EPAM Systems - Earnings Call - Q3 2025

Executive Summary

• Q3 delivered broad-based growth and a clean beat vs. consensus: revenue $1.394B (+19.4% y/y; +7.1% organic CC) vs. $1.375B consensus; non-GAAP EPS $3.08 vs. $3.03 consensus; GAAP EPS $1.91 (*Estimates: S&P Global). Sequential profitability improved again as non-GAAP operating margin reached 16.0% (Q2: 15.0%, Q1: 13.5%).
• Full-year guidance raised: revenue to $5.430–$5.445B (midpoint +100 bps to 15.0% growth), GAAP EPS to $6.75–$6.83, and non-GAAP EPS to $11.36–$11.44; Q4 guidance implies 11.1% y/y growth at midpoint with non-GAAP OI% 15.5–16.5%.
• Growth drivers: five of six verticals grew y/y; Financial Services and Emerging led; all regions grew (EMEA +24.9% y/y), with continued AI-native demand and foundational data/cloud work; pipeline shifting from proofs-of-concept to medium/large-scale programs.
• Capital returns: $82.1M repurchased in Q3; new $1.0B authorization announced Oct 21—a supportive near-term sentiment catalyst alongside raised guidance and AI product momentum (AI/Run.Transform, Agentic QA).

What Went Well and What Went Wrong

What Went Well

• Sustained organic momentum and broad-based growth: organic CC revenue +7.1% y/y (fourth consecutive positive quarter), with five of six verticals up y/y and all regions positive; EPAM exceeded the high end of prior Q3 revenue guidance.
• AI-native traction and pipeline maturation: 60–70% of AI projects expanded from PoC to larger programs; data/cloud practices grew faster than the overall business; internal AI literacy >90% and >95% of engineers completed foundational AI education. “Our third quarter results came in better than expected… clients are prioritizing their AI buildouts, turning to EPAM…” — CEO Balazs Fejes.
• Capital allocation and cash generation: record quarterly operating cash flow ($295M) and FCF ($286M); $82M buyback in Q3; new $1B repurchase authorization underscores confidence.

What Went Wrong

• Margin pressure vs. prior year: GAAP gross margin 29.5% (31.0% non-GAAP) down y/y due to acquisition mix and higher variable comp; prior year benefited from a Polish R&D credit catch-up; GAAP OI% fell to 10.4% (vs. 15.2% in Q3’24).
• EBITDA (S&P methodology) missed consensus despite EPS beat: Q3 EBITDA actual ~$180.3M vs. ~$233.8M consensus*; reflects differing methodology and acquisition mix headwinds (company doesn’t report EBITDA) (*Values retrieved from S&P Global).
• Seasonality and budget dynamics: management does not expect a year-end budget flush like last year; Q4 faces normal headwinds (fewer billable days, vacations, furloughs) tempering sequential growth into Q4.

Transcript

Operator (participant)

Thank you for standing by. My name is Rebecca, and I'll be your conference operator today. At this time, I would like to welcome everyone to the EPAM Reports Results for Third Quarter 2025 conference call. All lines have been placed on mute to prevent any background noise. After the speaker's remarks, there will be a question-and-answer session. If you would like to ask a question during this time, simply press star followed by the number one on your telephone keypad. If you would like to withdraw your question, press star one again. Thank you. I would now like to turn the call over to Mike Rowshandel, Head of Investor Relations. Please go ahead.

Mike Rowshandel (Head of Investor Relations)

Good morning, everyone, and thank you for joining us today on our Third Quarter 2025 earnings announcement. As the operator just mentioned, I'm Mike Rowshandel, Head of Investor Relations. We hope you've had an opportunity to review our earnings release we issued earlier today. If you have not, copies are available on epam.com in the Investors section. With me on today's call are Balazs Fejes, CEO and President, and Jason Peterson, Chief Financial Officer. I would like to remind those listening that some of the comments made on today's call may contain forward-looking statements.

These statements are subject to risk and uncertainties as described in the company's earnings release and SEC filings. Additionally, all references to reported results that are non-GAAP measures have been reconciled to the comparable GAAP measures and are available in our quarterly earnings materials located in the Investors section of our website. With that said, I will now turn the call over to FB.

Balazs Fejes (CEO and President)

Thank you, Mike, and good morning, everyone. It's a pleasure to be here with you on my very first earnings call as CEO. Please just call me FB, as Hungarian names are notoriously difficult to pronounce, and why FB? Because in my native language, family name comes first. This quarterly call arrived faster than even my standard double espresso shot in the morning. That is really the theme for the day. Things have been moving quickly since we spoke last, and today we have positive news to share. Our third quarter results came in better than expected, marking another quarter of broad outperformance and strong delivery execution. We continue to benefit from AI and AI-native lead demand, or TCs, that data and modern cloud architecture are critical for AI adoption and auto realization is broadly being confirmed by what we are seeing.

Our clients are prioritizing their AI buildouts, turning to EPAM to help them accelerate their investments and innovation in AI. The unique combination of our deeply rooted engineering DNA and our globally recognized best-in-class AI-native expertise continues to differentiate our offerings and help us further expand wallet share within our existing client portfolio and targeted new logo segments. At the same time, we are more focused than ever on upgrading our engineering skills advantage and investing for the future with new advanced AI playbooks and accelerators. Serving as client zero for AI adoption, we believe innovation starts from inside, which is why in parallel, we continue relentlessly to push our AI literacy and AI adoption rates. Looking at our progress year to date, more than 90% of EPAMers have completed their mandatory AI literacy education, and approximately 95% of our engineers have completed foundational AI education.

Additionally, our internal business processes are increasingly benefiting from AI-driven efficiencies. As you can see with the recent launch of EPAM AI Run Transform, which includes next-generation AI-managed services and EPAM Agentic QA, we are programmatically expanding new offerings and highly specialized capabilities, often in conjunction with our clients and strategic partners to help clients transform themselves into AI-native organizations. Our efforts are being recognized by our partners as well as industry analysts such as IDC Marketscape, who have positioned EPAM as a leader across experience engineering, custom-built, and AI consultancy capabilities. Further, Glassdoor ranked EPAM number seven on their 2025 best-led companies list, along with Forbes, who recognized EPAM as one of the world's best employers, our first time being recognized across both.

We will dive into the details a bit later in the call, but first, I would like to provide a quick update from my early days as CEO and my recently completed Vert tool. Over the past quarter, I met in person with many senior client executives, ecosystem partners, and of course, many, many EPAMers from all around the world. I experienced firsthand the high level of optimism and appetite for EPAM-proven quality of execution across our global, deeply specialized talent base. I am pleased with our continued and growing ability to assure higher levels of performance across a much more globally diversified footprint than ever before. Our work is not done. AI presents a permanent sea change in our industry and across our clients' businesses, driving the need for investment in modernization, data and cloud foundations, and critical AI-native skills.

EPAM is positioned to lead both the foundational and the transformational programs demanded by AI as clients need support from trusted partners who can reliably deliver through the need to simultaneously balance cost and productivity with an increasing need to reinvest, innovate, and keep pace with change. In line with Amara's law, we believe that as productivity grows and as cost to develop software declines, complexity will significantly accelerate, pushing the bleeding edge and resetting the boundaries of what is possible and triggering a flywheel effect of demand for EPAM's unique breed of capabilities and global scale. We believe that as complexity rises, so does enterprise risk, raising the importance of highly advanced engineering with proven enterprise-grade quality execution. While we have seen the flashy vibe-coding video shorts and headlines, in our view, the absolute need for true engineering expertise, risk management, fault tolerance, and reliability are overlooked and underestimated.

Now, let's turn to some Q3 highlights. In Q3, we delivered another quarter of double-digit revenue growth, including very strong year-over-year organic constant currency revenue growth of 7.1%, which exceeded our expectations set a quarter ago. This marks our fourth consecutive quarter of positive year-over-year organic constant currency growth, reflecting a steady build of improvement and strong execution of our core business as we continue to ramp our AI-native services. Our broad-based growth momentum carried forward in Q3 with five out of six verticals growing year over year. Notable standouts included emerging verticals, financial services, and software and high-tech. We also saw solid improvement in life sciences and healthcare along with consumer goods, retail, and travel, while business information and media remained steady. Geographically, all three regions delivered strong year-over-year growth.

We continue to add net organic headcount across key locations such as India, Central Europe, Eastern Europe, and South America, with increases partially offset by ongoing optimization in select pockets that we have discussed previously. Now, turning to demand environment. AI continues to trigger incremental demand and is driving positives in our pipeline globally. The majority of our top 100 clients remain highly engaged in AI-native initiatives. Clients engage EPAM to build out their data platforms and modernize their cloud, often redirecting work from other partners who successfully sold advanced capability but failed to deliver it. Overall, we continue to see improvement in the demand environment as we're seeing a continued upshift in investment towards everything that supports AI adoption and its deployment to production.

This is where reputation for trusted quality and execution remains a significant competitive advantage and a key enabler for us to continue to maintain our pricing integrity. We gain traction with our ongoing client-centric initiatives while at the same time continuously strengthening and optimizing our global delivery footprint, which has enabled us to better meet market demand. When you look across the AI project lifecycle, from proof of concepts to medium-sized use cases and then large-scale projects in production, we are seeing a continued shift in the volume of projects towards medium and large-sized projects, many making use of our own IPs such as Dial, AI Run, and other components, both open-source and proprietary.

Of the hundreds of individual AI-native projects we had active in Q3, between 60%-70% have expanded into larger programs from the origination as proof of concepts, illustrating our ability to scale and deliver AI-native solutions in production. We are also seeing positive signs at the top of the funnel, enabling us to replenish our pipeline as some projects come to a natural close. Our hard work and continuous effort to further position EPAM as a leader in AI-native services is serving us well as our pure AI-native revenues continue to grow nicely with a third consecutive quarter of double-digit sequential growth. As we have discussed before, the foundational services necessary to make AI work are a core fundamental to our business.

In both our data and cloud practices, we saw outsized growth in Q3 compared to the rest of the business, which is incremental and highly connected to the momentum we are seeing with our pure AI-native revenues. Now, turning to our AI Run Transform and Agentic QA announcements. First, a couple of core beliefs to frame our evolving AI approach. Number one, AI is not just a technology. It is a transformative force that is already redefining how enterprises innovate, operate, and create value in the future. In this context, advanced engineering, bleeding-edge AI technology, and toolsets along with deep knowledge across the software development industry and new product lifecycle are the core competencies that will drive the most tangible AI outcomes.

This will become even more evident as we see further rise of AI in the global and regional lineups of players in both Western markets and broadly across APAC, LATAM, and the Middle East. Number two, we believe in building AI responsibly with trust, transparency, governance, and measuring outcomes at the core. AI must deliver real outcomes with proper traceability and risk management. Number three, we believe we are creating a new AI-native engineering profile, or North Star, when it comes to talent development strategy, embedding AI intelligence and orchestration of agents directly into the development process. Over time, this role becomes the architect of AI-native products and experiences augmented by agents to expand the scope of what teams can achieve in the future. AI investments are in intense race.

Our approach is to invest in accelerators, tooling, and people who help us deliver reliable outcomes on the promise of AI. We do not sell foundation AI in a silo. Instead, we use our expertise and advanced IP to sell and deliver with AI, packed with proven quality and execution that guarantees outcome and value realization. This is true across our entire IP portfolio and is shaping into a structured blueprint we are calling AI Run. AI Run Transform represents our unified AI strategy that harmonizes our go-to-market notions with better activation across strategy and consulting frameworks and methodologies, talent, and advanced toolsets. We have two key offerings: AI innovation business transformation and AI-native engineering transformation. The first offering is focused on optimized expand-run across AI industry solutions, AI horizontal solutions, and AI product design and experience.

The second offering is focused on mastering the SDLC, advancing the agentic delivery lifecycle, known as ADLC, and preparing for product development lifecycle, known as PDLC. This encompasses AI-native delivery, AI-driven modernization, and PDLC agentic solutions. We will be talking more about these offerings in the quarters to come. Our AI Run blueprints encompass our AI frameworks and include our AI 360, AI Factory, AI Run SDLC, and AI adoption and education frameworks, which are agnostic and provide critical flexibility, which help EPAM deliver more enterprise-grade AI solutions at scale for our clients. Our AI Run talent houses our verticalized industry teams, ontologies, and accelerators, which include strategic advisory, data models, process modeling, and solution-built with partners. Most importantly, this is the scaffolding we are using to define the forward skills of the future and the paths for upskilling people and organizations.

Our AI Run tools combine our best-in-breed IP assets such as EPAM Dial and AI Run Platform with our strong AI data and cloud ecosystem partner solutions, and many are available today on our partner marketplaces. You may have seen we also recently announced one of these tools, Agentic QA, which bridges the gap between automated and manual testing, enabling clients to move faster by reducing lead times and costs. What's impressive is that our Agentic QA is shown to be 10x more efficient than manual testing, driving a 50% reduction in manual efforts and a 30% reduction in testing costs, covering 90% of the manual checks performed on standard releases while ensuring a high degree of quality and precision. Now, turning to some client examples to illustrate some of our progress.

This past quarter, we announced several collaborations with both new and existing clients, which illustrate not only the evolution of our client proposition, but also how EPAM is able to systematically address innovative needs while offering real value. A few notable examples. Agentic customer service breakthroughs are real with Eins und Eins, a major provider for telecommunications, cloud, and internet services in Germany. By deploying EPAM's AI Run Transform blueprint and leveraging Microsoft Azure, this client launched AI voice agents that handle over 100,000 calls weekly, with the first agents going live under three months into production. Two. Our collaboration with Hugo Boss and our Empathy Lab Studio is reimagining what it means to be a sports fan in the age of spatial computing. This innovation is shaping the next generation of motorsport fandom, lifting the bar, how luxury fashion, sport, and technology intersect.

We are blending our deep expertise in groundbreaking user experiences with gaming, fan engagement, advanced data, and analytics, and working with our clients to help package the 2025 Stevie Award-winning solution for the AI-native age. Three. Finally, we are putting EPAM and Neoris together in a way that goes beyond simple synergies. For a U.K.-headquartered global biopharmaceutical company, EPAM, with the addition of Neoris, recently became a global strategic supplier across a broad range of transformation pillars. A key joint win for us is in helping the client to build out a modern data and AI center of excellence, which spans across multiple programs and new locations, including Iberoamerica. To close, our operating momentum is strong. We are pleased with our performance throughout the year and continue to work on improving profitability.

We are confident in the upward trajectory we have been working hard to build and sustain over the past several quarters and feel good about our Q4 positioning, which has improved over the past 90 days. We are focused on what's right in front of us and finishing 2025 strong, which we believe should set up a solid foundation to build upon in 2026 as we continue to work on expanding our organic custom currency growth rate. We are prioritizing client-centric, disciplined execution while bringing a new level of intentionality on building verticalized and differentiated horizontal go-to-market offerings. Looking ahead, we see our investments in upskilling, differentiated AI playbooks, IP partnerships, and new lines of services such as agentic business process outsourcing helping us to further capture new demand. Jason, over to you.

Jason Peterson (CFO)

Thank you, FB. And good morning, everyone. In the third quarter, EPAM generated revenue of $1.394 billion, a year-over-year increase of 19.4% on a reported basis, exceeding the high end of our Q3 revenue guidance. On an organic constant currency basis, revenues grew 7.1% compared to the third quarter of 2024. We delivered another consecutive quarter of very solid year-over-year organic constant currency growth, reflecting ongoing steady execution. Our growth in the quarter was driven by a continued shift to quality and accelerating momentum across our AI-native data, cloud, and AI foundational initiatives. We're making early headway with the launch of our AI Run Transform strategy, which complements our underlying growth momentum, positioning us well to continue to capture demand. Our outperformance in the quarter was broad-based. We also recently announced a new $1 billion share repurchase program.

The underlying strength of our business and continued momentum, coupled with our efficient free cash flow generation and a strong balance sheet, enable us to take advantage of the current market dynamic while returning cash to shareholders. Moving to our Q3 vertical performance, five of our six industry verticals posted year-over-year growth, with four of the six growing double digits. Neoris and First Derivative continue to contribute substantially to our financial services and emerging verticals. Financial services once again delivered very strong growth, up 32.7% year-over-year on a reported basis, with 6% organic growth in constant currency. Growth came from banking, asset management, and insurance clients. Software and high-tech grew 19.1% year-over-year, driven by strong execution and broad improvement across large clients. Life sciences and healthcare increased 11.8% on a year-over-year basis. Revenue growth in the vertical continues to be driven primarily by clients in life sciences and med tech.

Consumer goods, retail, and travel delivered 9.9% year-over-year growth, marking a notable rebound relative to prior quarters. The vertical also delivered solid sequential growth, which was driven by growth in consumer products and retail. Business information and media was steady and delivered flat year-over-year revenue performance. Our emerging verticals delivered another quarter of very strong year-over-year growth of 38.9%, with Neoris continuing to contribute to the vertical's performance. On an organic constant currency basis, growth was 15.1%, primarily driven by ongoing strength in energy and materials. From a geographic perspective, Americas, our largest region, representing 58% of our Q3 revenues, grew 16% year-over-year on a reported basis and 3.9% in organic constant currency. AMEA, comprising 40% of our Q3 revenues, increased 24.9% year-over-year and 11.8% in organic constant currency. Finally, APAC, making up 2% of our revenues, increased 17.7% year-over-year and 14.2% in organic constant currency.

Lastly, in Q3, revenues from our top 20 clients grew 10.2% year-over-year, while revenues from clients outside our top 20 increased 24.4%. Moving down the income statement, our GAAP gross margin for the quarter was 29.5% compared to 34.6% in Q3 of last year. Non-GAAP gross margin for the quarter was 31% compared to 34.3% for the same period a year ago. As a reminder, the prior year period benefited from a cumulative catch-up related to the Poland R&D credit. The third quarter of 2025 includes a single quarter's benefit of $13.2 million. Additionally, for Q3 2025, we recognized higher variable compensation driven by expected stronger second-half performance. Combined with ongoing lower profitability associated with recent acquisitions, both contributed to the lower gross margin level. GAAP SG&A was 16.8% of revenue compared to 17.7% in Q3 of last year.

Non-GAAP SG&A in Q3 2025 came in at 14.1% of revenue compared to 14% in the same period last year. GAAP income from operations was $145 million, or 10.4% of revenue in the quarter compared to $177 million, or 15.2% of revenue in Q3 of last year. Non-GAAP income from operations was $222.8 million, or 16% of revenue in the quarter compared to $222.9 million, or 19.1% of revenue in Q3 of the previous year. Non-GAAP income from operations in Q3 2024 was similarly impacted by the Polish R&D credit. Our GAAP effective tax rate for the quarter came in at 25.6%, and our non-GAAP effective tax rate was 24.1%. Diluted earnings per share on a GAAP basis was $1.91. Our non-GAAP diluted EPS was $3.08 compared to $3.12 in Q3 of last year, reflecting a 4-cent decrease year-over-year.

In Q3, there were approximately 55.8 million diluted weighted average shares outstanding. Turning to our cash flow and balance sheet, cash flow from operations for Q3 was $295 million compared to $242 million in the same quarter of 2024. Although seasonality always has a positive impact on Q3 cash flow, cash flow from operations in the quarter exceeded the impact of typical seasonality, resulting in the highest level of quarterly cash flow from operations in EPAM's history. Free cash flow was $286 million compared to free cash flow of $237 million in the same quarter last year and also represented an all-time high. Cash and cash equivalents were just over $1.2 billion as of the end of the quarter. At the end of Q3, DSO was 75 days compared to 78 days for Q2 2025 and 74 days for the same quarter last year.

Share repurchases in the third quarter were approximately 493,000 shares for $82 million at an average price of $167 per share. Moving on to operational metrics, we ended Q3 with more than 56,100 consultants, designers, engineers, and architects, reflecting total growth of 17.5% and organic growth of 6.4% compared to Q3 2024. In the quarter, we added approximately 300 net delivery professionals. Our total headcount at quarter end was 62,350 employees. Utilization was 76.5% compared to 76.4% in Q3 of last year and 78.1% in Q2 2025. Now let's turn to guidance. Before moving to the specifics of our 2025 and Q4 outlook, I would like to provide some thoughts to help frame our guidance. Based on the strength of our Q3 and solid Q4 visibility, we are expecting a strong Q4 exit, ending the year with higher organic constant currency growth rates than we forecasted just 90 days ago.

At the same time, we are not expecting to see a significant release of excess client budgets, and typical seasonality will also have an impact. Compared to Q3, Q4 is negatively impacted by a higher number of holidays, vacations, and potential furloughs. As a reminder, we acquired Neoris and First Derivative in Q4 2024, in November and December, respectively. As per our usual reporting practice, revenues from these acquisitions will move from inorganic to organic in Q4 2025. As contemplated in our previous guidance. Based on our better-than-expected performance in the second half, coupled with improving visibility into Q4, we are raising the bottom end of the range for 2025 full-year organic constant currency revenue growth and now expect the midpoint of the range to be 4.6%. An increase from the guidance we gave 90 days ago, which was 4% at the midpoint of the range.

While driving top-line revenue growth, we also remain focused on improving profitability. While there is more work to be done, we have been pleased with the results of our ongoing focus on improving account profitability, which is evident in our improved profitability outlook for Q4 and full-year 2025. Lastly, we continue to work on improving utilization, and we continue to reduce isolated pockets of bench while adding net headcount to support growth. Our guidance continues to assume that we will be able to deliver out of our Ukraine delivery centers at productivity levels similar to those achieved in 2024. Moving to our full-year outlook, we now expect revenue to be in the range of $5.430 billion-$5.445 billion, reflecting a year-over-year growth of 15% at the midpoint, with inorganic continuing to contribute approximately 9.1% for 2025.

Based on current spot rates, foreign exchange is now expected to have a positive impact on revenue growth of 1.3%. We expect year-over-year revenue growth on an organic constant currency basis to now be 4.6% at the midpoint. We expect GAAP income from operations to now be in the range of 9.4%-9.7%, and non-GAAP income from operations to now be in the range of 15%-15.3%. We expect our GAAP effective tax rate to now be 25%. Our non-GAAP effective tax rate, which excludes the impact of benefits and shortfalls related to stock-based compensation, will continue to be 24%. For earnings per share, we expect the GAAP diluted EPS will now be in the range of $6.75-$6.83 for the full year, and non-GAAP diluted EPS will now be in the range of $11.36-$11.44 for the full year.

We now expect weighted average share count of 56.2 million fully diluted shares outstanding. Moving to our Q4 2025 outlook, we expect revenue to be in the range of $1.380 billion-$1.395 billion, producing a year-over-year growth of 11.1% at the midpoint of the range. Our guidance reflects an inorganic contribution of 4.3%. With a 2.4% positive FX impact during the quarter, producing a 4.4% organic constant currency growth rate at the midpoint of the range. For the fourth quarter, we expect GAAP income from operations to be in the range of 10%-11%. Non-GAAP income from operations to be in the range of 15.5%-16.5%. We expect our GAAP effective tax rate to be approximately 24% and our non-GAAP effective tax rate to be approximately 23%.

For earnings per share, we expect GAAP diluted EPS to be in the range of $2.00-$2.08 for the quarter and non-GAAP diluted EPS to be in the range of $3.10-$3.18 for the quarter. We expect a weighted average share count of 55.1 million diluted shares outstanding. Finally, a few key assumptions that support our GAAP to non-GAAP measurements for Q4. Stock-based compensation expense is expected to be $44 million. Amortization of intangibles is expected to be approximately $18 million. The impact of foreign exchange is expected to be $1 million. Tax-effective non-GAAP adjustments are expected to be around $16 million. We expect a tax shortfall related to stock-based compensation of around $1 million. Savings driven by our cost optimization program is expected to be around $10 million.

One more assumption outside of our GAAP to non-GAAP items, we now expect interest and other income to be $3 million for the remaining quarter. We remain focused on driving revenue growth and enhancing profitability. We are confident in our strong positioning as we enter Q4. We will continue to run EPAM efficiently, maintaining our focus on both growth and profitability throughout the remainder of the year. Thanks again to all our employees for their dedication and focus on serving our clients and driving results for EPAM. Operator, let's open the call for questions.

Operator (participant)

At this time, I would like to remind everyone, in order to ask a question, press star one on your telephone keypad. We'll pause for just a moment to compile the Q&A roster. Please limit to one question and one short follow-up question due to time restrictions. Your first question comes from the line of Maggie Nolan with William Blair.

Maggie Nolan (Partner and Equity Research Analyst in IT Services)

Hi. Thank you for taking my question. I wanted to start with the push that you mentioned into Agentic BPO. Do you intend to enter that space with proprietary products, or can you talk about maybe build versus buy decisions from clients for processes and then just the ability to automate this, how that may be or may not be any different from the robotic process automation wave that we saw several years ago that ended up being sort of difficult to accomplish given the variability of processes?

Jason Peterson (CFO)

Good morning, Maggie. Good afternoon. Thank you for the question. Actually, it is a really interesting subject. Early days for us. As you know, we made two acquisitions in this space. First was First Derivative, where it had a line of business which was in business services.

That's where we really went after that acquisition with the thesis that we could automate with Agentic AI the key BPO FinCrime elements of their business. The second acquisition was Linksys, which was this year, was a small BPO to really understand the space itself. Going back to what we are seeing, what we are seeing, our clients are keen to try out, but it's early days. We are using EPAM build platforms itself in order to really deliver the automation, but it's very different than RPA in the past. What we're trying to do is we're trying to experiment on simple and more complex agentic flows, which requires a high level of degree of engineering, not going beyond simple RPA or simple RPA capabilities. We don't know yet where the market will go. We don't know where it's leading.

Right now, we are seeing a bigger momentum from the clients which we are talking. But it's a very small sample which we're talking about. It's early days for us.

Operator (participant)

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

Bryan Bergin (Managing Director)

Hey, guys. Good morning. Thank you. I wanted to ask about, as we think on your 4Q exit rate considerations. And think beyond that as we move forward to 2026, how we should be thinking about growth potential. And specifically, if you can kind of comment on the impact of build days and furloughs and things like that as you go through 3Q into 4Q and then into 1Q, as well as just any other important factors such as how growth in Neoris and FD may affect your organic growth rate as you fold those in going forward.

Jason Peterson (CFO)

Yeah. Hey, Bryan. This is Jason. As I think most people know, there's a negative impact from a seasonality standpoint if you look at sequential Q3 to Q4. That impact is kind of three things, which is one, it's fewer build days. You've got more vacation. You also have a higher degree of furloughs. All of those things produce some tens of millions of kind of headwind on sequential growth Q3 to Q4. When I look at the performance of our business throughout 2025, Q1 to Q2, Q2 to Q3, and Q3 to Q4, if you adjust for foreign exchange and you adjust for sequential factors, our sequential growth rate has actually been surprisingly consistent. The other thing I would add is you've got probably a little bit of headwind on foreign exchange sequentially Q3 to Q4.

Just to sort of maybe answer a question that you hadn't asked, it's that our guide at the midpoint of the range contemplates, I think I said 4.4% organic constant currency growth. If we operate at the high end of the range in Q4, we'd be at about 5% organic constant currency growth.

Operator (participant)

Your next question comes from the line of Jason Kupferberg with Wells Fargo.

Jason Kupferberg (Senior Equity Research Analyst)

Good morning, guys. Thanks for getting the question. The organic constant currency in the quarter, obviously the 7%, I think kind of best in class now among the peer group. Just to kind of build on the last question, I guess, can you kind of break down the sources of what looks to be some deceleration in the Q4 on a year-over-year basis? You just walked us through the sequentials, Jason. We just want to make sure we have the puts and takes right there and then how we should just at least directionally be thinking about where the organic growth can go in 2026 versus, call it, a 4.5% exit rate for this year. Thanks.

Jason Peterson (CFO)

Okay. Good. Okay. So from a year-over-year standpoint, I think maybe the biggest difference is we see clients continue to make investments and move forward on programs. What we're not seeing is a release of excess budget at the end of the year the way we saw in Q4 of last year. I think that is probably the biggest difference. Clients continue to invest, but there just isn't a big kind of opening up the wallet at the end of the year. From a demand standpoint, it still feels broad-based.

It still feels, again, like we're continuing to see growth in financial services, high-tech, and also kind of the emerging energy portion of the portfolio. FBU, thoughts on 2026, or? I think we believe that the organic growth rate will be higher than this year. We see the momentum. It's driven by very much, as we mentioned, some of the AI and AI fundamental build-out, the foundational build-outs. We see the pipeline for 2026 building very, very nicely at this point of time. It's early days, but we see positive signals.

Operator (participant)

Your next question comes from the line of Jonathan Lee with Guggenheim Partners.

Jonathan Lee (Managing Director in Equity Research)

Great. Thanks for taking my questions. FB, welcome to the first of hopefully many earnings calls as CEO. It's interesting to hear that clients are redirecting work from partners who fail to deliver, effectively highlighting that you're winning share from peers. Can you help size that contribution and unpack your competitive advantage here versus your peers? How do you expect to maintain that gap going forward?

Balazs Fejes (CEO and President)

Jonathan, thank you very much for the kind words. I do not think we can size it yet, right? I do not think we can size it. How much work is actually redirected to us. We are seeing that in major programs, competitors who fail to deliver, now clients are redirecting the work to us. The reason why is because actually delivering these solutions in enterprises is much more difficult than what it seems on a YouTube short or a TikTok video. You need really deep engineering skill set, capabilities across the foundational elements on data, on data platforms, on cloud, or enterprise platforms themselves, or actually modernization in order to deliver on that. You also need to consider the cost.

You need to cost engineer because keypads, tokens are quite expensive. You need to consider risk elements and actual reliability and performance. All of these require deep engineering skill set. How are we going to keep our advantage? Because we are investing in our people. We are investing in our engineering talent. We're investing into tooling, methodologies. We're investing into the playbooks, and we're actually trying it out and experimenting on ourselves. That's why we believe being the customer zero, EPAM being the client zero, is so important for our future.

Operator (participant)

At this time, I would like to remind everyone, please limit to one question and to one follow-up. Your next question comes from the line of Jim Schneider with Goldman Sachs.

Good morning. Thanks for taking my question. FB welcome.In some of your public commentary and interviews recently, I think you've kind of struck a chord about focus on costs for the company. Can you maybe give us a sense about how that focus on cost is being manifest across the company and how that might materialize in terms of SG&A or other kind of cost savings or margins over time? Thank you.

Balazs Fejes (CEO and President)

Thank you very much. I think in the last recent months, I was talking about the focus on pyramids. We focused on actually balancing the pyramid. When we diversified our delivery, we actually went into certain geographies, went into certain locations. We were not able to really create the ideal pyramid structure. Now we're working on that, and we're trying to rebalance the pyramid itself. Rebalancing the pyramid is actually allowing us to really focus on and bring down some of the cost.

Second is, as a CEO, I'm putting more emphasis on profitability on the deals, emphasis on the capabilities to actually deliver profitable projects, profitable growth, which really manifesting for us, selecting the right clients, being more picky, more selective on the deals that we take, which we can only do because our demand is changing and the demand is up for us. That's what you are seeing the effect of that.

Jason Peterson (CFO)

Yeah. I think I'll add a piece on this as well. As FBU indicated, with that focus, we are seeing an improvement in account margin in the second half of the fiscal year. I think probably what is most notable is throughout the year, we've been talking about a 15% midpoint of our profitability range. At this time, we feel pretty strongly that we'll operate in the upper half of the 14.5%-15.5% range. As we talked about in my prepared remarks, we expect to operate in the 15-15.3. That is a result of a number of things, including a better account margin as we work through the fiscal year.

Jim Schneider (Senior Equity Analyst)

That's helpful. Thank you. Maybe as a follow-up, you gave many data points relative to your AI project traction and increasing size of deals in AI. Can you maybe give us or level set any kind of quantification in terms of the size of your average AI project today and then where you hope it may go in, say, two years?

Balazs Fejes (CEO and President)

I think our prepared remark, we actually kind of explained how the projects are evolving, moving from proof of concepts to medium to large-scale engagements. It is an evolving set. We have hundreds of engagements right now. Most engagements typically start on a small side as typical in the proof of concept. As they are scaling up, some of them get into the tens of millions of dollars range as we go forward. We do see most of our top 100 clients are actually engaging with EPAM with large AI initiatives, which looks like an AI initiative. It does not mean that right now we are executing large AI projects, but they have large AI initiatives. We are hoping that most of our revenue would be coming in the coming years from these initiatives, either because of AI transformation, either because of introducing AI, or creating the foundational elements for the AI deployment, which is right now one of the main drivers for our business.

Jim Schneider (Senior Equity Analyst)

Thank you.

Operator (participant)

Your next question comes from the line of Jamie Friedman with Susquehanna.

Jamie Friedman (Senior Fintech and IT Services Research Analyst)

Hi. Good morning. FB, some of your comments were considerably more technical than what some of us are accustomed to. We appreciate that because that is where the industry is going. We will adjust. I wanted to ask specifically about agentic delivery, lifecycle management. Yeah, that one. In terms of the vectors or phases that the customers need in order to proceed with agentic delivery, how would you describe the chronology of that aspect and the relative size of that one and delivery lifecycle relative to some of the others that you mentioned?

Balazs Fejes (CEO and President)

Jim, thank you for the question. What we need to start considering is when clients were just delivering software products, they still had to master SDLC cycles. Most clients, most of our clients, and most of the industry have not really fully mastered the SDLC itself. When you start deploying agentic capabilities and trying to automate.

Large scale of processes and works using agentic AI capabilities, you need to really follow an agentic development process, agentic lifecycle. It is as complex or even more complex than, compared to SDLC. It is not as simple as it looks in order to make it really work in the enterprise realm. I know that everybody believes and keep believing that this is a simple step. Agentic development lifecycle is as complex as any SDLC lifecycle, and you need to really master it. This is going to rethink, it is going to be a bigger problem going forward than mastering SDLC. The reason being is now you are no longer just touching on certain elements of your software spec, but now you really need to consider how you automate processes which you never automated before. You are going to step into automating previously very manually intensive.

Components, and automating those are very, very complicated and difficult. And error-prone. Process itself. As you're migrating in this, it's not just you need to upskill your teams who are doing it. And it's no longer just engineers who we're talking about. You need to also introduce the right tooling, the right processes across the enterprise. If you want to really get the benefit, you need to also start considering not just the element of, "Can I actually automate this process?" During automation, I really also want to achieve a certain ROI. Because if the automation results into a higher cost, then you kind of didn't deliver on your promise. We think this is a large shift. It is not going to be a very quick one. It requires a tremendous amount of work to make it happen.

Companies will have to partner with organizations such as EPAM to actually do that in what we call an AI factory model where you introduce the foundation, build the foundational component, build the right process in place, the right governance itself, and then you go process by process. Building the agentic automations. That's great. That's a great answer. Thank you. I'll drop back in the queue.

Operator (participant)

Your next question comes from the line of David Grossman with Stifel.

David Grossman (Research Managing Director)

Good morning. Thank you. Just kind of looking at high level at the business and how it's been performing. It looks like the revenue per head is up for the first time in three quarters despite flat utilization. Maybe you could just talk a few minutes about what's going on under the covers here. Is it geographic mix? Where are you delivering from? Is it perhaps growth outside the top 20, which has been very strong and historically a pretty good leading indicator of kind of new business activity and funnels? Maybe you could just illuminate what's happening there.

Jason Peterson (CFO)

Hey, David. Good question. One of the things about the revenue per head count number is that there is a lot of, I guess, what I'd have to call noise in it. Utilization, as you pointed out, is one of the factors. Foreign exchange actually impacts as well. It is a number that I think most people will look at and try to sort of draw a conclusion from how much price uplift is a company getting as they increase revenue per head count. I think there's more noise in it than I think people realize. Okay.

As we look at our number and I subtract out some of these factors that I just referred to, what we're really seeing is we are getting somewhat better price than we've gotten in the past. Some of that is probably mix-related. Again, I would say more kind of customer mix. As we talked about, it's consistent with the account margin improvement that I referred to earlier in the call. Some of it is foreign exchange, okay, but some of it really is actual price improvement.

David Grossman (Research Managing Director)

When did the contract profitability—when is this the first quarter that it really inflected, or has it been inflecting and just not visible?

Jason Peterson (CFO)

Yeah. I think we've been working on it throughout the year. I think we've talked about it a fair bit last quarter. It's all the things that drive that, including pyramids. Again, the pyramid that they will probably have more of an impact on 2026. I think it's just really beginning to probably show up in this discussion as we see both solid profitability in Q3 and what we're now expecting is much better profitability in Q4 than we originally anticipated 90 days ago. Again, some of that is account profitability improvement. I think, as I said earlier, I was very convinced that we were going to operate at about 15% this year. Now, as you heard me, we're talking about operating in a 15%-15.3% range.

David Grossman (Research Managing Director)

Got it. All right. That's it for me. Thank you.

Operator (participant)

Your next question comes from the line of Bryan Keane with Citi.

Bryan Keane (Head of North America Payments Processors and IT Services Research)

Hey, guys. Congrats on the solid results. Jason, let me just follow up on that discussion. What does that mean for headcount growth going forward in this model and maybe even revenue per head going forward? How should we think about that as we get into the fourth quarter and into next year? My second question is just the organic growth of FD and Neoris. Maybe you can help us with that. Thanks.

Jason Peterson (CFO)

Yeah. You would expect to see us add headcount in Q4. It will be similar to what we have been doing throughout the year, where we do have some pockets of excess bench that we continue to kind of reduce, and we will be making net additions globally. You will see an increase in headcount in Q4. From an FD and Neoris standpoint.

As we talked about very early in the year, the lead customer at Neoris was impacted early by U.S. tariffs and kind of generally kind of political and economic instability in Mexico. We definitely see a decline in that customer on a year-over-year basis. It probably has a modestly negative impact on organic cost and currency growth. Both of those businesses have kind of stabilized at this point, and we think there is a lot of strategic benefits. Again, particularly with that lead customer from Neoris, it has a modestly negative impact on organic cost and currency growth in Q4.

Bryan Keane (Head of North America Payments Processors and IT Services Research)

Okay. That is helpful. Any comments on revenue per head on what that might look like going forward?

Jason Peterson (CFO)

It is always utilization and foreign exchange really moves the needle on this one. It is difficult for me to tell. What I will just do is comment on pricing. What we do think is that pricing is better at this time than it was last year at the same time. Maybe it has not improved a lot over the last 90 days, but it is a somewhat better pricing environment. We are expecting modest price increases as we enter 2026. Again, maybe not at the level that we would have gotten four or five years ago, but in kind of the low single-digit kind of range. Thank you. Which is, again, a better environment than certainly 2023 was. 2024.

Bryan Keane (Head of North America Payments Processors and IT Services Research)

Yeah. No doubt. Thanks so much.

Jason Peterson (CFO)

Next comes from the line of Sean Kennedy with Mizuho.

Sean Kennedy (Director and Senior Analyst in Payments and IT Services Equity Research)

Hi. Good morning. Very nice results. Great to see the growth momentum in the business. I have a follow-up on the AI projects. I appreciate it's still early, but how does the AI work differ from EPAM's non-AI projects in terms of duration and profitability now? And how do you think that could evolve in the future? Also, are you seeing certain clients in terms of size and industry engage in AI projects more than others? Thank you.

Balazs Fejes (CEO and President)

Sean. Nice meeting you. I think the project, I don't think, is fundamentally different in AI. It requires the same engineering discipline, engineering capability. It might have more scoot towards data or scoot towards data platform or the capability around AI. It requires a little bit different engineering skill set, different understanding. On the other hand, it also requires a combination of deep business domain understanding. Really, you need to combine it because now you are starting automating processes which were not automated before.

Building the platform is probably very similar to what we've done in the past. Preparing the foundation, cloud migration, data platform build-out, data engineering, or modernizing the backends. This is very typical for EPAM. When you are really starting automating new processes, that's where domain capabilities, select capabilities, and understanding of how to automate that process and understanding the specific industry is where needed. Profitability-wise, at this point, probably similar to others. I think there is a clear potential for better profitability as you are potentially not just delivering the projects maybe on a time and material basis, but maybe you can explore alternative business models too. I'm hoping, of course, that with this kind of projects, we are able to charge probably higher rates to begin with.

Sean Kennedy (Director and Senior Analyst in Payments and IT Services Equity Research)

Great. Appreciate the color. Thank you.

Operator (participant)

Your next question comes from the line of Darrin Peller with Wolfe Research.

Paul Obrecht (Equity Research Associate)

Hi. Thanks. This is Paul Obrecht on for Darrin. Jason, I appreciate all the color on headcount. Just curious, longer term, if you think the greater usage of AI will perhaps impact the need to hire in any way? And just to what degree, as you increasingly embed AI internally, is that perhaps allowing for lower delivery requirements?

Jason Peterson (CFO)

I'm going to turn that one over to FB.

Balazs Fejes (CEO and President)

We continue to believe that although AI does create efficiency against your demand increase, we'll outstrip any kind of efficiency gains that we are seeing. We believe that going forward, as we continue to hire, continue to grow, the organization will grow. We need to bring in maybe differently trained teams. And we're also bringing on. I'm anticipating your next question. We're going to continue bringing in junior engineers because with the right training, with the right background, with the right education, we do believe that the balanced pyramid is the best serving not just our clients, but also EPAM too.

Operator (participant)

Your next question comes from the line of James Faucette with Morgan Stanley. Your line is open.

Antonio Jaramillo (Equity Research Senior Associate)

Hey, guys. How are you? It's Antonio on for James Faucette. I wanted to ask more back to the AI part of the equation. Just on your build versus buy strategy. I know that you had touched on that earlier, but I'm just trying to get a sense of what is the growth of your Gen AI revenue.

Balazs Fejes (CEO and President)

Yeah. A little bit hard for me to tell exactly what you're looking for there, but what we continue to see is this strong sequential improvement in revenues for what we call the Gen AI native. That continues to be kind of double-digit sequentially. We saw, again, here in this quarter. As FD has been talking about, and maybe once we add some color, we continue to see strong growth in what we call the foundational side, which is the cloud modernization and data in that piece of the business.

We continue to see lots of demand coming in. As Jason mentioned, the AI native revenue is growing sequentially very strongly, with double digits. We are seeing our clients building more solutions. Actually, they're taking advantage of AI, a software engineering feature or a functionality because the functionality close to it is decreases. They're actually building more. We believe that, going forward, people will build more than buy. The equation or the percentages will start scooting towards build versus the buy side. That's our thesis, and we are seeing evidence around that.

Antonio Jaramillo (Equity Research Senior Associate)

Got it. Got it. That's helpful. As a follow-up, I wanted to ask on the software and high-tech vertical, what are some of the key drivers for that growth? I know it's grown pretty nicely sequentially. Any one-time factors there, or is this just a broadening out of demand there?

Balazs Fejes (CEO and President)

Yeah. I mean, we've had a few large customers that are growing nicely. We've got one client that particularly has a large kind of platform program that they've been investing in. You won't see the growth rates stay like that forever in that space, but we've been pleased with the ongoing revenue generation from the high-tech portion of our portfolio.

Antonio Jaramillo (Equity Research Senior Associate)

Great. Thank you, guys.

Operator (participant)

At this time, there are no further questions. I will now turn the call back over to FB for closing remarks.

Balazs Fejes (CEO and President)

Thank you very much for attending my first earnings call. Really, I would like to thank all the EPAM employees for delivering a successful quarter. We talk next time in 90 days, approximately. Thank you.

Operator (participant)

Ladies and gentlemen, that concludes today's call. Thank you all for joining. You may now disconnect.