on that next-generation financial framework and will be available on Verisk Synergy Studio. There is not a specific platform upgrade fee that we're monetizing. Some customers may choose to have the platform hosted and operate on a hosting basis. Some customers have migrated to a hosted model previously, which has added revenue. Taken together, this is contemplated in the Investor Day guidance for the Catastrophe and Risk business at a long-term rate of about 8.5% to 9%.
Your next question comes from the line of Curtis Nagle with Bank of America.
Could you comment on growth in Catastrophe and Risk in the quarter? I think it was up double digits in Q1. What did it look like in the second quarter, and how should we think about expectations in the back half of the year, particularly with the launch of Synergy Studio?
We don't provide specific ongoing disclosure on that business, but it has been a strong contributor. Q2 benefited from securitization activity on the transactional side. The business is seeing steady performance in line with its long-term guidance.
Your next question comes from the line of Jason Haas with Wells Fargo.
A high-level strategic question: you mentioned an agentic underwriting platform and XactAI, which sounds like a workflow tool. Is there an opportunity to use AI to push more into workflow software? Historically you've been more of a data company with some software. How are you thinking about where you want to take the business?
Thanks, Jason. There are clear opportunities to integrate our data and insights into workflows. The predominant path will be integrating those datasets into our clients' AI solutions or existing workflows. Some clients have asked us to develop platforms that integrate their workflows more effectively, and we're excited because we have the expertise and data to do that. XactAI is part of our Xactware platform servicing claims professionals. There is a software opportunity for us to deliver on clients' needs for data integration, automation and modernization of processes, and to connect participants within workflows. We can deliver that connectivity in partnership with clients' preferred platforms, in partnership with AI solutions developed by model companies, or in specialized platforms we develop. Ultimately, we are focused on leveraging our datasets and expertise to create more value for clients, and we have multiple means to do that.
Your next question comes from the line of Ashish Sabadra with RBC Capital Markets.
Thanks for sharing the client engagement and monetization for AI. Previously it was noted that sales cycles had elongated due to AI. Have you seen any shift in that trend recently? Also, insurance companies are launching industry-specific LLMs—could they move some of these processes in-house? How should we think about AI as a net positive or as posing headwinds from in-housing?
I'll take the second part first. Yes, some clients are developing their own LLMs. The opportunity for us with those clients remains strong because the utilization of our unique, standardized, cleansed data is highly relevant to them. Our semantic layer and RAG capabilities remain important. The more sophisticated clients are from a technology standpoint, the more data they consume and the broader the relationship they have with us. On sales cycles, we've seen elongation remain in some cases due to legal discussions around data usage rights and protection. Those discussions are important and we view our strong governance and long history of trust as a competitive differentiator.
On sales cycles, the elongation pattern remains in some areas. Clients have intense focus on data usage rights and data protection, and we have a strong legal and governance framework to ensure confidence. In the long run, that is a competitive differentiator for us.
Your next question comes from the line of Jeff Silber with BMO.
I was hoping you could help us on the EBITDA margins for the quarter. I know there was a tough comp with FX year-over-year, but could you walk through other drivers for margins? And looking forward, are you still anticipating expense ramp for the second half of the year?
Thanks, Ryan. It was a strong margin quarter. Q2 of 2025 included a 120 basis point benefit from foreign exchange translation. Excluding that nonrecurring benefit, our margin expansion is real. On a trailing 12-month basis we're at 56.3%, which is in the midpoint of the guidance. Factors for the strong margins this quarter include seasonality—Q2 is often the strongest quarter because ILS transactional revenue comes in at a high incremental margin and expenses build over the course of the year. We've also maintained strong expense discipline and delivered margin expansion across prior quarters despite transactional headwinds. For the balance of the year, seasonality and timing of expenses will impact margins, and we may invest further behind innovative products as revenue growth accelerates.
Your final question comes from the line of George Tong with Goldman Sachs.
You highlighted AI as a driver of stronger renewals and improved price realization. Can you help quantify the contribution that AI is making to organic growth today? Is the benefit measured in tens of basis points? Or has it already become large enough to contribute more meaningfully to organic growth?
Thanks, George. We're still at an early stage, so it's hard to quantify current impact precisely. Based on what we see, we clearly believe AI should lift our organic growth rate to some degree. It's difficult to parse out across overall renewal and other value we're providing, but we feel more confident that this is additive to our growth rate. We see increasing client engagement and recognition of tangible value from integrating our datasets with AI models.
This concludes today's call. Thank you for attending. You may now disconnect.