Grid Dynamics Holdings, Inc. (NASDAQ:GDYN) Q2 2025 Earnings Call Transcript

Grid Dynamics Holdings, Inc. (NASDAQ:GDYN) Q2 2025 Earnings Call Transcript August 1, 2025

Cary Savas: Good afternoon, everyone. Welcome to Grid Dynamics Second Quarter 2025 Earnings Conference Call. I’m Cary Savas, Director of Branding and Communications. [Operator Instructions] Joining us on the call today are CEO, Leonard Livschitz; CFO, Anil Doradla; CTO, Eugene Steinberg; COO, Yury Gryzlov; and SVP Americas, Vasily Sizov. Following the prepared remarks, we will open the call to your questions. Please note that today’s conference call is being recorded. Before we begin, I would like to remind everyone that today’s discussion will contain forward-looking statements. This includes our business and financial outlook and the answers to some of your questions. Such statements are subject to the risks and uncertainties as described in the company’s earnings release and other filings with the SEC.

During this call, we will discuss certain non-GAAP measures of our performance. GAAP to non-GAAP financial reconciliations and supplemental financial information are provided in the earnings press release and the 8-K filed with the SEC. You can find all the information I just described in the Investor Relations section of our website. I’ll now turn the call over to Leonard, our CEO.

Leonard Livschitz: Thank you, Cary. Good afternoon, everyone, and thank you for joining us today. I’m delighted to report another record quarter in revenue. Our second quarter revenue of $101 million was another all-time high, driven by the continued growth in our engineering billing headcount. More importantly, we’re witnessing a strong pipeline of opportunities across industry verticals. I will talk more about it in my prepared remarks. Grid Dynamics is aligning every aspect of its business with an AI-first approach. This includes infusing AI into go-to-market strategies, service offering, delivery and talent management. We’re doing that while preserving and expanding our core assets around high-caliber technology, consulting and engineering services.

While traditional programs face increased scrutiny, innovation-centric initiatives are being prioritized from a spending perspective. Enterprises are actively seeking AI-native partners capable of driving and leading adoption within the enterprise environment. Furthermore, traditional functional structure within large enterprises are often lacking the adaptability needed for efficient cross- functional decision-making regarding AI implementations encompassing both technology and business aspects. This is precisely where Grid Dynamics plays a crucial role, empowering organizations to accelerate AI adoption at enterprise scale. I’m happy to report that the first half of 2025, AI and Data was 23% of the company’s overall organic growth. The AI and Data practice is growing almost 3x faster than our overall organic business.

I’m excited to see the growth in pipeline of opportunities as we enter the quarter with accelerated business momentum. This is the basis for our positive business outlook, even though macroeconomic uncertainties persist. I’m also pleased to report on the progress with our recent acquisitions. JUXT has significantly elevated our industry expertise in banking and financial services, attracting considerable interest from global banking based in the United States. In the second quarter, a U.S.-based global bank continued to be a top 10 customer. And that was the reason the financial services vertical remained our second largest. Mobile Computing has enhanced our follow-the-sun capabilities and talent acquisition efforts, successfully integrating engineering teams to support our U.S. enterprise accounts.

Q&A Session

Our partnership-influenced revenues reached 17.9% of the total revenue in Q2 2025, and we continue to experience increased traction with all hyperscalers, notably only with Google. In our European business, we begin implementing a modular B2B digital search solution built on Microsoft Azure for one of the largest worldwide brewing companies. We also have launched AI expert agents at a Tier 1 investment bank to perform in-depth code quality and security reviews as a part of the software development life cycle. Our India expansion continues to be a strategic highlight. India is now among our 2 top countries by the headcount and has emerged as a hub for multi-agent, multi-modal platform engineering, demonstrating a strong talent attraction and upskilling.

Our internship program also saw strong momentum with over 16,000 applicants and high placement rates into the billable roles. Across the majority of our customers, there is a profound impact of AI on the way they are planning their future initiatives and programs. Customers now expect a flavoring of AI across all of their service offerings, even traditional ones. We firmly believe the workforce pyramid of the IT industry space is shifting to our senior talent and AI-centric agents. As you know, Grid Dynamics workforce pyramid is more weighted towards senior more experienced engineers in comparison to our peers in IT industry. Our alignment in the workforce, along with the technology-centric DNA, positions us well as enterprises embrace AI. Given the critical role of AI, I would like to emphasize Grid Dynamics’ unique market position.

It’s important for investors to remember that our company’s core values have built upon a strong foundation in data and data platforms as well as expertise in large-scale data engineering for Fortune 1000 enterprises. I will now turn the call over to Eugene Steinberg, our CTO, to elaborate on the important topic of AI, Eugene?

Eugene Steinberg: Thank you, Leonard. Good afternoon, everyone. I’m delighted to share how Grid Dynamics is actively retooling for an AI-first future, where AI capabilities are embedded in every aspect of our operations and service delivery from the ground up rather than added as an afterthought. We are methodically building on a strong foundation of a decade of data and AI experience. We are expanding our key AI capabilities and strategic partnerships. We are delivering production-ready solutions with proven ROI for enterprise clients. We are innovating with major customers and building considerable experience across Agentic AI platforms and solutions and AI-first Software Delivery Lifecycle. The investments we made are yielding positive results, many of which I’ll discuss today.

Grid Dynamics’s whole AI framework is based on 4 foundational pillars. Let me walk you through each. First, AI-powered business transformation. We are delivering immediate and measurable impacts from our engagements in the areas of customer engagement, enterprise operations and manufacturing. Conversational commerce is redefining customer engagement by driving hyper personalized customer experiences. Our search solutions have become a key entry point for many new client relationships in retail and CPG industries. From there, we often expand to build conversational commerce capabilities. Our efforts routinely yield conversion improvements of over 5%. This success leads to a follow-on engagement that average 2, 3x the initial project value as clients expand the AI adoption across additional business units.

We have specialized domain solutions for many sub-verticals that have been particularly differentiating. Take auto-parts search, for example. Our fitted auto-parts search capabilities have established Grid Dynamics as a preferred partner for most leading auto-parts retailers. With over 150 AI search specialists deployed across customer projects, we are demonstrating our ability to grow with an existing accounts while delivering measurable business impact. Agentic AI significantly enhanced efficiency for a major financial services company by automating intelligent processes in facilitating data-driven decisions. Previously, the cost of covering the vast number of Tier 3 customers was prohibitive for digital sales. Our B2B customer 360 agent now conducts exhaustive research, aggregating client data from diverse sources such as CRM, contract databases and spreadsheets.

These detailed profiles integrate automated risk alerts, AI-powered insights and intelligent recommendations informed by prior interactions, ultimately leading to improved customer retention and business growth. This initiative is expected to free up about 20% of seller capacity, allowing for high-touch approach with more clients and accelerated time to revenue. Within the manufacturing sector, we implemented the remaining useful life prediction system for a leading industrial equipment manufacturer, which enhances maintenance planning and reduces unplanned downtime. We have also delivered facility modeling with G-code generation for a global technology company, optimizing manufacturing processes and improving production efficiency. The rise of physical AI is fundamentally transforming the industrial robotics landscape, leading to the replacement of legacy robotic platforms with modern AI enabled solutions.

We collaborate with innovative platform providers such as Wandelbots, enhancing their offerings with advanced AI components for inspection, welding and precision manufacturing applications. This represents a new and promising revenue stream for our company. Second, AI and agentic platforms. We partner with large enterprises to develop in-house bespoke agentic AI platforms. For instance, we are collaborating with a leading global payment technology company and a multinational beverage giant to construct comprehensive AI platforms. These platforms empower our clients to create a full spectrum of AI agents both low-code and high-code within a secure, scalable environment. They offer an expanding ecosystem of tools for agents to access enterprise data and systems.

This platform-first strategy enables us to see substantial expansion opportunities by building AI solutions atop of the platforms we develop. Third, AI-first SDLC. As enterprises embrace AI-first mentality, the entire approach to the software development life cycle, SDLC is shifting. Last month, we introduced our proprietary AI-centric Grid Dynamics AI-Native or GAIN engagement model, and we’re driving strong adoption of AI-first SDLC methodologies across the dynamics. What is particularly exciting is that this enables our expansion into previously unaccessible market segment. Labor-intensive legacy modernization projects that additionally require large volume of relatively low skilled labors are now within our reach. This represents a significant market expansion opportunity as we can now compete for projects that were previously economically unfeasible.

For example, we are migrating 16,000 data processing jobs for a global technology leader using a small specialized team equipped with AI-first SDLC tooling. AI-first SDLC has dramatically improved our presales and client acquisition. We now create high-quality proof of concepts and prototypes in hours, not weeks, significantly boosting conversion rates and shortening sales cycles. For example, when a leading specialty pet retailer requested a computer digital solution to automate fish counting in aquariums, previously requiring manual fish transfer between tanks, our AI-first development team delivered a working prototype the next day. And finally, fourth, Grid Dynamics process efficiency. Beyond client-facing applications, we are leveraging AI to transform our own internal operations.

Our in-house Agentic AI platform is transforming and automating every aspect of our operations, including knowledge management, talent sourcing, project management, contract reviews and HR functions. AI is fundamental to driving our client business forward. Our continued commitment to the AI-first future is unwaering, and I am excited about the road ahead. I will now turn the call over to Vasily Sizov, our SVP of Americas to discuss some notable project highlights from the quarter.

Vasily Sizov: Thank you, Eugene. Good afternoon, everyone. I am pleased to highlight some important accomplishments from the quarter that illustrate the value of our work. For a leading global technology company, we modernize their data processing infrastructure by migrating Spark and Scala workflows from a legacy scheduling system to a next-generation cloud platform. We developed a comprehensive data validation framework to ensure data consistency, optimize compute resource usage and created reusable templates that accelerate future migrations to containerized environments. This initiative significantly improves platform stability and performance, reduced operational risks and established the foundation for scalable, efficient development of future data- driven capabilities.

Another example, we partnered with a leading multinational technology company to develop hermetic C++ toolchains for their ML portfolio. This foundational initiative established a highly reproducible, reliable and efficient C++ build environment across their machine learning programs. Our team led the strategic architectural shift to a fully hermetic C++ build system, delivering a tenfold improvement in build reliability, a 25% reduction in operational costs and significant developer velocity gains for complex CPU and GPU accelerated workloads. We are developing an AI platform for a leading home improvement retailer, serving as the foundation for generative AI tools that assist customers with how-to guidance and product inquiries already in production, this virtual assistance offers project inspiration, design concepts, product comparisons and expert recommendations for both do-it-yourself and professional users.

The solution is expected to drive significant improvements in conversion rates and average order value, particularly in maintenance and repair and aesthetic upgrades. For one of the top fintech companies, we developed a spectrum of AI initiatives to showcase advancements across domains. A multi-agent marketplace validates temporal as a scalable execution platform for complex multi-agent interactions, offering robust observability and reliability. The travel desk agent creates stateful agents with advanced memory components that generates personalized long-term itineraries overcoming context limitations through self task execution. Another AI-based solution leverages public reviews to accurately categorize miscoded merchants, identifying system misuse and potentially increasing revenue through corrected interchange fees.

Thank you. With that, let me turn the call to Anil, who will talk about our financials.

Anil Kumar Doradla: Thanks, Vasily. Good afternoon, everyone. We recorded the second quarter revenue of $101.1 million, slightly higher than the midpoint of our $100 million to $102 million guidance. On a year-over-year basis, this represents a growth of 21.7%. Excluding the impact of our recent acquisitions, the year-over-year growth was 6.3%. Both on a quarter-over-quarter and year-over-year basis, there were roughly 73 bps and 40 bps of FX-related tailwinds, respectively. Non-GAAP EBITDA came in at $12.7 million, within our guidance range of $12.5 million to $13.5 million. In the second quarter of 2025, negative impacts on our cost from FX fluctuations both on a quarterly and year-over-year basis. As you know, over the past months, the U.S. dollar has weakened against most of the currencies.

Grid Dynamics is exposed to currency basket across Europe, Latin America and India. We have a natural hedge against some of these currencies and the net impact of it was approximately $1.4 million. Looking at the performance of our verticals. Retail remained our largest vertical, contributing 29.2% of total revenues for the second quarter of 2025. Revenues in this vertical grew 10.4% year-over-year, primarily driven by demand from our existing specialty retail customers and new customer engagements. On a sequential basis, however, revenues declined by 6.2%, largely from home improvement customers. The finance vertical accounted for 25.1% of total revenues in the quarter and remained our second largest vertical. Revenues grew 1.4% sequentially and doubled year-over-year.

The substantial year-over-year growth was primarily driven by increased demand from our fintech customers, a lot of with contributions from our 2024 acquisitions that brought in global banking customers. TMT accounted for 24.9% of total revenues for the quarter, with a growth of 6.7% quarter-over-quarter and 8.4% compared to the same period last year. Largest growth driver was increased demand from our technology customers. Turning to the remaining verticals. CPG and manufacturing represented 10.5% of quarterly revenues while revenues remained flat in absolute value sequentially. It increased 7.7% year-over-year, primarily due to contributions from our recent acquisition. Other vertical contributed 7.8% of total revenues, reflecting sequential growth of 10.1% and 4.6% increase compared to the second quarter of 2024.

The year-on-year increase primarily came from customers tied to agriculture, marketplace and service provider subverticals. And finally, the healthcare and pharma made up 2.5% of our revenues for the quarter. We ended the second quarter with a total headcount of 5,013, up from 4,926 employees in the first quarter 2025 and up from 3,961 in the second quarter of 2024. At the end of the second quarter of 2025, our total U.S. headcount was 359 or 7.2% on the company’s total headcount versus 8.8% in the year ago quarter. Our non-U.S. headcount located in Europe, Americas and India was 4,654 or 92.8%. In the second quarter, revenues from our top 5 and top 10 customers were 37.5% and 57.3%, respectively, versus 38.5% and 57% in the same period a year ago, respectively.

During the second quarter, we had a total of 194 customers down from 204 in the first quarter of 2025 and 208 in the year ago quarter. The decline in the number of customers was primarily driven by our continued efforts to rationalize our portfolio of nonstrategic customers. Moving to the income statement. Our GAAP gross profit during the quarter was $34.5 million, or 34.1% compared to $37 million or 36.8% in the first quarter of 2025 and $29.6 million or 35.6% in the year ago quarter. On a non-GAAP basis, our gross profit was $35.1 million or 34.7% compared to $37.6 million or 37.4% in the first quarter of 2025, and up from $30.1 million or 36.2% in the year ago quarter. On a sequential basis, the decline in the gross margin was largely from FX headwinds, increased engineering headcount to support future growth and timing of costs related to some fixed price contracts.

Non-GAAP EBITDA during the second quarter that excluded interest income, expense provision from income taxes, depreciation and amortization, stock-based compensation, restructuring, expenses related to geographic reorganization and transaction and other related costs was $12.7 million or 12.6% of revenues, down from $14.6 million or 14.5% of revenues in the first quarter 2025 and up from $11.7 million or 14.1% in the year ago quarter. Sequential decline in EBITDA was largely due to the decline in gross profit and FX segments. The increase on a year-over-year basis was largely due to higher revenues, partially offset by an increase in operating expenses and FX fluctuations. Our GAAP net income in the second quarter was $5.3 million or $0.06 per share based on a diluted share count of 86.4 million shares compared to the first quarter net income of $2.9 million or $0.03 per share based on a diluted share count 87.8 million and a net loss of $0.8 million or $0.01 per share based on 76.6 million diluted shares in the year ago quarter.

On a non-GAAP basis, in the second quarter, our non-GAAP net income was $8.3 million or $0.10 per share based on 86.4 million diluted shares compared to the first quarter non-GAAP net income of $10 million or $0.11 per share based on 87.8 million diluted shares and $8.5 million or $0.11 per share based on 77.9 million diluted shares in the year ago quarter. On June 30, 2025, our cash and cash equivalent totaled $336.8 million, up from $325.5 million on March 31, 2025. Now coming to the guidance. Over the past couple of quarters, the majority of our enterprise clients across industry verticals have taken a certain degree of caution with traditional digital transformation spending. This is something we’ve seen across our customer base. That said, innovation led projects are client priorities from a spending point of view and Grid Dynamics has been one of the key beneficiaries of this trend.

Coming to the third quarter guidance, we expect revenues to be in the range of $103 million to $105 million. We expect our recent acquisitions contributing approximately 12% of the revenues. We expect our third quarter non-GAAP EBITDA to be in the range of $12 million to $13 million. For the third quarter of 2025, we expect basic share count to be in the range of 84 million to 85 million and our diluted share count to be in the range of 87 million to 89 million. We are maintaining our full year revenue outlook of $415 million to $435 million, all of this despite an estimated low double-digit annual percentage reduced revenue from cautionary spending on traditional business, which we projected early in the year and it was affected by uncertainty with the macro environment.

In spite of these events, we continue winning innovation led projects and grow overall revenue. As Leonard pointed out, roughly 23% of our business is tied to AI and data. This momentum around AI business is growing, and we expect this to be higher in the quarters to come. That concludes my prepared remarks. We are ready to take questions.

Cary Savas: Thank you, Anil. [Operator Instructions] First up is Mayank Tandon of Needham.

Mayank Tandon: I wanted to just maybe focus a little bit more on the pipeline and the pace of deal conversion. And if you could just talk about — as you look at the guidance for the rest of the year, and let’s take the midpoint, for example, how much of the revenue would you say is in the bag under contract? And how much do you actually still have to go out and win? Just kind of give us a sense of your confidence level in terms of getting to your guidance range.

Anil Kumar Doradla: So maybe, Leonard, if you want, I can kick off and then we can add. Yes. So Mayank, look, last quarter, there was this question, when we talked about $415 million to $435 million. And what we talked about is when you look at the low end of our guidance, that will be reached by some of the working time benefits that we see. And we still maintain that. There is obviously organic growth. So if you look at the low end of our guidance, for example, if you model something to the effect of high single digits for the full year in organic growth and we maintain this momentum of about 12% from our acquisitions that gets you to the low end of the guidance, which is a good place to be. Now we also said that there is — there are 2 things that are happening.

There are other new pipeline business, which I’m sure Leonard and the team will talk about, but there’s other opportunities that are there. As you go from the low end of the guidance to the high end of the guidance, obviously, there is a little bit more on expectations on the acquisitions. So with that, I’ll pass it on.

Mayank Tandon: Should I continue? Or…

Anil Kumar Doradla: Yes, yes, go ahead, continue on.

Mayank Tandon: Okay. I guess my follow-up question would be around just some of the key underlying drivers of the model. So how should we think about the pricing climate? How much more leverage do you have on utilization and what are your hiring plans just given some of the demand trends you talked about. So if you could just touch on those 3 metrics, that would be helpful from a modeling standpoint.

Leonard Livschitz: Let me take it because Anil is not in the room, [indiscernible] connection itself. So first of all, let me finish the first part of your question. So — the pipeline is very robust. We have a little bit of a conservative point of view because if you remember, when we were last time in the earnings call, we were talking about a very good April. And indeed, there was a very optimistic part of the growth, but we wanted to be cautious because you don’t know. Then a lot of macro impact happened next couple of months. And even though we finished the quarter, had a solid record number, still, it was not to the full expectation where it would be. The pipeline which was created in Q2 is extremely good. Again, jumping forward, the July numbers look positive, but I don’t want to jinx it again, but we look optimistic for the second half of the year quite a bit because the convergence of the projects, especially related to any kind of data and AI platforms is growing fast.

We’re talking about 3x more than regular business, but reality is almost all the customers across our universe are taken on the business associated either with a innovative projects or with a substantial migration. And we’re talking about not just POCs, quite bug projects. Some hyperscalers help us with it as well. So when we look at the — how much in the bag versus how much is a bit of a stretch, Anil mentioned to you about the lower end — getting from low end to the midrange require quite a bit of effort. And at this point, we just stay in the range. I think we’ll have much better view by the end of the quarter. But right now, we’re very optimistic on the new big deals. At the same time, coming back to your second part, how we structured the pricing around the business associated with it.

There are several aspects. So first of all — and we will about GAIN model. So when we are engaged with the technology innovative projects, as you can imagine, the price points are favorable because clients are trying to reach the goal of their internal value add to the business and to — also the cost system. So there’s a little bit of a competition for the talent. And as you know, Grid Dynamics is quite well positioned. Now, on the traditional business, there are various factors. We see a lot more pressure right now from the clients to scale the business with Grid Dynamics particularly in India. The cost structure, the price structure in India is different. So as you see, we’re continuing to grow our headcount but it’s not — it’s highly more proportional at least at this point to what we were when we were purely driven by the European engineering.

Also with this global follow-the-sun model, we’re signing the deals across various regions, particularly in Europe and in LatAm, which again has a different pricing model. So it’s very hard to kind of create a common denominator for all these factors. But we see the vector is solid. We went through negotiations, most of all, for this year. But very soon we’ll start negotiations for 2026. What’s very important and we also address it that with the weakening U.S. dollar some of the value factors for the European engineering cannot be addressed by purely time and materials. So fixed bid, our pads and now the GAIN work, it’s becoming very, very much into the solution base and that kind of gives us more positive attributes how do we build the business.

But overall, to summarize it, we embed heavily on our AI data business to grow, and it’s a fantastic positioning where we are today with our technology capability. If you want clarification, I’ll do more, but I tried to cover a very broad base base in 1 answer.

Cary Savas: Next up is Puneet Jain of JPMorgan.

Puneet Jain: I wanted to ask about how — like, Eugene, I think you talked about like how AI is changing the nature of work, specifically in this traditional SDLC cycle. But can you talk about like need for training or hiring employees differently? And it feels like your high experience within your workforce could be helpful, but I’d like to know your thoughts, like I’d like to hear like how you think you might have to hire or train employees differently as you prepare Grid Dynamics for these changes to traditional SDLC.

Eugene Steinberg: Yes. Thank you, Puneet. Great question. And this is not something which we started to do just today, we started to do it quite some time ago preparing for AI-first future. And as we already said, we’ve always been a little bit of hiring on our talent, making a strong preference to the more senior, more capable, more, I would say, broad-view engineers. And in the new AI-first software development lifecycle, the engineers who are working on the actual projects supported by the AI agents, which are actually writing code and making modification to the code base, they are acting as the judges. And their deep experience as engineers help to determine whenever the suggestions which are made by their agent is good or bad suggestion.

And this is where we see a lot of value, which is coming from more senior engineers. And at the same time, we invested quite a bit into their, I would say, AI native engineers who’ve grown with those AI agents from the very beginning of their careers, and they very natively are coming out of our internship, already armed with the understanding of those solutions. Our platform, which we are developing right now, which we call GAIN, it’s a combination of the technology. It’s not only about coding, it’s all through the whole cycle of development, starting from requirement understanding and gathering and ending with deployment and testing of the solution in production. It’s all supported by the different kinds of AI agents. And engineers, seniors engineers and AI engineers are supervising those agents and correcting them and guiding them through this process and of course, our review training program is preparing those engineers with like, we traditionally said prompt engineering, but right now, it’s more like a context engineering, a little bit new term in the industry, which everybody is using, to help those agents to be successful and to drive on further.

So that’s a shorter vision.

Puneet Jain: No, that’s great, like context engineering. We’ve been hearing that term a lot these days. No, I appreciate the response. And obviously, from investor standpoint, we take a lot of questions on the reasons for slower growth in broader IT services, whether it’s macro, whether it’s AI, let me ask that question this way, like you have, say, verticals, say, for example, financial services, which has been doing great, and then verticals like retail, healthcare, CPG, not as great. So are there any differences in AI adoption across these verticals? Or would you say like that growth difference across those verticals is purely like a function of macro or sector-specific challenges?

Leonard Livschitz: So Puneet, I’ll let other people talk about more specifically. I think it’s a very fundamental question. There’s no slow down on AI adoption across all the verticals. What happened is in certain verticals, we’re gaining momentum because the existing business, the traditional business, the cloud migration, new platforms and software development continues to expand while they adopt AI. And the AI platform, we have own homegrown platform, we use our partnership. There’s a lot of stuff that’s going on. We even started getting some press from our participation in the adoption of physical AI. So we’re really at the cutting edge of all this technology. There are some other verticals where the traditional business has been somewhat muted.

And obviously, because the retail and CPG was a substantial part of our business, and there are some traditional large legacy business, which has participation in brick and mortar, not mentioning all these promotions around the tariff strategies. They’re slowing down on a traditional software development, infrastructure and expense, all the stuff. They continue to invest into AI front. But what — if you noticed Anil brought in some flavor talking about what could have been if that business would be slowed down, it will be way above the upper range of the guidance. But it wasn’t. And what we see right now is the redeployment of resources in a more traditional conservative field, in which the, I would say, potential expansion is little bit limited where the other more aggressive expansion combined with the traditional spending in other industries.

So the bottom line conclusion, AI growth supports very dynamic growth wholeheartedly. We have more than 1 platform. We have internal platform, we have external, we participate in many activities, which actually pay decent dollars. But that’s muted current business in some of the more traditional areas started dragging a little bit down and is driven by those macros.

Cary Savas: Next up is Brian Bergin of TD Cowen.

Bryan C. Bergin: I wanted to ask on the AI-powered engagement model. Can you talk about the early client testing and reception to that model? And how are you thinking about how fast this ultimately gets adopted in your business? So what I’m specifically curious about it as it gets adopted by more, what’s the impact going to be on the financial profile of the business as we think about growth and gross margin?

Vasily Sizov: All right. Maybe let me address these questions. Thank you, Bryan, for the question. So I would say first that we definitely see increasing demand for new types of engagements and AI-powered engagements. So this definitely should fuel future growth. So that’s the first statement. As for the particularly gain implementation, so as Eugene mentioned, it’s a very comprehensive, I would say, holistic approach on how you approach the software development lifecycle by embracing those processes, technical tools, team composition and also the new commercial model. As a matter of fact, right now, we already apply certain aspects of this new platform at selected customers. Primarily the easiest thing for us is basically to bring this platform and processes into fixed-price engagements, which basically doesn’t require the customer to rethink the VMO process on how they engage us.

So from that perspective, we already see benefits by reducing — primarily reducing the time lines, which allows us to be more competitive also on the pricing side. As for the full-fledged GAIN implementation, including the commercial, we are in the phase of fine-tuning this whole model because it’s truly innovative thing. So VMOs are not ready yet, it will a learning curve for them and it will be a learning experience for us to fine-tune that. But the good thing is that we are talking to actually, 2 out of our top 15 customers right now on starting piloting this model as soon as we make the process as smooth as possible to go into production.

Leonard Livschitz: Margin portion, Bryan. I think what Vasily is — Vasily is kind of a foundational father of the model and running Americas, gives him a bit of upper hand with others. And when I was in Chicago at the conference, first time I briefly mentioned the ideas and turns out obviously we’re not alone, but what’s important is the proof is in the pudding. How much business we generate but what’s more important, how beneficial it becomes to the client and Grid Dynamics. I would not talk about directly gross margin. I was talking — I would refer to the profit margin as we grow. And you can really translate it back because it includes the partial ownership of the people, the platforms are proving the conceptual business basically become a technology consultant to the client, understanding their business flavor of those verticals.

So we’re trying to prove very interesting points which you guys were torturing us from the beginning of AI. Well, the engineers will disappear and how the new world is going to work. What is going to do to us, and this is going to be very important, if we scale this program properly. It will substantially increase revenue per person. And why would it be is because we can use our top talent, which is growing, but obviously never enough. I mean, you see some of the notable big companies growing with 8-digit, 9-digit numbers, into the people. Now for us, since they have such a good pyramid going up, we force the clients to think what’s important to them. And what’s important to them is not only individual talent but having a partnership with Grid Dynamics which makes measurable results.

And those 2 clients, which Vasily mentioned, they are far along the way. And the reason they realize why it’s important is because the pace of innovation substantially increases for the time and surpass their ability internally to conceptualize the business. So we’re innovating, deploying and analyzing business at the same time. And the key point of that today, actually, if you look at the crux of the GAIN offering and value is the data platforms because without a reliable and logical dedicated data platform on the client side, the business will be risky because the conversion may not be as valuable for the business. So I would look at the revenue per person as we scale our company rather than fuel the margin, which obviously will be addressed by increasing margin as well.

Bryan C. Bergin: Okay. Okay. Makes sense. I also follow up just in the near term, the — we’ll talk about near-term margin and just understanding demand is choppy, you do have — you’ve increased headcount again, so how are you balancing keeping quality bench for a growth recovery and potentially investing around nonbillable R&D right now versus kind of optimizing cost structure? Can you just talk about that dynamic in the near term, specific to ’25?

Anil Kumar Doradla: As you would expect — I’m laughing, Bryan, because this is exactly what we’re doing day in and day out right now. So it’s something very interesting that we are doing within the company. There are 2 very important things that I’m working on. One as you pointed out is there is a certain degree of financial discipline that we have to embark upon right, in the short term to ensure that as a public company we have to just work on. But there’s another mandate that Leonard has given me, which is, we have to double down and invest into future technologies, into future platforms and future personnel. So there are 2 parts of my whole kind of balancing act, so to speak. The focus that we’re looking at is we are creating specialized pools of labor that are targeting certain specific technologies, and I’m sure the group here, they can talk a lot more whether it is a hyperscaler, whether there’s some AI specific things.

We’ve developed internal platforms. So there’s a lot of activity going on, on the tooling side to build these accelerators and platforms on the AI side. While we’re doing that, we’re taking a very closer look at our utilization bench on our more traditional side of business. And from that point of view, obviously, we’re ensuring that we are a little bit more cost optimized. So that’s how we’re working on it. And actually, if you look at from Q2 to Q3, some of my increased costs is because of investments in some of these engineering talent, too.

Leonard Livschitz: So Bryan, let me be very blunt because Anil was trying to be a little bit elusive. This is my absolute concrete determination, we will need to be the top leaders in AI implementation offering to the clients. I know as Anil mentioned about a company need to do a certain cleanup, and we’re doing the cleanup. But we’re doing the cleanup only to open up more capabilities. As far as I’m concerned, as you know, how Grid Dynamics stock has deteriorated for whatever reason you guys decided to, so to me, it’s like, I don’t want to go back to where I was 3 months ago. I want to go 5x more than I was 3 months ago. And the reason being is the value we’re going to add to the system is going to be disproportional to everything we’ve done until the digital transformation with the cloud migration and public clouds took off.

And we’re actively participating by codeveloping the key products with our major partners. So answer your question, we will do the housekeeping, and Anil will have to make sure that we don’t do it randomly. But I have a hugely aligned strategy with technology organization. The development of the delivery capabilities, following the sun, we have here on the table, next time we’ll hear also from some Indian representative, Rajeev was there. But we have Yuri and Vasily to make sure that in the U.S. and Europe, we’re going to have a bespoke application AI platform. And as of today, we won a major program actually in LatAm as well. So you will not hear for me for a quarter or 2 that we’re going to be cautious. We’re going to be aggressive, intelligent, Anil will hold us back when things have become a little bit newer but the way how we’ve been doing it for the last 18 months, even going through the Liberation Day and all on the other macro parts and run through ups and downs, it would not deflect our determination in the modern AI, Agentic AI, physical AI, the solution practices, we’re going to be at the leadership goal.

Cary Savas: Next up is Maggie Nolan of William Blair.

Matt S. Dezort: This is Matt on for Maggie. Congrats on the quarter. I wanted to ask about the partner program and the impressive growth there. What’s your outlook for partner growth into the second half of the year? Can that continue to accelerate? And I guess where are you seeing the most new traction today amongst partners outside of Google and the hyperscalers? Or is it primarily those hypers?

Leonard Livschitz: All right. So you kind of accelerated, Matt, the question. So that was my agenda for the next quarter. We’re going to bring you ahead of the global partnerships to get you very specific details on the partnership program because the modes of operation today for Grid Dynamics is the focus on innovation, customer partnerships and a wide distribution of the GAIN model, so regions AI technology, internal platforms and a scaling partnership. So you asked a question where are we beyond the hyperscalers. First of all, hyperscalers also [indiscernible]. There are no longer hyperscalers traditionally fighting for the space on the spending dollars on the cloud platforms only. They’re very actively participating on merging the platforms they built with the AI tools they offer.

Second part is we are also partnering with the Colossus guys, the major players, which enable the foundational capabilities, like NVIDIA’s of the world because you need to have another layer without the — physical layers without capabilities, it’s hard to scale. But we are partnering with the robotics companies. We are participating in a revolutionary way of changing the industrialization and again, industrialization is very important. We’re building agent tools and agent factors within the clients and partnering with their own teams as well with third parties, which are bringing AI tools. Not to forget the fact that some of the very, very innovative creative ideas come from the myriad of the new formed AI, I would called them startups. Some of those startups are getting capitalization way above where Grid Dynamics is today.

This is where — I love the VC world. People throw money and then something works. But these guys are brilliant and our job, Eugene’s job and some of the key people on the team job is to select the ones who actually will make sense. Now I don’t want to offend any of the VCs. It’s great that you guys have so much money. But for us, we need to select the winners. So it’s the three ways, the hyperscalers into new models, the big players who are bringing their own homegrown models, and it’s our customers. We’re enjoying efforts with partners and their own stuff, we’re helping them to build the solution. And the fourth one, which is kind of exploding, it’s the industrialization part of the world with the physical AI.

Eugene Steinberg: Just want to add on the more traditional hyperscalers of the world, I’m talking from the European perspective, we definitely see traction, for example, with Google price. We are a little bit later compared to the U.S., but definitely, I see the traction right now. And it’s tied to — even with Google, it’s tied to both our more traditional search capabilities and collaboration around that. And on the other side, that Agentic AI platforms as well. So that’s definitely a big portion of what we are doing right now. And Anil also reported the numbers that we got from partnerships in terms of the revenue.

Matt S. Dezort: Great color. Can I follow-up with a question on client count. I think obviously declined quarter-over-quarter and year-over-year, I think, primarily due to the rationalization of your portfolio. How long is that going to take Anil? And when do you expect we can see stabilization in that line item?

Anil Kumar Doradla: So Matt, very good question. If you see over a pattern of several quarters, this has been kind of marginally going down. So if you look at it, there are a couple of parts to it. The first part is that most of the decline comes from our acquisition-related clients. We’ve had 6 acquisitions in the past several years. And many of them have smaller clients. And our whole focus is to look at the world through the lens of whether they are an enterprise customer or commercial customers. Our focus tends to be more on the enterprise, which is where we manage the program. We have more focus, whereas the commercial side of it could be just a cost-plus as we had through some of our acquisitions or some of the smaller ones.

So at this stage, what we do is that as those projects roll off, many at times, we don’t invest back in, there are some cases from quarter-to-quarter, some of our enterprise customers are also falling off, but that’s not a very big portion of it. And we do have a little bit of a flavor of when we even come through the partnerships. There are some clients that try out, work with us through the hyperscalers, and there’s a little bit of an infant mortality there, as they’re pausing into the next round. So the way we define, we have a little bit of a structured approach towards defining what a client is. If I do not get a dollar revenue in the quarter, and I just don’t call them a client. Although there might be an MSA, there might be something out there.

And within the 12-week month period, if they come back again, they’re not a new client, so to speak. So a client can go now 3 quarters later, they come back, they’re not a new client for me, they’re within that 12 months. So we have a little bit of a strict approach towards this. So I think you will continue seeing some of these things. At the end of the day, we can look at our top 30, top 40 customers that going to draw most of the value. Some of these smaller clients over time have come in the top 30 for us, but do expect at least in the near term, to see a trend like that.

Leonard Livschitz: Let me just conclude that part with a very important message. As you go to see Grid Dynamics, we believe that AI implementation in various forms will remain to be the key business. And there will be a huge value with a proper combination of the platform and the service providers. And I would say in a classical form, service consult with very strong technology flavor. Remember, Bryan mentioned about what is utilization, how you manage it. We look at some of the clients from the tail, and we just don’t invest into that relationship, too, because we feel they don’t have capabilities to become a viable player in the near future. It doesn’t mean we just tell them goodbye. But if they don’t fit in the model of the new AI era, and we try very hard to convince them there, they just don’t have that priority from our folks.

So you will see that some of the clients will grow exponentially because it’s a meeting of the mind. Some of the clients either stagnate or leave — fade as we go forward.

Cary Savas: Next up is Surinder Thind of Jefferies.

Surinder Singh Thind: Big picture question here. As you look forward to all of the changes that you guys are making, the transformations, what is the scenario that you’re actually solving for? Meaning what is the level of AI capability? Are you planning for a scenario where 50% of software development, 75% is done by AI? Like what are the kind of the framework of what you’re moving towards?

Leonard Livschitz: Well, first of all, software development is not the only area of AI.

Surinder Singh Thind: Just as an example, that’s a conceptual idea of the environment that — because one of the things that’s happening is changes have to be very rapidly and so if you’re solving for something here in X, but by the time you get to X or at Y, you’re going to be faced with the constant evolution, so how are you thinking about the evolution of the firm over the next 3 to 5 years?

Leonard Livschitz: Yes. So the evolution will take more than 3 quarters. This is going to — it’s a combination of adoption of the technologies, capabilities of the distributed systems and ability to generate the value of each individual in each individual case. 50% is just a number, 30, 50, 80 and used software development, but please, in my opinion, again, it’s a small percentage of the change. There’s a change of creating the value for the business and just trying to improve the quote. I think that human touch will continue to be a big part of all the key decisions, but it’s going to be in a different forms. Writing the code is always allowed. And I think Eugene can comment more on that. There is ability to understand the deployment of the systems, reverse compatibility of the systems, security to manage and control the future expansion of the systems.

And we will have to nail each of them separately. On a low level, it is going to take way more — it will take way lower than 50%. It’s going to go substantially a large percentage of the co-development. As we continue to monitor and expand on the capabilities, the progression will take a much longer time than some of those people want to make sure they have instant conversion. So the people, the humans who are properly trained in the systems will have to define with ability to what is the future? Otherwise, you’re going to move with — we will have to face the moving target, which is not a good solution because you continue to invest in something which is going to be aged in 6 months. So to summarize that, on the basic coding level, it’s going to be way more than 50%.

On a system management integration, data compliance, it’s going to be lower, but the effect has to be proven. On analytical part of the systems, it is going to be always a combination of people and projects. And just another point, when people are talking about tasks some of them will be 100% task by the agents. And when we’re talking about the total system implementation, the percentage would be lower. So I think it would be a good segue that Eugene, you can make comments of your experience.

Eugene Steinberg: Yes, of course. And I understand that currently internet is full of the examples of vibe coding, generating full-fledged Facebook clone in 15 minutes. Anybody who really kind of performed production bound projects understand that building a prototype and putting something in production underload for mission critical systems is very different things. And what we observe in our business is that I’m in charge of presales, for example, as a company. And our presales is completely transformed by AI-first development. We are able to turn really very good prototypes and pilots very, very quickly to the customers. But then we are going into the real thing in implementation of the scalable systems, implementation of the system underload in secure environment with deployment and scaling requirements, which are needed for production, it’s completely different ballgame.

And even the most sophisticated AI agents are very quickly losing their contracts and starting to float to the side. And they still need a lot of the human supervision and human design and thinking and the activity behind them to steer them. Very simple like they are savants. They are smart, but aimless and this is why we are still kind of in — we’re still…

Leonard Livschitz: Eugene, you opened the pandora box. This is our strive for excellence.

Vasily Sizov: We had multiple conversations on the topic. And 1 color I wanted to add is that AI has different flavors that it’s in different stages of adoption because things like, for example, co-pilot when still an engineer writes a code, but AI suggests things, it will be widely adopted. Maybe if it’s not 100%, maybe 95% or whatever, it’s more kind of dependent actually on the security protocol within the customer, whether they want to be exposed to external LLMs. But this thing is, like it’s a done. So what Eugene was and Leonard was talking about is more of agent-based AI coding, and that’s kind of the more complex topic, which will hash through the adoption curve for sure.

Cary Savas: Ladies and gentlemen, this concludes the Q&A session of our call today. I will now turn it over to Leonard Livschitz for closing comments.

Leonard Livschitz: As we conclude our second quarter earnings call, I want to leave you with 3 key takeaways. Number one, Grid Dynamics AI-first strategy is driving our growth. The AI and data initiatives now account for a quarter organic revenue in the first half of 2025 growing nearly 3x faster than our overall organic business. Number two, AI is fundamental to driving our clients’ business forward. Enterprises are seeking AI-native partners with the expertise to lead AI adoption at scale. This is Grid Dynamics strength. Our expanded pipeline aligns with enterprise investments. And finally, Grid Dynamics is built for sustained differentiation. We have a proven track record of emerging stronger through industry transitions.

Based on reaccelerating client demand, we are confident in our outlook and our ability to empower Fortune 1000 enterprises in their AI journey. We’re excited about the path ahead and the value we’re creating. I look forward to giving you an update on the next earnings call.