Thomson Reuters Corporation (TRI) Earnings Call Transcript

Good day, everyone, and welcome to the Thomson Reuters Second Quarter Earnings Call. Today's conference is being recorded. At this time, I'd like to turn the conference over to Mr. Gary Bisbee, Head of Investor Relations. Please go ahead.

Thanks, Jennifer. Good morning, and thanks, everybody, for joining us today for our second quarter 2026 earnings call. I'm joined by our CEO, Steve Hasker; and our CFO, Gary Bischoping. Steve and Gary will discuss our results, and then we'll take your questions following the prepared remarks. [Operator Instructions] Throughout today's presentation, when we compare performance period-on-period, we discuss revenue growth before currency as well as on an organic basis. We believe this provides the best basis to measure the underlying performance of the business. Today's presentation contains forward-looking statements and non-IFRS and other supplementary financial measures, which are discussed on the special note slide. Actual results may differ materially due to a number of risks and uncertainties discussed in reports and filings that we provide to regulatory agencies. You may access these documents on our website or by contacting our Investor Relations department. Let me now turn it over to Steve Hasker.

Thank you, Gary, and thanks to all of you for joining us today. Our strong start to 2026 continued in the second quarter, with revenue growth ahead of our prior expectations and margins in line. Total company organic revenues rose 8%, with the Big 3 accelerating to 10% organic growth, up from 9% in recent quarters. The acceleration was driven by Legal Professionals and Corporates, which both also accelerated to 10%, up from 9% last quarter. We are raising our full year 2026 outlook for total inorganic revenue growth to approximately 8% or the high end of the prior 7.5% to 8% range. We're also raising our Big 3 total and organic revenue outlooks to a range of 9.5% to 10%, up from the prior approximately 9.5%. We continue to forecast margins rising year-over-year to approximately 40%. On July 14, we were pleased to announce the signing of a definitive agreement with KKR to form a joint venture to operate the Global Print business, where we will sell a 51% stake for approximately $500 million. We're excited about this transaction, which will sharpen our focus on content powered AI solutions serving fiduciaries, while setting up an independent business to serve our customers' print needs. The transaction provides attractive proceeds to TR and will be modestly accretive to organic revenue growth. Gary will provide additional details in a few minutes. We continue to invest heavily and remain encouraged by the growing success of our innovation engines. Commercial momentum across our AI-enabled offerings continues to build, and our pipeline of features and offerings in development continues to grow. In a moment, I'll provide an update on Thomson, our proprietary large language model, which we see as an increasingly important tool to deliver accurate and cost-effective AI solutions. In addition, I'll highlight the successful completion of the next-generation CoCounsel legal beta, the introduction of AI-driven capabilities into ONESOURCE and our excitement about a next-generation version of CoCounsel for tax and audit currently in development. To support our product investments, last month we launched The CoCo, our largest brand campaign in more than a decade to accelerate awareness and demand for CoCounsel. The campaign reinforces our differentiated position in professional AI by highlighting what our customers value most, trusted fiduciary grade AI grounded in authoritative content, domain expertise and the accountability required in professional workflows. Our capital capacity and liquidity remain a key asset that we are focused on deploying to create shareholder value, and we made solid progress on this during the quarter. In May, we executed a $605 million return of capital. And on July 21, we completed the $600 million share repurchase program announced in February. Together, these transactions have reduced our share count by approximately 3%. We remain committed to a balanced capital allocation approach, and we continue to assess a number of inorganic opportunities. With approximately $9 billion of estimated capital capacity through 2028, we are positioned to be both aggressive and opportunistic. Turning to the second quarter results by segment. The Big 3 segments accelerated to 10% organic revenue growth, up from 9% in recent quarters. Legal organic revenue accelerated to 10%, driven by continued strong law firms momentum and improved government growth. Legal, excluding government, continued to grow at the 11% pace we saw in Q1, driven by momentum from Westlaw and CoCounsel Legal. Corporates organic revenue accelerated sequentially to 10%, driven by offerings in our legal, tax and risk portfolios and the segment's international businesses. Pagero was particularly strong and continues to drive market share gains for Thomson Reuters in the transactional compliance space. A recent significant Pagero win with Google is one example. Tax, Audit & Accounting organic revenues grew 8%, driven by CoCounsel for tax and audit and Latin American business and SafeSend. Reuter's organic revenues rose 4%, driven by growth in the agency business and our contract with LSEG. And lastly, Global Print organic revenues declined 3% year-on-year, in line with our expectations. And in summary, we're pleased with the building revenue momentum we've delivered in the first half of 2026. I'll now discuss our continued portfolio evolution and provide several product innovation updates. The Global Print transaction I mentioned earlier continues the positive evolution of our portfolio. As you know, we have invested heavily in innovation in recent years, both organic and through strategic M&A. We've also pursued targeted divestitures, including Elite, FindLaw and now a majority stake in Global Print. These efforts leave us with a stronger, more focused and more strategically aligned portfolio, with improved growth prospects versus the TR of just a few years ago. Adjusting our last 12 months performance for the Global Print transaction, the Big 3 segments would contribute 87% of our revenue, up from 81% in 2023. Our Big 3 revenue growth has accelerated from 7% in 2023 to 9% on a last 12-month basis, and we remain focused on building upon the 10% growth this quarter. Total TR improved from 6% in 2023 to 8% on a last 12 months basis. The quality of our revenue mix has also improved, with recurring revenue rising to 86% of total on an as adjusted basis, up 6 percentage points from 2023. When including repeat transactional revenue, we have good visibility into over 90% of our annual revenue. Looking forward, our focus remains on driving an accelerating pace of innovation as we deliver authoritative content powered AI solutions that provide fiduciary grade outcomes for our professional customers and markets. Let me close with a few thoughts on our innovation roadmap. If this chart looks familiar, it is an updated version of one we shared a year ago. Like last year, we are delivering a significant portfolio of innovation in 2026, including new offerings, additional capabilities and geographic expansion. Let me share a few highlights. In June, due to the strength of customer feedback, we completed the beta for the new generation version of CoCounsel Legal ahead of schedule, and began providing early access to all existing CoCounsel Legal customers. Customer usage is ramping, and we remain on track for the broader launch by the end of this month. Outside of Legal, we have added several AI features into our ONESOURCE portfolio, including touchless compliance, which automates the creation of U.S. sales and use tax returns and AI research for global trade, which leverages our authoritative content to simplify trade research. Pagero has continued its geographic coverage expansion with the addition of 5 more countries, including France, Poland and Belgium, building on its market leadership position, and we are working on an agentic next-generation version of CoCounsel for tax and audit expected this fall. Let me now provide an exciting update on Thomson. As a reminder, in mid-2024, we made a modest but highly strategic acquisition of Safe Sign Technologies, a start-up that was developing legal specific large language models. Over the last 2 years, highly talented teams from Safe Sign and TR Labs have continued the development of these models, leveraging TR content and expertise along the way. They recently completed development of the first production-ready version of the model, which we call Thomson. Joel Hron, our Chief Technology Officer, recently issued a blog post discussing the results of a detailed benchmarking study of [ Thomson ONE ]. Despite relatively modest investment of approximately $40 million and training Thomson on less than 10% of our legal content to date, the benchmarking study indicates that Thomson delivers results on par with the latest versions of the leading frontier models on a broad range of general domain tasks. And as expected, Thomson performed strongly for legal tasks, with further improvement potential as we add more TR legal content. This best-in-class performance is delivered at a meaningfully lower cost, and in many cases, at significantly reduced latency versus third-party models. One might ask how we can deliver results on par with frontier models at a fraction of the cost. The answer lies with our content and our expertise. When building on leading open source models, the quality and sophistication of training data matters far more than the volume of data used. Our deep repositories of expert curated or authoritative content across Westlaw, Practical Law and Reuters are a key advantage as are our attorney editors and practice experts. The benchmarking results embolden our strategy for Thomson and provide growing confidence in its potential. We are on track to power tabular analysis, a bulk document review tool in CoCounsel Legal with Thomson later this month. And we see an opportunity to port over a broader range of capabilities in the future to leverage Thomson's cost and speed advantages. In addition, initial conversations with our largest and most sophisticated customers indicate potential for additional commercialization opportunities. The success to date with Thomson demonstrates the value of our content, expertise and talent in this AI environment. It also provides important optionality for TR as we work to deliver market-leading and cost-effective AI solutions for our professional markets. I'll now turn it over to Gary for a review of our financial results.

Thanks, Steve. As a reminder, throughout my remarks, I will talk to revenue growth before currency and on an organic basis. Second quarter organic revenues grew 8%. Organic recurring and transactional revenue grew 9% and 11%, respectively, while print revenues declined 3%. Adjusted EBITDA increased 10% to $745 million, with a margin of 38.1%. Moving to the Big 3. Organic revenue growth accelerated to 10% in the second quarter, improving from the 9% pace in recent quarters. Legal Professionals organic revenue accelerated to 10%, as underlying law firm momentum continued and government growth improved sequentially. Key drivers from a product perspective remain Westlaw and CoCounsel Legal. Legal Professionals, excluding government, again, grew 11%, matching the first quarter growth rate and up 9% in the second half of 2025. The strength was broad-based with our large, mid, small law and international subsegments, all at or near record growth rates. Government growth improved to 5% year-over-year from 1% in Q1, though we anticipate a softer growth rate in Q3 as certain transactional revenue in the quarter is not expected to recur at the same level. Our Corporates segment accelerated to 10% organically, up from 9% in recent quarters. Recurring revenue grew 9% and transactions revenue grew by an impressive 24%. Pagero, indirect tax, CLEAR, CoCounsel Legal and our international businesses were key contributors. Tax, Audit & Accounting organic revenue increased 8%. Recurring and transactional revenues grew 9% and 6%, respectively. Our Latin American business, CoCounsel for Tax and audit, SafeSend and the cloud audit suite of offerings were key drivers. The Tax, Audit & Accounting second quarter transactional growth rate fell short of our expectations due in part to timing, but also due to go-to-market execution challenges. We have made several talent additions and leadership changes and expect to get back on track in the second half. In addition, the second quarter growth rate was again impacted by 2 product updates that shifted revenue recognition toward the second half of the year. This was an approximate 1% drag, but is expected to largely normalize in the second half. We continue to expect [ CAP ] revenue growth to accelerate in the second half, driven by rising revenue contribution from our newer AI-driven offerings in the U.S., a key product line extension at Dominio in Brazil, and the benefit from the revenue recognition timing change I just mentioned. Moving to Reuters. Our organic revenue rose 4% for the quarter, driven primarily by growth from the news agreement with the Data and Analytics business segment of LSEG in our agency business. Finally, Global Print revenues decreased 3% on an organic basis. On a consolidated basis, second quarter organic revenues increased 8%, slightly ahead of our expectation from a quarter ago. At the end of Q2, the percent of our annualized contract value, or ACV, from products that are Gen AI-enabled was 32%, up from 30% last quarter. Turning to our profitability. Adjusted EBITDA for the Big 3 segments was $691 million, up 12% from prior year period or 10% constant currency, with a margin of 42.7%. Reuters' adjusted EBITDA was $48 million, with a margin of 20.8%. Global Print's adjusted EBITDA was $42 million, with a margin of 37.7%. In aggregate, total company adjusted EBITDA was $745 million, a 10% increase versus Q2 of 2025, reflecting a 30 basis point year-over-year margin increase to 38.1%. Our Q2 results included $8 million of severance expense related to our initiatives to reimagine [indiscernible]. Turning to earnings per share. Adjusted EPS was $0.99, up from -- up 14% from $0.87 in the prior year period. Currency added $0.01 to adjusted EPS in the quarter. Let me now turn to our free cash flow. For the second quarter, our free cash flow was $727 million, up 29% from $566 million in the prior year period. EBITDA growth and working capital changes were the primary drivers of the year-over-year increase. I'll also provide a quick update on several capital allocation items. We completed our $605 million return of capital transaction on May 4 and repurchased 100 million of our shares in the quarter. In July, we repurchased an additional $238 million, completing the $600 million NCIB announced in February. In aggregate, these transactions have reduced our share count by approximately 3%. We also paid down $500 million of maturing notes in the quarter. Now let add some incremental color on the Global Print transactions mentioned. In mid-July, we reached agreement to sell a 51% stake in our Global Print business to KKR for cash proceeds of approximately $500 million. As Steve indicated, we see this as a positive development as it will leave a stronger and more focused portfolio with improved growth and a higher quality revenue mix. We anticipate the transaction closing in the fourth quarter, subject to satisfaction of regulatory approvals and customary closing conditions. After the close, Global Print will be deconsolidated from our financial statements, with our 49% stake treated as an equity method investment. Beginning with our Q3 results, we intend to report Global Print as discontinued operations in our financial statements. To help with your modeling, we plan to issue a schedule with restated historical results based on this discontinued operations treatment ahead of our Q3 report. As part of the transaction, Thomson Reuters will maintain intellectual property rights and full editorial control over its content portfolio. The joint venture will hold an exclusive license to publish and distribute the content in print, on ProView -- on print and on ProView Global Print's eBook platform. In return, the JV will pay a royalty to Thomson Reuters equivalent to 20% of its professional revenue, which is 85% to 90% of the total Global Print revenue. The royalty will be reported within a new revenue line in our segment reporting. The royalty plus a multiyear transition services agreement will likely offset -- will largely offset stranded costs from the separation. As a result, we see the Global Print transaction being 60 to 70 basis points accretive to organic growth -- revenue growth and approximately neutral to our margins following the transaction close. I'll conclude with my thought -- with a few thoughts on our outlook. Let me start by noting that our guidance is based on the current reporting format, including the Global Print segment. Following the close of the transaction, we will update our outlook and -- to incorporate the financial impact. As Steve outlined, we are raising our full year outlook for both total and organic revenue growth to the high end of the prior 7.5% to 8% ranges, incorporating the stronger first half performance. We are also raising the total and organic revenue growth outlooks for the Big 3 to a range of 9.5% to 10% from the prior approximately 9.5%. Our other outlook metrics remain unchanged. We continue to see 2026 adjusted EBITDA margins of approximately 40%, and we expect free cash flow of approximately $2.1 billion. Turning to the third quarter. We expect organic revenue growth of approximately 8%, and our adjusted EBITDA margin to be approximately 36%. Included in this outlook is an expectation for $19 million of severance expense related to our initiatives to reimagine how we work. We are confident in the full year margin outlook and see strong year-over-year margin expansion in the fourth quarter, driven by the impact of severance actions in recent quarters, growing automation savings, moderating M&A dilution and underlying leverage on our strong revenue growth. Savings from severance actions and automation efforts are expected to be approximately $40 million in the fourth quarter. Now I'll turn it to Gary Bisbee for the Q&A.

Thank you. Jennifer, we're ready to begin the Q&A session.

[Operator Instructions] We'll take our first question from Manav Patnaik with Barclays.

I was just wondering if you could give us a sense of how CoCounsel was doing in terms of its growth rate and size? And just previously, the build out of Thomson, is that -- can be beneficial to CoCounsel ? Like how interconnected are those 2?

Yes. Manav, thanks for the question. So I'll start. Gary will likely add. So a quarter or so ago, we reported that CoCounsel had reached the 1 million user mark. We see healthy growth beyond that. I think equally importantly, though, we see exciting growth in terms of its usage. It's one thing to get it in the hands of people. I think it's another to see daily usage tick up in really healthy ways. And that's across the legal and the tax and order versions of CoCounsel. So that's my first comment. The second comment is we put a completely rebuilt version, fully agentic version of CoCounsel legal into the market in beta form and we're ramping that up as we go through here. We'll be talking a lot about it at the ILTACON conference in a couple of weeks. And the feedback on that has been extremely strong. I haven't seen feedback as strong in my career for a new product. And specifically, the accuracy and the breadth of agentic capabilities, and maybe most significantly, the transparency. It is the opposite of a black box. And for the fiduciary professions that we serve, particularly the legal profession, -- the idea that a young mid-tenure senior lawyer can see that 12 or 20 or 30 steps that the agent is going through and see all the citations and references, and in a sense, access a product that is verifiable, auditable and they're able to validate each and every step, I think that's a step forward for the profession, and that's one of the things that the customers are excited about. So that's the first part of your question. The second part of the question, we're going to put, as I mentioned, tabular analysis, which is an important bulk document analysis feature across onto the Thomson model later this month. And over time, I would see us supporting more and more capabilities as we develop the Thomson model and its capabilities are enhanced. And that will give us, we think, the benefit of reduced latency, so greater speed, more scalability, and a cost advantage. And all of those things, we think will be compelling in the marketplace.

The only thing I would add is, I think relative to the CoCounsel, the pipeline is building nicely here coming out of beta and heading into what we've seen in June and July. And so the team is out there selling all the benefits through, and we're seeing also good conversion early on in that pipeline. So that's the only thing I would add.

Got it. Maybe just asked a slightly different way. I mean the acceleration in the legal organic growth has been impressive the last 2 quarters. Is there any way to disaggregate that growth by the market that you described before seeing any other content through research and then the workflows or any way to just aggregate to help us appreciate the growth rates there?

So -- it's a great question, Manav. The answer today is no. We don't have that ready for you. And I think one of the reasons we don't is that these things are becoming increasingly intertwined. So -- the most exciting thing for us about this agentic AI environment, for example, is it expands the role we play in a meaningful way. So you take a product like Westlaw Advantage and CoCounsel. It gets us into the draft -- it gets us into the drafting business. It gets us into the brief building business. It gets us into sort of advanced litigation analytics. And so we start to do more and more advanced lawyering tasks for the professionals that we serve using our content, using our expertise, using the access to the best AI tools. And so that sort of clear distinction between research and workflow software is meaningfully, and I think, purposely blurred. And that's sort of what's driving the uptick in growth that you cite.

We'll go next to Vince Valentini with TD Cowen.

Can I focus on the margins for a second? If you do 36% in the third quarter, obviously, that's down from 37.7% last year. $19 million in severance would be just about 100 basis points of the impact, but there's still seemingly year-over-year decline even after the severance. So is there anything else onetime in nature, timing-ish in nature that you can point to? And you mentioned earlier the largest ever brand campaign you've done for the next generation of CoCounsel, so maybe there's some somewhat nonrecurring costs there? And secondly, just keying off of that, if your full year guidance hasn't changed, it would still be up around 100 basis points full year. Unless my math is wrong, you need to do just about 45% EBITDA margin in the fourth quarter. I just want to make sure that's what you're telling us is possible given the automation savings from reimagining work plus other benefits of leverage from the good revenue growth you're doing. So if you can clarify those margin things, that would be very helpful.

Yes, you got it, Vincent. Thanks for the question. I'll address your third quarter question first and then go to the full year. So in the third quarter, it's really a continuation of factors we've described in the first half, the severance that we mentioned here in the third quarter. We are continuing to make investments, in addition to the increase in the marketing spend Steve mentioned, to drive automation and innovation. And then we also have some modest dilution from an M&A perspective. So it's the amalgamation of all those things that, like you rightly pointed out, is driving that guide in the third quarter. The thing that I would then pivot to in terms of the full year, margins are progressing largely as we have planned. And so this has kind of been how we thought about the year as we've gone through it. We remain confident in delivering that full year outlook. Year-to-date, we've had a number of targeted investments as we focus on, like I said, innovation and automation. We have absorbed a lot of that acquisition dilution, and there's been meaningful severance. Like I said, in the third quarter, we expect $19 million of severance. In the fourth quarter, to your question, maybe not quite as high as you just referenced, but kind of in the low 40s EBITDA margin is what we have line of sight to. And we expect to deliver $40 million of savings from our efforts to drive automation or reimagine how we work. And we expect to have significantly lower year-on-year severance than last year, we had $19 million of severance in Q4 of 2025. So you kind of put that all together, Vince, what that says is that combining the underlying operating leverage from our healthy revenue growth. And the factors I pointed out, that provides confidence we feel and we do have line of sight to the strong year-over-year margin expansion implied for the fourth quarter and our full year outlook.

We'll go next to Drew McReynolds with RBC.

Yes. First question on the increase in Big 3 organic revenue growth for 2026. Obviously, great to see. Wondering if you could unpack the key driver, key drivers underneath. You're clearly seeing strong transaction revenues on a year-over-year basis. And just wondering, is there a segment that's driving it? Is it transaction revenue? And then, Steve, in your prepared remarks, you talked about repeat transaction revenue. Can you just explain that forth? And then second question, just a follow-up on the Thomson LLM, was it [ trained ] on less than 10% of your content? Is its capability presumably going to grow with a function of training it on more of the content? And what's the gating factor for, just to date, training it on less than 10%?

Yes. I'll handle the growth drivers here quick. Yes, I mean the Corporates segment had a standout transactional revenue growth quarter. And that was ahead of our expectations. They had a bit of an easy compare embedded in that. But Pagero and the global expansion really has accelerated that business. Global trade indirect tax, we're seeing strong traction there. We continue to add agentic capabilities to those platforms, which is driving kind of a new conversation for the field to have and therefore, delivering strong growth. Look, our legal professional business, the size and scope of that, for it to grow and accelerate its growth rate, that's a lot of dollars of growth as well in addition to growing and accelerating kind of from 7, 8, 9 to 10, 11. And so I just want to make sure we all understand that we're driving that kind of growth at scale across that and its Westlaw, and it continues to be CoCounsel as we continue to drive that forward. So those are kind of some of the growth drivers that we saw delivering that growth rate here in the second quarter. Steve, did you want to talk about the other part of the question?

Yes. Well, just to repeat the 2 parts. To repeat transactional revenue through, that was really -- this is the reference to the improving quality of our book of business as we divest 51% of print. As you know, we're very focused on the long-term customer relationships and driving our NPS up and translating that into multiyear agreements. And that forward visibility, we think, is important in terms of the way we manage the business and the way in which investors view us. So that was really just a reference to that shift. In terms of the Thomson model, there's no particular constraint that led us to use less than 10% of the legal content, other than the architecture of the model wanted to create something that is rock solid in terms of its foundation. And as I said, ended up creating something which has produced extraordinarily strong general domain results. And I think that was sort of the order of business in terms of creating the model. So there's no particular constraint in here that we're trying to navigate. What we will now do is continue to invest in that and start to open the spigot in terms of that legal content. We expect the legal specific results to improve as we apply more and more content and expertise to it. And we'll keep you apprised as to sort of what that looks like as the quarters roll through here. Let me make a couple of comments about why we're calling out the Thomson model and why we think there's reason for real sort of optimism and that we're encouraged by what the team has been able to achieve. The first thing, Drew, is I think it speaks to innovation at Thomson Reuters. So -- if you look at the last couple of years, we've been able to put a fully agentic deep research version of Westlaw into the marketplace, which has been, by far and away, the leading legal research product in this deep research environment. We've reinvented CoCounsel legal. And as I said, in response to Manav's question, we're very encouraged by the early feedback there. CoCounsel audit and tax is performing well. We've started to add agentic capabilities to our ONESOURCE, into our transaction compliance portfolio. And then for us to create a large language model, it performs at the level it does in its first version, I think, speaks to us owning our future. I think it gives us a level of sort of leverage with suppliers, that's healthy. And importantly, as we've taken this out in the very early going to customers, our most sophisticated customers are increasingly seeking advanced AI models operating within their own environments, where they retain control over their IP and their data. And this is incredibly important we serve as we do fiduciaries, right? Their preference for what -- what I would call sovereign AI is strongest where they have the concerns over compliance, and those concerns are particularly [ our cue ] and the sort of consequences of IP bleeding out from their environment, but particularly acute. So I think there's an opportunity here for us to meet and exceed that demand for sovereign AI within the legal community within the tax accounting and audit communities using Thomson. And then lastly, I referred to this as -- in response to Manav's question, but I'll just reiterate it. It does provide us some really compelling optionality as a potential to CoCounsel. But we'll start with tabular analysis. We'll run that in August. And then based on our evaluation of that, well then, I think, put more capabilities across -- within the CoCounsel suite to Thomson, and that will give us reduced latency, a significant cost advantage and this ability to provide sovereign AI solutions that will meet the needs of our most sophisticated customers. So look, what I've said to the teams here is I think they've done something amazing, but we're just getting started.

Drew, the only thing I would add to that, just to your -- reoccurring transactional revenue, I would call it reoccurring, if you will, in some of our tax products, a few of our tax products. Well, we call it transactional, it's a repeat kind of year in and year out, and kind of what goes up and down is a little bit of the volume elements of that. And so that's what I would call reoccurring. And again, we have good visibility to that here as we get closer to when that demand shows up.

We'll go next to Andrew Steinerman with JPMorgan.

This is Rohan Kalra on for Andrew Steinerman. I just wanted to touch maybe on LLM costs, maybe how you guys are thinking about this going into the back half and also maybe seeing if there's any interest in shifting to a consumption or subscription and overage model for any of the AI offerings?

Yes. Thanks, Ryan. Great question. So a couple of thoughts, and I'm sure Gary will build here. So the first thing is we've built our agentic solutions to be optimally efficient. And kudos to [ Kirsty ] and [ Joel ] and everybody involved here because we do see more efficient usage of tokens than some of the other sort of competitors and in-market products. So that's one thing. The second thing is, as I just referred to, the Thomson model gives us a lot of optionality here. And if we can reach anywhere near capacity of our GPU [ fact ] compute around that model, we're going to have a meaningful cost advantage, which means we can keep our pricing and our sort of propositions, very, very simple, clean and clear, to our customers. Having said that, I think like many others, we are considering a consumption-based component to pricing, some of our options in the future. And certainly, customers are open to that. And we're making a series of infrastructure investments to support this as soon as the new year. Gary, what would you add?

Yes. I mean I think the only point I would make here is that our AI solutions are typically sold as a premium tier that includes the agentic capabilities. And the pricing for that key AI offering like in Westlaw Advantage is [ comfortably ] covering the growth in the LLM costs and the customer usage here we're seeing in 2026. And that pricing structure that we have today really also is aimed at supporting the cost of growing customer usage over time. So I feel good about the current mechanisms, like Steve said, that are simple for customers to understand, but also providing the right economic outcomes as we move forward.

We'll go next to Aravinda Galappatthige with Canaccord Genuity.

I'll start with a quick follow-up on CoCounsel. Steve, I think you've said in the past that you've observed with many of your law firm clients that the level of experimentation and trying out new products remains high. I know that you speak to some very good feedback on CoCounsel, including sort of the recent beta version. But vis-a-vis the competitors, the startups, any kind of head-to-head feedback that you can share? I'm not sure how easy that is to extract, but I'm keen to maybe hear your thoughts on that. And then in terms of sort of my main question, with the buyback programs completed, maybe just how you're thinking of capital allocation? I know the stock has recovered a bit off the bottom, but obviously, I think in the minds of many, it remains attractively priced. I wanted to hear your thoughts on that as well.

Yes. Thanks, Aravinda. So I'll defer the buyback question to Gary, but let me address the CoCounsel legal question. So I think you've captured the environment correctly up until now, which is law firms, small, medium, large and to some extent, general counsels offices have been experimenting with most multiple tools and have signed up for trials or 1-year agreements. And so it remains a pretty fluid market. And it was with that as the backdrop that we completely rebuilt CoCounsel, and we've done that under the leadership of [ Emily Colbert ] and [ Rabia Ashwath ], who I think have done a wonderful job with the Head Engineer, [ Viola ], in redoing it. And it's given us enough confidence to launch The CoCo campaign and spend some real money in terms of getting the merits of that product into the mind -- the hearts and minds of our customers. It is the first time that we have used the depth and breadth of our content and expertise. So Westlaw Practical Law, 2,600, 2,700 attorney editors and practice experts to train that product. And it's the first fully agentic version. And so the early feedback we're getting, as customers compare it to other offerings in the marketplace, is that it is different -- highly differentiated because of its access, it's native access to our content and our expertise. So it's barely going in terms of launch. It's only literally a number of weeks that it's been in the marketplace. But the customers that have ported across from the prior version of CoCounsel and the new customers to this offering are showing a very, very strong usage growth, which for me is the most important thing that I look for in terms of are people coming back all day long and using the product and getting value from it. And that looks really encouraging. So how it plays out from a competitive landscape, I think, remains to be seen. But we're very optimistic about this new offering. We're going to keep investing behind it with [ bob ] document review capabilities, with collaboration tools increasingly linking it to our other propositions. And bear in mind one last comment. This is white space for us. So this whole sort of legal AI assistant is a white space growth opportunity for us. And we're confident we'll capture more than our fair share as we go through the next 12, 24, 36 months, but it is new spend and a new growth opportunity over and above the existing components of our legal business. Gary, what else would you add on CoCounsel and buybacks?

Yes. On CoCounsel, I would just emphasize Steve's point around usage with just a click down. And so we're seeing the number of interactions go up but also the duration and persistence of those interactions sustained. And so they're in it more often and they're using it for longer in kind of getting to end of job, if you will. And so it's demonstrating a real impact, I think, by those usage patterns. So encouraging for sure. In terms of your question around kind of capital allocation, I'll just kind of go up periscope in a minute and then directly answer your question. So I definitely support a balanced capital allocation approach that TR has followed in recent years. Our objective, as Steve outlined, is to drive innovation, really to beat our competitors, delight our customers while maintaining a key focus on shareholder value creation. Against that objective, our first priority for capital allocation is -- remains investing in our business, in innovation, both internally and through strategic M&A. We remain committed to growing the dividend over time, and we'll consider kind of capital returns through share repurchases and another return of capital to shareholders. And so that's the broad philosophy. And directly related to the rest of the year, but we don't have an approved program in place right now. We just completed that program. We'll continue to evaluate that relative to that prior kind of stated set of priorities and determine if it would be prudent to get back in for a share repurchase or not, but no plans at the moment.

We'll go next to Tim Casey of BMO.

Steve, could you outline how we should think about the Thomson LLM model in terms of discrete product offerings? Like is this going to be a product that is marketed in itself? Or will it power and complement existing platforms? And there has been a lot of -- or some stories of major law firms deciding to do it alone and protect their IP and content. Is the Thomson LLM something that could help them do that? And just -- I'm just trying to -- if you could help us think about how we should place Thomson LLM in your product set.

Yes. Thanks, Tim. So we're working through that now. The first production version eval results came out 10 days ago, and we've been hard at work to sort of look at what's the best way for us to take advantage of that, which we've added and built. And I think it provides at least 2 parts. So one is, as you say, major law firms who want to create and -- a sovereign AI environment and run a version of the model, co-mingle their own information within their own environment and potentially run CoCounsel on the top of that. So that is one part and we're in conversation with a number of firms now around what that might look like and what that sort of primary use cases will be and how we would implement that. The second that I've referred to a couple of times is to build upon tabular analysis running on Thomson and add more and more of the CoCounsel capabilities. And as I said, I think that gives us a sort of a degree of ownership over our future and sort of independence and autonomy as well as speed and cost advantages. And so we'll be thoughtful about that because CoCounsel is working well. And so we don't want to be overly disruptive. But it's pretty exciting as to the options that it puts in front of us.

But notionally, when do you think you'll be in a position to monetize Thomson LLM on a discrete basis?