ExlService Holdings, Inc. (EXLS) Earnings Call Transcript & Summary

Good day, and thank you for standing by. Welcome to the Second Quarter 2024 ExlService Holdings, Inc. Earnings Conference Call. [Operator Instructions] Please be advised that today's conference is being recorded. I would now like to hand the conference over to your first speaker for today, John Kristoff, VP of Investor Relations. John, please go ahead.

Thanks, Felicia. Hello, and thank you for joining EXL's Second Quarter 2024 Financial Results Conference Call. On the call with me today are Rohit Kapoor, Chairman and Chief Executive Officer; and Maurizio Nicolelli, Chief Financial Officer. We hope you've had an opportunity to review the second quarter earnings release we issued this morning. We also have posted an earnings release slide deck and investor fact sheet in the Investor Relations section of our website. As a reminder, some of the matters we'll discuss this morning are forward looking. Please keep in mind that these forward-looking statements are subject to known and unknown risks and uncertainties that could cause actual results to differ materially from those expressed or implied by such statements. Such risks and uncertainties include, but are not limited to, general economic conditions, those factors set forth in today's press release, discussed in the company's periodic reports and other documents filed with the SEC from time to time. EXL assumes no obligation to update the information presented on this conference call today. During our call, we may reference certain non-GAAP financial measures, which we believe provide useful information for investors. Reconciliation of these measures to GAAP can be found in our press release, slide deck and investor fact sheet. And with that, I'll turn the call over to Rohit.

Thanks, John. Good morning, everyone. Welcome to EXL's Second Quarter 2024 Earnings Call. I'm pleased to be with you this morning sharing our strong financial results. In the second quarter, we generated revenue of $448 million, an increase of 11% year-over-year. We also grew second quarter adjusted EPS by 11% to $0.40 per share. The execution of our data and AI led strategy enabled us to accelerate our growth momentum across both our Analytics and Digital Operations & Solutions businesses during the quarter. In Analytics, we delivered revenue of $194 million for the quarter, up 2% sequentially and 6% year-over-year, driven by strong double-digit growth in health care payment services and data management. I am pleased to share that this marks the second consecutive quarter of sequential growth for Analytics, which positions us well to accelerate our growth rate in the second half of the year. Data management continues to be a major focus area for EXL and for our clients on their AI journey. The ability to generate insights and outcomes using AI hinges on accessible quality data, both structured and unstructured. In line with our strategy, I'm thrilled to announce our acquisition of ITI Data, a leading information and data management company serving the world's largest banks, financial services and health care firms. This acquisition not only bolsters our data management capabilities, but also greatly complements EXL's existing vertical markets, expanding our data and AI partner ecosystem and grows our geographic and Global 1000 client footprint. This acquisition is an example of our commitment to further enhance our competitive differentiation by acquiring wider skill sets and deeper expertise in data management, as we strengthen our position as a data and AI-led company. In our Digital Operations & Solutions business during the second quarter, we once again delivered strong double-digit growth as we leverage our domain, data and AI capabilities to win in the market. We grew revenue 4% sequentially and 14% year-over-year to $255 million. This acceleration of growth was driven by our Insurance and Emerging business segments, which grew 16% and 15%, respectively. Our clients continue to focus on cost efficiency and digital transformation, creating a favorable demand environment for us. Last month marked EXL's 25th anniversary, and we commemorated the milestone by ringing the opening bell on NASDAQ. As I reflect on our journey over the past 2.5 decades, our transformation from a traditional outsourcing company to a recognized leader in data and AI has been truly remarkable and stands out. Our strategic decisions to make timely pivots and evolve into a data and AI led company has contributed to our industry-leading growth rates. More importantly, we have positioned EXL for success in the rapidly evolving and expanding AI era. As we shared in our Investor Strategy Update event in May, AI has changed the business transformation landscape from one that was historically technology led to one that is now increasingly domain and data led. This plays to our core strengths in domain expertise, data mastery and AI implementation capabilities. It has significantly expanded our total addressable market and accelerated our growth rates. Data and AI-led revenue now accounts for more than half of the company's total revenue, and we anticipate that percentage to grow over time. We recently announced a strategic collaboration with NVIDIA to create enterprise-wide data and AI applications for insurance, healthcare, banking, retail and other industries. EXL's adoption of advanced AI technologies powered by NVIDIA AI helps our clients with fast and scalable AI in complex enterprise production environments. As a leader in helping clients redesign customer journeys and reinvent business models by integrating data, analytics and AI directly into critical workflows, EXL will play a pivotal role in fine-tuning and applying the NVIDIA AI platform to highly specialized use cases. I am happy to share that our first insurance-specific LLM, which has been fine-tuned on 2 billion tokens curated from insurance domain data is now production ready. Our fine-tuned model outperforms several foundational models such as GPT 4.0, Claude Sonnet and Llama-3 70 billion on insurance-specific tasks in the medical claims space. We leveraged a number of proprietary data preprocessing and fine-tuning techniques in close collaboration with NVIDIA's engineering teams to achieve this outcome. Given our success, we will continue to create additional domain-specific LLMs across our chosen industries. Our growing ability to tangibly improve our clients' businesses through domain, data and AI is reflected in the continued strength of our sales pipeline. We are particularly encouraged by the year-over-year growth we have experienced in the data and AI portion of our pipeline. Let me share a couple of recent examples of how we are harnessing our differentiated capabilities to deliver meaningful value for our clients. We partnered with a global personal and commercial lines insurer to help them improve underwriter utilization and drive higher premium growth. As part of the engagement, we deployed our SaaS-based underwriting solution on our proprietary AI platform. This included our patented gen AI-based XTRAKTO.AI solution for structured and unstructured data processing, an advanced scoring engine to generate a risk score for all incoming leads and automated decision-making algorithms. The whole underwriting process now just takes a few hours where it used to take multiple days. This solution not only improves our client quote to policy conversion, but it also helps them improve the broker experience and drives better underwriter productivity and higher premium growth. This is a great example of how EXL is combining our insurance domain expertise along with our data and AI solutions to help our clients transform their operating model. In another example, we have been working with a Fortune 50 consumer goods company to help them solve challenges in their rapidly growing e-commerce business. They were experiencing a surge in data volume, which has strained their internal resources, resulting in frequent data pipeline failures and unreliable data quality. Leveraging our data engineering experience and capabilities, we help the clients streamline data flow and improve data quality. This included combining data in various disparate formats across supply chain, marketing and sales into a comprehensive data warehouse. We also optimized their data extraction algorithms and developed various actionable executive dashboards empowering leaders to make informed decisions faster. This resulted in a reduction in pipeline failures of approximately 95% with consistent data availability and smooth processing. In addition, we were able to reduce their computing costs related to extracting, transforming and loading data for those pipelines by over 1/3. These impressive results demonstrate our ability to leverage our deep data management capabilities to become the data and AI partner of choice for some of the largest companies in the world. We continue to see growing opportunities in data management and engineering as more clients focus on complex foundational work necessary to prepare for broader AI deployment in the future. And we continue to successfully attract new specialized talents to EXL as well as train and develop our existing talent in data and AI. In summary, we delivered strong results in the second quarter, and we are encouraged by the acceleration of growth across our analytics and digital operations and solutions businesses. The consistent execution of our data and AI-led strategy, combined with our growing data and AI sales pipeline puts us in a strong position for further growth in the second half of the year. With that, I'll turn the call over to Maurizio to cover our financial performance in detail.

Thank you, Rohit, and thanks, everyone, for joining us this morning. I will provide insights into our financial performance for the second quarter and the first 6 months of 2024, followed by our revised outlook for the full year. We delivered a strong second quarter with revenue of $448.4 million, up 10.7% year-over-year on a reported basis. On a constant currency basis, we grew revenue 10.8% year-over-year and 2.8% sequentially. All revenue growth percentages mentioned hereafter are on a constant currency basis. Revenue from our Digital Operations & Solutions businesses as defined by three reportable segments, excluding Analytics, was $254.6 million, which represents year-over-year growth of 14.3%. Sequentially from the first quarter of 2024, we grew revenue 3.6%. In the Insurance segment, we generated revenue of $149.3 million, an increase of 16.2% year-over-year and 2.8% sequentially. This growth was driven by the expansion of existing client relationships and new client wins. The Insurance vertical consisting of both our Digital Operations & Solutions and Analytics businesses grew 12.9% year-over-year with revenue of $185.7 million. In the Emerging segment, we grew revenue 15.1% year-over-year and 4% sequentially to $77.2 million. This growth was primarily driven by the expansion of existing client relationships. The Emerging vertical consisting of both our Digital Operations & Solutions and Analytics businesses grew 5% year-over-year with revenue of $159.7 million. The Healthcare segment reported revenue of $28.1 million, growing 3.5% year-over-year and 7% sequentially. This growth was driven by higher volumes in our Clinical Services business. The Healthcare vertical, consisting of our Digital Operations & Solutions and Analytics businesses, grew 16.8% year-over-year with revenue of $103 million. In the Analytics segment, we generated revenue of $193.8 million, up 6.4% year-over-year and 1.7% sequentially. Growth in Analytics was driven by higher volumes in payment services and growth in our data management business. SG&A expenses as a percentage of revenue increased by 230 basis points year-over-year to 20.5% driven by investments in sales and marketing, generative AI and digital solutions as well as restructuring and litigation settlement costs. In the second quarter, we incurred $6.2 million in restructuring and litigation settlement costs to realign a portion of our employee workforce to better meet the evolving needs of our clients. This was largely completed in the second quarter. We do not expect to incur any additional material costs in the third quarter. Our adjusted operating margin for the quarter was 19.8%, down 20 basis points year-over-year, primarily due to increased SG&A investments. Our effective tax rate for the quarter was 23.2%, down 70 basis points year-over-year. This was driven by higher profits in lower tax jurisdictions. Our adjusted EPS for the quarter was $0.40, a 10.8% increase year-over-year on a reported basis. Turning to our first half performance. Our revenue for the period was $884.9 million, up 9.8% year-over-year on a reported and constant currency basis. This growth was driven by both our Digital Operations & Solutions and Analytics businesses. Adjusted operating margin for the period was 19.4%, down 30 basis points year-over-year. Our first half adjusted EPS was $0.78, up 9.9% year-over-year on a reported basis. Our balance sheet remained strong. Our cash, including short- and long-term investments, as of June 30 was $285 million, and our revolver debt was $335 million for a net debt position of $50 million. We generated cash flow from operations of $75 million in the second quarter compared to $48 million for the same period in 2023, driven by improved working capital management. During the first 6 months, we spent $23 million on capital expenditures and repurchased approximately 4.3 million shares at an average cost of $30 per share for a total of $128.3 million. This includes 3.35 million shares received upfront as part of our previously announced $125 million accelerated share repurchase plan. We received the remaining shares in July. Now moving on to our outlook for 2024. Although the macroeconomic environment remains unpredictable, we are raising our guidance range for revenue based on our strong first 6 months results and the recent acquisition of ITI Data. We now anticipate revenue to be in the range of $1.805 billion to $1.83 billion, representing year-over-year growth of 11% to 12% on both a reported and constant currency basis. This represents an increase of $13 million at the midpoint from our previous guidance of $1.79 billion to $1.82 billion. The current revenue guidance includes $7 million to $9 million from ITI Data for the remaining 5 months of 2024. We expect a foreign exchange gain of approximately $1 million, net interest expense of approximately $5 million to $6 million and our full year effective tax rate to be in the range of 23% to 24%. Based on this, we anticipate our adjusted EPS to be in the range of $1.59 to $1.62, representing year-over-year growth of 11% to 13%, which is an increase from our prior adjusted EPS guidance of $1.58 to $1.62. The ITI Data acquisition is expected to be neutral to adjusted EPS in 2024. We expect capital expenditures to be in the range of $50 million to $55 million. In summary, we are pleased with our second quarter results and encouraged by the acceleration of our growth momentum. By continuing to execute on our data and AI-led strategy, we are confident in our ability to maintain our double-digit growth trajectory. With that, Rohit and I would now be happy to take your questions.

[Operator Instructions] The first call comes from the line of Maggie Nolan of William Blair.

I'm hoping you can break down your Analytics segment in a little bit more detail for us, where you're seeing incremental tailwinds versus where you're seeing headwinds by kind of offering and vertical alignment within the Analytics segment as well.

Sure, Maggie. So as we mentioned, the places where we are seeing a tremendous amount of growth is around some of the work that we do on healthcare payment services as well as data management. Both of these continue to drive the growth in our Analytics business very nicely in double digits. We are also seeing some of the strength come back across some of the industry verticals. So what do we do in Analytics, particularly in some of the other industry verticals like retail and banking, and that's encouraging for us. And then as we mentioned previously, marketing analytics for us continues to be a tailwind and that's something which has been a drag in terms of our growth rate for the second quarter.

And can you talk in a little bit more detail about the adjusted operating margin expansion expectations on a quarterly basis and in particular, what we should expect in terms of SG&A investments over the remainder of the year?

Sure, Maggie. So when you take a look at our adjusted operating margin, in 2023, our margin expanded 100 basis points to 19.3% and we had guided this year to be flat -- fairly flat with 2023. If you look at the first half of the year, we ended the first half of the year at 19.4%. We had a lower first quarter because of a number of investments that we planned on making and we made that brought that AOPM down to 18.9%. But now that we've made those investments in the first quarter, we get to much more of a normalized AOPM for the rest of the year. So you can anticipate or expect the second half of the year AOPM to be right in line with our guidance to being flat with 2023, right around that 19.3%, 19.4% range. In terms of SG&A investments, we made a number of investments, both in AI and also in our front-end sales in the first half of the year. So our second half investments in SG&A should be fairly reasonable, nothing significant in SG&A.

The next question comes from the line of Bryan Bergin of TD Cowen.

I mean on ITI, can you give more detail on how that target complements your existing data management practice? I'm curious if it's about -- more about scale or more so additive solutions and kind of the industry penetration? And then just on the financial side, has that -- can you comment on how growth has been performing there?

Sure. So first of all, we are really excited to welcome the ITI team to become part of EXL. As we've kind of gotten to know them, we find that there's exceptional talent within ITI Data. I think what it does for us is from a strategic lens, it further strengthens our capability in data management, and it allows us to be able to provide a wider array of services to our clients around data management. It also brings to us a much more Global 1000 customer base. And so a number of new large client relationships that we think we can leverage together and add to the capabilities out there. It also brings some IP as well as some partnerships, which we think would be additive and would be helpful. And then finally, it's got a very nice global mix of business that complements the capabilities that EXL has. So we are excited about this combination. And we think with this combination, we should be able to grow the business at a much faster pace and be able to add to the service offerings that we currently have for our clients around data management.

Okay. And then my follow-up, just on the gen AI front. The insurance specific LLM you trained here is interesting. Can you comment on how you intend to contract engagements that may leverage that as well as just any other details on kind of the productivity and delivery implications?

Yes. So that's exciting for us. It's our first domain-specific trained LLM and we naturally chose to do that in our insurance vertical, which is where we've got the maximum amount of domain knowledge as well as access to data. And I think the -- there are a number of things that we are learning as we are experimenting with this and as we are building up these LLMs. Number one, what we are finding is, foundational models are good. But when you have to deploy them into the workflow, these need to be specifically trained on the domain and data from that domain and that's what allows them to be a lot more effective and be able to be a lot more productive. So one of the things which we have done is to create our own insurance LLM, which can now be a plug and play. So it actually allows us to be able to offer that service to multiple clients simultaneously. And it requires a onetime effort upfront, which is what we have done and trained that LLM on 2 billion parameters. And we are finding that the efficacy of that LLM is far superior to the foundational models that we are kind of testing it out against and so it allows us to be able to deploy the solution with speed, at much lower cost and get great results. I think in terms of the commercial model, that's something which is going to evolve as we go forward. It will depend upon the kind of engagement that our clients want to kind of have with us. If they want to have an engagement, which is a total outsourcing engagement, we will embed this as part of the cost structure and be able to offer them a benefit and an outcome and it might be a commercial model that integrates in our operations management capabilities, our analytics capabilities and the LLM. In other situations where clients may want to adopt this as a stand-alone capability, there would be perhaps a licensing agreement as well as a payment for the use of this LLM. And obviously, this is just the beginning for us. We hope to be able to continue to kind of keep improving on this and also building out LLMs in other industry verticals. I think our viewpoint is that, that's where the market is headed. And we are excited that we are able to kind of take the lead in terms of going forward with one of these LLMs and be able to demonstrate its performance and its efficacy.

The next question comes from the line of Surinder Thind of Jefferies.

Rohit, just following up on the prior question about proprietary LLMs that are industry specific. Can you help me understand how the training of the models work in the sense that where are you getting that data from? And then are you able to then go to the client? And does it have to be further trained with client data? Or where is the IP for some of that exists?

Sure, Surinder. So first off, as you know, for us, we go deep into industry verticals, and we work with multiple clients across each of our industry verticals. So take insurance as an example. We work with several leading carriers in that space, and we have a deep knowledge and understanding about the workflow, the processes, the data and everything that needs to happen there. Second, we do have contractual rights to be able to use the data across some of our client relationships. And those situations where we have permissioning to be able to use the data as well as to be able to use derivative actionable data, we're basically using both of this. So we are using the primary data as well as some of the derivative data to be able to train our models on these insurance processes. Now going forward, in terms of the work that we will do, certainly, in terms of areas where we have invested in terms of creating that model, and we have trained that data, that IP will belong to us. But as we deploy it in client workflows, that IP will jointly be owned by us and by our clients. So it's going to be no different than other technology solutions that we create or we deploy across our client portfolio. But I think the exciting part out here is the efficacy that we are able to get, I mean, that is beating the foundational models, which are trained on much, much more larger parameters. I think that's what is exciting for us and for our clients.

That's helpful. And then just in terms of a bigger picture question here. I think you've noted that the current environment from a digital ops perspective has been highly supportive with considerations around cost and efficiency. But as we move towards maybe rates coming down, perhaps a bit more growth on the discretionary spend in part, how are you thinking about the potential impact on the digital ops business? So how much of that tailwind should go away as we think over the next 2, 3 years or whenever the demand normalizes?

Sure. So fundamentally, when we think about our business and what we are helping our clients with at the highest level, there are two things that we are helping them with. Number one is helping them drive down cost; and number two, helping them transform their businesses. That's what we focus in on the digital operations side. We've said always that for us, the demand environment continues to be strong as long as the penetration is low and the adoption of our capabilities can be used to leverage for clients being able to benefit from the cost reduction as well as transformation of their operating processes. In an economic environment where the growth rates are slowing down, the interest rates are coming down, the propensity of clients to focus in on a lot more cost savings becomes even more significant. So frankly, at that point of time, we would expect more and more immediacy in terms of client actions outsourcing their work and entrusting that work to us so that we can actually lower and reduce their cost structure in a very significant way. In terms of transformation, the way in which clients are approaching this is they want to be able to get the benefit of the transformation in the current period itself. And therefore, they want to work with those providers that have the necessary tools and the skill sets and the capabilities that can be deployed immediately where the return on investment can be very, very high, very quickly. So I think we end up being a very credible partner for our clients in terms of helping them engage on the transformation initiative as well. So that's why I think for us, the Digital Operations & Solutions business is strong, and we would expect it to continue to remain strong. Our expectation is that our growth rate has accelerated in the second quarter, both in analytics and in digital operations and we do expect that acceleration to continue in the second half of this year.

The next question comes from the line of Puneet Jain of JPMorgan.

I have like a follow-up on digital ops. So it seems like in the industry, clients' in-house operations are ramping up on an overall basis. Are you seeing that trend in operations management as well? And broadly, what do you expect for in-sourcing versus outsourcing decisions by your clients in digital ops?

Puneet, yes, I think clients do use multiple avenues for being able to manage and run their operations efficiently and certainly creating in-house operations or captive operations or global captive centers, that is certainly part of that portfolio. I think the place where we tend to have an advantage is where there is a leading and cutting-edge technology that needs to be applied, where data flows need to be integrated in the customer workflows, where AI needs to be deployed, that's where I think we have an advantage because we tend to see this across multiple clients and across multiple use cases, and therefore, our understanding of being able to deploy this effectively in a shorter time period and quickly tends to be a little bit better. So it actually works quite well because we end up complementing some of the work that the in-house captives are doing. And many times, we might do the work initially and subsequently, the captive might do that work or vice versa, right? So it actually is a good ecosystem. I think the key is, can we continue to remain competitive and continue to innovate and continue to be adding value to our client relationships across the board. And I think by virtue of us going deep into the domain, by going deep into data, by going deep into AI, which is our strategy, we are able to maintain that differentiation and that competitive edge.

Got it. No, that makes sense. And then you had like accelerated repurchase this year just completed an acquisition. From here on, how should we think about the use of cash, and also, like how much you are paying for the ITI acquisition?

Sure, Puneet. So to answer your latter question, the ITI acquisition was $26 million in cash that we purchased the company for. So we use that from existing cash resources. Going forward, we have been pretty aggressive on share repurchase in the first 7 months of the year. We closed the ASR in July. And we've repurchased right around $160 million worth of stock so far this year right around $30 per share. So it's been very -- we've been fairly aggressive, and we've done a good price on our share repurchases for the year. We got approval of $0.5 billion back in March by the Board for the next 2 years for share repurchases. So at least for 2025, our minimum for share repurchases that we would plan on is at least half that number, which would require us to do about another $90 million in share repurchases in the final 5 months of the year. So when we allocate capital for the remainder of 2024, you'll see a portion of that go towards share repurchases and then any other opportunities that come up on the M&A side going forward.

[Operator Instructions] The next question comes from the line of Dave Koning of Baird.

Nice quarter. And I guess my first question, new wins were really strong, new client wins in the quarter. I think the best I think we've ever seen, certainly a lot higher than many of the recent quarters. I guess, is that an indication in part that macro is getting a lot better? And maybe what are the size of those? Is the size also bigger? And what's the lag again? Like when you win this quarter a lot, is that kind of really 2025 impact more so than the next 6 months?

Yes. Thanks, Dave. So we're very pleased with the number of new client wins that we had in the second quarter. And also, if you saw our first quarter new client wins was pretty healthy. So I think in the first half of the year, we've signed up 39 new clients, 23 of them coming in the second quarter. I think what this is showing us is that the adoption of our business model and working with clients seems to be resonating nicely. And therefore, we have been able to widen the number of client relationships that we work with. We're also very happy about the quality of these new clients that we are signing up because some of these are very significant client relationships, and we expect to be able to do a very significant and meaningful amount of revenue with them. But as you correctly noted, most of this is going to be a revenue growth that takes place in 2025 and beyond. We get very little growth from new clients in the current period. Keep in mind that we've had a good track record of signing up new clients in all of '23 and '24 as well. And some of the clients that we signed up in '23, we are still in the process of ramping those up right now and into the second half of '24. So that should be adding on to our volume of business going forward.

Got you. And maybe as my follow-up, you guys don't do -- you guys have very, very limited like restructuring type actions over time. You've super clean financials for many years and this was pretty small too, the $6 million or so. But maybe describe a little more. Was this part of why employee growth was a little slower sequentially than normal? And maybe what are the savings going to be around it? Or just maybe a little more data around it.

Sure, Dave. So in the second quarter, we did have a restructuring and legal settlement charge of $6.2 million. The restructuring piece was $4.8 million. It involved about 450 employees that were affected by the restructuring, so less than 1% of the overall workforce. And it was really just to realign a small portion of our workforce during the quarter. And so it did contribute to a slower headcount growth number during the second quarter. We grew just slightly north of 800 employees during the quarter, and so they did contribute to that. But it was really just a onetime exercise in the second quarter, and we don't see that reoccurring at all for the rest of the year.

The next question comes from the line of David Grossman of Stifel.

I just wanted to follow up on two of the questions that were already asked. The first one was just, Rohit, going back to the backlog and the pipeline. You talked about the pipeline at least for data analytics being fairly strong and AI. I'm just curious on the ops -- digital ops side, what does the backlog look like year-over-year and as well as the pipeline and where are we in terms of rolling out '23 wins in '24? Do you think that's going to extend into '25? Or do you think most of those contracts will ramp in '24?

Sure, David. So the backlog and the pipeline for us across the board continues to be pretty healthy. It continues to be strong. We're not really seeing any material change in that. I think for us, in terms of the implementation of deals that we've already won, we've only partially implemented the backlog that we had at the beginning of the year. We've got a tremendous amount of backlog that we still need to implement on in the second half of this year. And there are a number of large client wins that we will be implementing in the second half of the year. The pipeline for us is basically strong across verticals. But what I would say is while the large deal pipeline within our overall pipeline, that ratio is about the same as what it's been previously. The place where we've seen, obviously, a tremendous amount of growth in the pipeline is the pipeline attributable to gen AI and that's increased in a pretty material way over the last 12 months. So we are encouraged by the quality of that pipeline and the areas in which clients want to partner with us.

And how should we think about the revenue conversion of a gen AI deal? Is that fairly small and quick turn or are those going to be a longer duration type relationships?

Yes. So I would say that what we are seeing is a bipolar effect on the gen AI part. In some cases, we are seeing gen AI being embedded into the operations and into the workflow and the client outsourcing the entire operation to us along with the gen AI implementation, and those tend to be the large-sized deals. And on the other side of the spectrum, they want us to implement gen AI solutions into existing operations, which they might be running or some of our competitors might be running and those tend to be typically much smaller implementations as such. So we are seeing deals at both ends of the spectrum that are there. Some which are bundled in with the operations, which tend to be large, and some of which are pure gen AI implementations, which tend to be much smaller in size.

And just as a follow-up on your comments about your domain-specific model, I think you mentioned that how you're going to monetize that is still a little bit up in the air. However, as you look beyond the near-term, Rohit, how do you think a domain-specific model impacts your business longer term? Over time, the market has a tendency, at least the BPO market has a tendency to price in a lot of this, and they want to recapture a lot of the benefit, the client wants to recapture a lot of the benefits. So is there something different here that would suggest that EXL can retain a larger piece of the economics as it relates to these models? Or do you think it follows a more traditional pattern?

Yes. So I think you're right. I think there will always be a tussle between clients and service providers in terms of retaining economic value. But I think where we stand out is our focus and our expertise and our domain knowledge in select industry verticals. I think in those areas where we have deep knowledge and we have deep access to data, our ability to train these models and to be able to leverage these models will perhaps be a little bit better. And the networking effect of being able to see workflows across multiple clients, I think that is very powerful when the use of gen AI is involved. So I think that's what gives us a little bit of an advantage and a little bit of pricing power that can leverage the value that we can create out there.

I see. If I could just squeeze one more in. Maurizio, can you give us what -- post all the share repurchases to date, what the fully diluted share count should look like?

So David, so far, we have purchased right around 4.5 million to 5 million shares overall. So if you deduct that from our overall share count going forward in 2025, you're going to find us right around -- you can calculate what that reduction in the share count will be. But that's what we've done so far.

So you're talking about 4.5 million versus where you were at the beginning of the year then?

No, you'd have to average that out for the year. You're going to get a much -- when you do the diluted average shares for the year, and you do your calculation, you're going to have to weigh that over the year. But we've done that in the first 7 months of the year. So you have to break that -- you have to average that out over the first 7 months and calculate and recalculate that. So you're not going to get that whole benefit this year. You'll get that next year.

The last question comes from the line of Vincent Colicchio of Barrington Research.

Yes, Rohit. The ITI deal looks quite like a value add. You haven't been very acquisitive for some time. Will we see more ITI type deals as a way to accelerate your positioning?

Yes, Vincent. I think we've been very clear that for us, we'd love to be able to grow our business organically and inorganically. And frankly, the amount of cash flow that we generate and the strength of our balance sheet allows us to be in a fortunate position of being able to acquire assets. But we're going to be very selective in terms of the strategic fit, the financial discipline and the cultural fit of the assets that we acquire. And therefore, we do think that there's a lot of opportunity for us to be able to deploy capital and integrate in assets. And as things continue to evolve and progress, I think you should be expecting to see us continue to be active in this area.

And how should we think about this NVIDIA partnership in terms of how long it will take to meaningfully benefit the company?

I think it's a very strategic partnership, which -- where we are going to be making a tremendous amount of investment of talent and resources, time and commitment and building up capabilities on the NVIDIA AI stack. We think their software and their platform is actually highly differentiated and provides for much better business outcomes. So as we deploy it and jointly go to market, I think you're going to see us being able to kind of benefit from that and our clients being able to benefit from that. Traditionally, NVIDIA has been seen as a hardware company and as a gaming company and their use of their advanced computing capabilities into the workflow is something that we think we can be very, very additive on. And so for us and for them, I think it's going to be a very meaningful partnership.

This concludes the question-and-answer session. I would now like to turn it back to John Kristoff.

Okay. Thank you, everyone, for joining our call today. And as always, feel free to reach out directly to me with follow-up questions, and hope everyone has a great day. Thanks, and bye-bye.