EPAM Systems: A Value Investing Deep Dive

On billing, EPAM still relies mainly on time-and-materials. In 2025, time-and-materials contracts were 80.0% of revenue and fixed-price contracts 19.5%; by Q1 2026, time-and-materials fell to 77.8% and fixed-price rose to 21.7%. This carries two implications. First, revenue is highly tied to delivery capacity and client demand and is by nature not the high-stickiness subscription of SaaS. Second, as long as clients keep advancing digitization, platform upgrades, and AI deployment, EPAM's client relationships still carry a degree of repeatability, because many projects gradually expand from single-point development into ongoing collaboration.

Is the revenue recurring, stable, and predictable? My judgment: moderately recurring, not highly predictable. It does not fit the model where "a contract signed this period auto-renews for the next decade," but it is also not a "one-and-done" sale. The company itself emphasizes long-term client relationships and discloses that top-client concentration is falling; in 2025, revenue from clients outside the Top 20 rose to 68.1%. Still, revenue fluctuates with the start and end of multi-year projects, client budget cycles, and the macro environment, as the 2023 revenue decline, near-zero growth in 2024, and recovery in 2025 directly show.

On cost structure, the core cost is people. In 2025, the most important cost components were delivery-personnel compensation, bonuses, benefits, project travel, and subcontractor fees; the main drivers of higher costs in 2025 included the 2024 acquisitions, wage increases, higher variable compensation, reduced Polish R&D tax incentives, and unfavorable foreign exchange. In other words, this is not a model where "the bigger you get, the closer marginal cost approaches zero"; it remains labor-intensive but capital-light. Capital spending is in fact very light: 2025 CapEx was only $42.2 million, under 1% of revenue, which is an important reason its cash flow can look quite good.

As for dependence on a few suppliers, channels, policies, or key people, EPAM's most sensitive exposure is not "supply chain" but talent supply, geographic footprint, client budgets, and the regulatory/tax environment. At the end of 2025, the company had roughly 56,600 delivery personnel; India is now its largest delivery location with 12,200 delivery staff, Ukraine still has about 8,750, Poland about 5,050, Belarus about 3,400, and Mexico about 2,950. This means its talent and delivery system is far more diversified than before 2022, but Ukraine and Eastern European risk have not fully disappeared.

If the stock market were closed for five years, would I be willing to keep holding this business? Yes, but only if the entry price is not too high. I understand how it makes money and acknowledge it is a capable engineering services firm; but I would not hold it as a Coca-Cola- or Visa-style "naturally very wide moat" enterprise. It is more like an excellent but still cyclically and competitively exposed high-end services provider. Business understandability score: 4/5.

Industry, Competition, and Moat

EPAM's industry is more accurately the intersection of digital engineering, software product development, technology consulting, and IT services. Long-term demand has not disappeared and is in fact still growing: in early 2026, Gartner projected global IT spending would reach $6.15 trillion, up 10.8% year over year; IDC projected global AI spending would reach $632 billion by 2028; and ISG's description of the U.S. digital engineering services market explicitly notes that generative AI and agentic AI have already permeated R&D, design, build, test, run, and platform services. So the long-term direction is right, but budget cadence, procurement cycles, and project priorities will be highly volatile.

This is not a declining industry, but it is by no means a "win-by-default industry." Its demand is stable and rising over the long run, yet easily disrupted in the short run by technology shifts, client budget cuts, and changing delivery methods. EPAM's own growth slowdown in 2023 and 2024 is the clearest example; the recovery in 2025 shows demand is still there, but this looks more like "cyclical and execution repair" than an industry with a permanent tailwind.

The main competitors are not hard to identify: the largest in scale and breadth is Accenture; large traditional IT services rivals include Cognizant; closer to the digital engineering and high-end development positioning are Globant, Endava, and others. The current valuation spread the market assigns these companies reflects its view of the industry: Accenture trades at about 24.9x earnings, Cognizant about 17.6x, Globant about 15.6x, Endava about 36.0x, and EPAM about 14.7x. On absolute valuation, EPAM is no longer treated as a "high-flying growth stock"; the market has placed it in the "growth recovery yet to be proven" camp.

I would define EPAM's industry position as: not the largest, but with a real seat at the table in high-end digital engineering and complex delivery. It lacks Accenture's ubiquitous top-level CEO advisory relationships and has no platform-style network effects; but its engineering reputation in product engineering, complex software delivery, and financial and enterprise digitization projects is genuine. After acquiring First Derivative and NEORIS in 2025, the company clearly strengthened its financial services, Latin American coverage, and European business, with 2025 Americas and EMEA revenue growing 12.9% and 19.7% respectively, both clearly benefiting from these acquisitions and recovering client demand.

On the moat, my judgment is moderate-to-weak to moderate, not strong. Brand advantage: Yes, but mainly as "engineering credibility" within the B2B technology community and at the enterprise procurement level, not a mass-market brand. Cost advantage: Some global delivery cost advantage, but not impossible to replicate; the India, Eastern Europe, and Latin America footprint improves cost structure, but peers can do the same. Scale advantage: Moderate scale advantage. At the end of 2025 total professional staff were 62,850 and delivery personnel 56,600, stronger than many mid-sized engineering services firms but far below giants like Accenture. Network effects: Essentially none. Switching costs: Yes, and this is EPAM's most genuine moat. Once complex products, core systems, and long-term engineering teams are embedded in a client's processes, switching vendors brings delivery risk, knowledge-transfer cost, and organizational friction. But this switching cost is at the project/account level, not the whole-business-model level. Channel advantage, patents/licenses, data advantage: None of these are strong. Culture and operating capability: This is the part of EPAM that genuinely deserves attention. The company has long emphasized engineering culture, delivery capability, and utilization management, with 2025 utilization averaging about 76.8% for the year, and it launched a cost optimization program in 2025 to improve utilization and profitability, showing management clearly understands that "for a services firm the most critical thing is not concepts but delivery and utilization."

Is the moat widening, stable, or narrowing? My judgment: stable overall, but the traditional part faces narrowing risk while AI-related new capabilities offer widening opportunity. On one hand, AI and automation will weaken pure labor-stacking development demand, and the company itself acknowledges in its risk factors that AI tools may improve its own and clients' productivity and thereby reduce demand for external services; on the other hand, for enterprises to truly land AI at the process, data, platform, and security layers, they need exactly the engineering integration capability EPAM offers. In other words, EPAM's future lies not in "selling more person-days" but in "upgrading the labor model into AI-native engineering services." If this transition succeeds, the moat holds; if it fails, the moat narrows.

Can the company raise prices? The answer: yes, but not strongly. In 2025 the company explicitly noted that rising costs were "not fully offset by price increases." This shows EPAM is not without pricing power, but its pricing power is insufficient to cover all wage and FX pressure. In other words, it can raise prices, but it does not have the strong pricing power to dictate whatever it wants.

Can the company stay profitable in a downturn? Historically, yes, but margins get compressed. Revenue declined in 2023 but it stayed profitable; in low-growth 2024 it still earned $455 million in net income; in 2025, under the amortization and integration pressure from acquisitions, it still earned $378 million in net income. But margins are well below the 2021 peak, showing it has resilience without being a "cycle-defying champion." Industry attractiveness score: 3/5. Moat strength score: 3/5.

Management and Capital Allocation

The most important management change is the 2025 CEO transition. In May 2025 the company disclosed that founder and long-serving CEO Arkadiy Dobkin would become executive chairman, and that Balazs Fejes would serve as CEO and president from September 2025; the 2026 proxy described this transition as a "deliberate and thorough succession planning process." This is a positive in itself: it was not a reactive change of leadership but a planned transition from founder leadership to a professional successor.

But for long-term shareholders, the key is not only "whether the leadership change went smoothly" but "whether the successor has been tested through a full cycle." Fejes previously ran global business and revenue and has ample market and delivery experience; the issue is that his verifiable track record as CEO is still very short. For an investment held 10-plus years, this means: governance continuity is good, but the new CEO's capital allocation ability remains in the observation phase for now.

On alignment of interests, founder Arkadiy Dobkin held roughly 1.523 million shares as of March 2026, about 2.9% of shares outstanding; Fejes personally held about 71,700 shares. This means the company is not the "very high management ownership" type, but neither is it without alignment. Meanwhile, the company revised its stock ownership guidelines in November 2025: the CEO must reach a holding of 6 times annual salary within five years, and must retain 100% of net shares acquired until the target is met; other executives and directors face stricter net-holding requirements as well. Overall, the governance structure is moving in a shareholder-friendly direction.

Capital allocation should be viewed separately across reinvestment, M&A, buybacks, and dividends.

On reinvestment, EPAM's organic reinvestment needs are low. Annual capital spending is small, with more money going to hiring, training, sales, and M&A. This is normally a good thing: capital-light means that if management is disciplined, shareholders can receive more cash; the downside is that management can also more easily deploy surplus cash into M&A or buybacks at the wrong price.

On M&A, the big moves were First Derivative and NEORIS in 2024. The company disclosed that the two acquisitions contributed a combined $65.9 million in revenue in 2024, but that 2025 growth in both Americas and EMEA clearly benefited from the new clients and capabilities these acquisitions brought; the company also provided pro forma figures for NEORIS, noting that if consolidated for full-year 2024, EPAM's 2024 pro forma revenue would have been about $5.015 billion and pro forma net income about $407 million. My read: the strategic logic holds, but the financial return is still being proven. They do not look like obviously "junk acquisitions," but they are also far from "M&A master" territory.

On buybacks, the assessment must be more demanding. In October 2025 the company authorized a new $1 billion buyback; it repurchased $660.6 million in full-year 2025 and $398 million in 2024; in Q1 2026 it executed a $300 million ASR, repurchasing 1.835 million of the 18.35 million shares within the quarter for $264 million of cash, with an initial settlement price equivalent to about $140.9 per share, while the average open-market repurchase price in Q4 2025 was also between $180–205 per share. Looking back from today's $102.46 price, these buybacks were clearly not "Buffett-style shrewd." They show management's willingness to return capital but not necessarily a strict discipline of deploying capital at undervalued moments. This is my most important reservation about management's capital allocation.

On equity compensation, the stock-based compensation expense recognized in 2025 was $176.8 million, which is not low; the good news is that, by results, total share count has not faced runaway dilution. On the contrary, shares outstanding at year-end fell from 56.869 million at the end of 2024 to 54.274 million at the end of 2025, and further to 52.757 million by Q1 2026. This shows that although management issues a meaningful amount of stock, the buyback intensity is enough to cover the dilution. Still, for shareholders this remains a real cost that cannot be pretended away.

Does management speak candidly about risk? On the whole I think it is acceptable. The company consistently discloses the Ukraine impact, the cost optimization plan, M&A integration, Polish tax incentives, FX, and AI risk, and does not package everything into an "only opportunities, no challenges" story. Management and capital allocation score: 3/5. I trust its operating ability and credit its governance continuity, but I would not give capital allocation a high mark.

Financial Quality and Owner Earnings

Start with the most important financial profile of the past five years: EPAM's revenue grew rapidly from 2021 to 2022, came under clear pressure in 2023 and 2024, and returned to double-digit growth in 2025; margins fell back from the 2021 peak and have not returned to old levels. Cash flow performance is overall better than the income statement, capital spending is very low, and the balance sheet is very sound.

The revenue, profit, cash flow, balance sheet, and share data in the table are compiled from EPAM's 2021, 2022, 2024, and 2025 annual reports; free cash flow is operating cash flow minus capital expenditures; ROE is a rough estimate based on average equity; net cash is cash and short-term investments minus interest-bearing debt.

This table has three important conclusions. First, EPAM's return on capital is trending down. 2021 was a very strong year, but since 2022 ROE, ROA, and operating margin have all stepped down a tier, driven by the Ukraine shock, personnel and geographic restructuring, M&A amortization, FX, and softer industry demand. Second, cash flow quality is overall good, with FCF in most years close to or above net income. Third, growth does not require heavy capital investment; it is not a company where "the more it grows, the more cash it needs."

But seeing only "FCF is very high" would make the analysis too optimistic. The reason 2025 free cash flow reached $613 million does not fully mean the "cash printing machine" essence has greatly strengthened. At least two items deserve to be unpacked. First, the company recognized $55.2 million of Polish R&D tax incentives in 2025, a direct cost offset; second, the company's stock-based compensation expense in 2025 reached $176.8 million, which is added back in cash flow but is not free to shareholders. In other words, 2025 free cash flow has quite good components, but it may not be replicable in the same form indefinitely.

Now the balance sheet. At the end of 2025 total assets were $4.902 billion, total liabilities $1.224 billion, and total equity $3.678 billion; by Q1 2026, the ASR and financing arrangements pushed interest-bearing debt up to $165 million, but the company still held $1.037 billion in cash and $4.3 million in short-term investments, with net cash still about $876 million. This shows the company has no financial leverage risk, net debt/EBITDA is actually negative, and the traditional "interest coverage ratio" is not very meaningful either, because the company's "interest and other income, net" was still positive in 2025. For a value investor, this balance sheet greatly reduces the probability of permanent capital loss.

On working capital, inventory is almost irrelevant since this is not manufacturing; the focus is receivables and contract assets. Receivables and contract assets were $1.002 billion at the end of 2024 and $1.108 billion at the end of 2025, with DSO rising from 70 days to 72 days, a change that is not out of control. The company also disclosed in Q1 2026 that operating cash flow turning negative was mainly due to higher variable compensation payments, not a broad deterioration in client collections. Such Q1 swings are not unusual at services companies, but they still bear watching, because once the IT services cycle weakens, receivables, utilization, and cash collection come under pressure together.

On accounting quality, I do not see very typical red flags. Deloitte has served as the company's auditor since 2006 and issued an unqualified audit opinion in the 2025 report; the company consistently discloses geopolitical events, tax incentives, restructuring charges, M&A amortization, and stock-based compensation. What investors truly need to watch for is not "signs of financial fraud" but mistaking tax incentives, SBC add-backs, and a temporary working-capital tailwind for permanent owner earnings. This is not a fraud issue but the optimism bias most easily committed in valuation.

Owner Earnings Estimate

Here I deliberately distinguish four categories of information:

Facts: 2025 net income was $377.68 million; operating cash flow was $654.93 million; capital expenditures were $42.24 million; free cash flow was $612.70 million; stock-based compensation expense was $176.76 million; Polish R&D tax incentives were $55.2 million.

Assumptions: For a capital-light services company like this, maintenance capital expenditure is usually not far from reported CapEx, so under a conservative lens I treat all CapEx as maintenance CapEx; at the same time, I do not treat all of 2025 free cash flow as permanently distributable cash flow, but instead discount for the tax incentives, the economic cost of SBC, and part of the working-capital tailwind.

Inferences: Under the most lenient owner-earnings lens, 2025 owner earnings roughly equal FCF of $613 million; under a more conservative lens closer to long-term shareholders, I would put sustainable owner earnings between $500 million and $550 million, with a midpoint of about $525 million. This implies the current market cap corresponds to about 10.6x conservative owner earnings; calculated directly on 2025 reported FCF, it is about 9.1x.

Views: For a company with net cash, low CapEx, and still mid-single-digit-plus long-term growth potential, conservative owner earnings of around 10x is not expensive. But the premise is that you accept EPAM is not an "ironclad moat" but a services firm that must keep proving its engineering value, utilization, and competitiveness in the AI era.

Valuation and Margin of Safety

Owner Earnings Discount Method

I am unwilling to give a single "magically precise value," because the key to EPAM's valuation is not the formula but your judgment of long-term sustainable owner earnings. Below are my three cases, all using a free cash flow to equity / owner earnings approach rather than profit alone:

The core meaning of these ranges: if EPAM is merely a "stable cash flow but ordinary moat" engineering services firm, then today's price is roughly near the conservative valuation; if it can deliver on growth recovery, M&A integration, and AI transformation, then today's price still carries a meaningful discount to fair value. So the current share price is not "obviously mispriced," but it is by no means "unattractive" either.

Relative Valuation Method

Placing EPAM among its peers makes clearer what the market is pricing in now. Current EPAM P/E is about 14.7x; dividing current market cap by 2025 FCF, P/FCF is about 9.1x; using market cap plus debt minus cash, EV is about $4.68 billion, and against 2025 operating income of $520 million, EV/EBIT is about 9.0x; on year-end 2025 shareholders' equity, P/B is about 1.5x. For a company with net cash that returned to double-digit growth in 2025, this set of figures is on the low side overall.

Compared with peers, Accenture trades at about 24.9x earnings with 2025 free cash flow of $10.87 billion, corresponding to about 18.7x P/FCF at current market cap; Cognizant trades at about 17.6x earnings with about $2.23 billion in 2025 net income and a 120% free-cash-flow conversion rate, implying about 13.8x P/FCF; Globant trades at about 15.6x earnings with 2025 revenue up only 1.6% and an IFRS operating margin of 7.0%, but on a rough calculation using the four quarters of official 2025 cash flow data, P/FCF is a bit above 8x. On a relative basis, EPAM is far cheaper than Accenture and Cognizant, and no more expensive than Globant, which has slower growth and lower margins. Its discount mainly corresponds to: a weaker moat, more geopolitical and delivery risk, and margin repair that has not yet fully materialized.

Asset and Liquidation Value Method

EPAM is not an asset stock suited to buying on a "liquidation revaluation." At the end of 2025 total equity was $3.678 billion, of which goodwill was $1.211 billion and intangible assets $407 million; excluding both, tangible net assets are roughly $2.059 billion; as of Q1 2026, net cash is about $876 million. So this company's true value comes mainly from its talent organization, client relationships, and delivery capability as a going concern, not from land, inventory, or hard assets. If I forced a very conservative liquidation lens, I think the "cold-blooded floor" its tangible net assets and net cash could provide is roughly $35–45 per share; this is not a buy thesis, only an indication that the downside is not entirely without an anchor.

Margin of Safety Judgment

So, is the margin of safety sufficient? My answer: yes, but mainly because the price is already not expensive, not because the business model is flawless. There are two most fragile valuation assumptions: first, whether 2025's $500–550 million of conservative owner earnings can hold; second, whether EPAM still has at least mid-single-digit long-term growth. If growth comes in below expectations and margins keep falling, today's valuation appeal will clearly contract; if growth merely slows but cash flow stays stable, the current price most likely still offers a long-term return above the 10-year U.S. Treasury.

My conclusion: Conservative intrinsic value range: $95–110 Fair intrinsic value range: $125–145 Optimistic intrinsic value range: $160–180 Current price relative to fair value: roughly a mid-single-digit to low-double-digit discount; relative to conservative value, close to fair. Ideal buy price: $80–95 Acceptable holding price: $95–130 Clearly overvalued range: above $160.

Risks, Comparisons, and Checklist

The most important risk is not short-term share-price volatility but permanent capital loss. From this angle, I would rank EPAM's risks into the following categories.

The first is competition and technology substitution risk. EPAM's value is built on "high-end engineering delivery still mattering." If AI substantially raises clients' internal development efficiency, or automates a large share of traditional outsourced development work, EPAM's labor-intensive model will be squeezed. The company has explicitly acknowledged in its risk disclosures that AI tools may reduce client demand for external services. Conversely, if EPAM can upgrade into an AI-native engineering services provider, it may also benefit. This is the biggest double-edged sword.

The second is geopolitical and delivery risk. At the end of 2025, Ukraine still had about 8,750 delivery personnel, and the company explicitly acknowledges that the situation in Ukraine and related sanctions and countersanctions could have a material adverse effect on its business and delivery; the company also retained a $100 million humanitarian commitment to Ukrainian employees, of which $10.1 million remained to be expensed as of the end of 2025. Although the company has scaled up India, Latin America, and other regions, Ukraine and Belarus remain variables that cannot be ignored.

The third is margin and utilization risk. The company's 2025 operating margin was only 9.5%, well below the 14.4% of 2021; while Q1 2026 recovered to 8.3%, it is still not high. The company also launched a cost optimization plan in 2025 and expects related restructuring actions to continue into Q2 2026. This shows management believes margins can improve, but it also proves margin recovery is not easy.

The fourth is capital allocation risk. The buybacks of 2025 and early 2026 mostly occurred, in today's view, at prices above the current share price; if management continues to repurchase heavily at elevated valuations or pursue expensive M&A, shareholder value will be eroded. Buybacks are not a virtue in themselves; buybacks at undervaluation are the virtue.

The fifth is accounting and valuation illusion risk. EPAM has no obvious signs of financial fraud, but 2025 free cash flow was very high, including a large SBC add-back and $55.2 million of Polish tax incentives. If investors simply extrapolate 2025 FCF linearly and apply a high-growth multiple, the valuation can easily be overstated.

The Strongest Bear Case

The strongest bear logic is actually simple: EPAM may not be a "wrongly punished quality company" but merely "an IT services vendor not yet proven capable of rebuilding high margins in the AI era." If AI reduces the labor needed for much traditional development work, client procurement will increasingly favor cheaper, more platform-based, more automated delivery; this would gradually compress EPAM's differentiated engineering advantage over several years. At the same time, the revenue scale added by the large 2024 acquisitions does not automatically equal added intrinsic value per share; if margins cannot hold, SBC stays high, and buybacks remain price-insensitive, the current "low valuation" may be only the entrance to a value trap. I believe this bear case deserves to be taken seriously.

Which facts would make me admit I was wrong? Four are most critical: First, revenue growth over the next two to three years falls back to low single digits or even negative, while free cash flow cannot stabilize around $500 million; Second, operating margin stays at 8–9% for the long run, unable to prove that the 2025 pressure was only temporary; Third, AI-related business fails to offset pressure on traditional development demand, leading to continuously declining utilization and repeated layoffs and restructuring; Fourth, management keeps making large buybacks or expensive acquisitions at elevated valuations. Once these facts appear, the premise of the current "Cautious Buy" would be broken.

Comparison with Other Opportunities

Against its strongest competitor, Accenture, EPAM is far cheaper but its business quality is also somewhat weaker. Accenture's scale, client relationships, consulting entry point, and AI commercialization are all steadier, with fiscal 2025 revenue of $69.7 billion and free cash flow of $10.87 billion, and its valuation is also markedly higher. EPAM's upside is that you do not have to pay so much for "certainty" today; the downside is that you must bear more execution and model risk.

Against broad indices, EPAM's current valuation is clearly lower. FactSet's early-May 2026 forward P/E for the S&P 500 over the next 12 months was about 20.9x, above the 5-year and 10-year averages; while EPAM's current trailing P/E is about 14.7x. But cheapness is not the only consideration here: the index's core advantages are diversification, stability, and "mistakes not being fatal"; EPAM's core advantage is a higher probability of structural undervaluation. For someone who has no wish to study companies at all, I would not say "buying EPAM is clearly better than buying the index"; but for someone willing to bear company-specific risk, EPAM's potential risk compensation is higher.

Against the risk-free rate, the U.S. 10-year Treasury yield was about 4.45% on May 28, 2026. If my base-case judgment for EPAM roughly holds, its expected annualized return over the next decade should have a chance of reaching low double digits, above Treasuries; but if the conservative case holds, the long-term return may be only slightly above Treasuries. So it carries a risk premium, but that premium is not given away for free.

If I could hold only five assets, EPAM most likely would not automatically enter my top five, unless my circle of competence happens to be concentrated in software engineering, IT services, and enterprise digitization. It is more like a mid-to-high-quality name with an odds advantage than the highest-certainty "ballast." This is also why it is better suited to "Cautious Buy" rather than "heavy buy."

Checklist

The core meaning of this checklist: EPAM passes the three gates of "understanding, financials, and valuation," but fails the three gates of "strong moat, strong pricing power, and supreme peace of mind in holding." For a value investor, this means it is better bought when the price is attractive, not because the story sounds appealing.

Final Judgment and Limitations

【Final Rating】 Cautious Buy

【One-Sentence Investment Thesis】 EPAM is a high-end digital engineering services firm with solid cash flow and a strong balance sheet whose valuation has already pulled back significantly, but its long-term return depends more on "the right price plus continued execution" than on an indisputably deep moat.

【Core Bull Reasons】

• Valuation is not high: P/E about 14.7x, P/FCF about 9.1x, EV/EBIT about 9x, attractive for a net-cash company.

• Strong balance sheet: still about $876 million in net cash in Q1 2026, markedly reducing the risk of financial permanent loss.

• Capital-light with good cash flow: 2025 operating cash flow of $655 million and CapEx of only $42.24 million.

• Revenue recovered to growth in 2025 and still grew 7.6% in Q1 2026, showing the business has not lost demand.

• A relatively diversified client base, with the Top 5 at 13.7%, and acquisitions that strengthened financial services, Latin America, and Europe.

【Core Bear Reasons】

• The moat is not deep enough, with no brand monopoly, network effects, or strong licensing barriers.

• Limited pricing power; in 2025 the company explicitly stated that rising costs were not fully offset by price increases.

• The margin center has clearly fallen since 2021, with 2025 operating margin of only 9.5%.

• Ukraine/Eastern European risk persists, and execution continuity is going through a new-CEO transition.

• Buyback timing is not pretty, with large repurchases in 2025 and early 2026 at prices well above the current share price.

【Key Assumptions】

• Conservative owner earnings can hold above $500–550 million over the next 3–5 years.

• AI's effect on EPAM leans toward "enhancing engineering value" rather than "wholesale substitution of outsourced development."

• The 2024 acquisitions can bring long-term client relationships and profit synergies, not just amortization and complexity.

• Operating margin can at least stabilize from 2025's 9.5% and gradually repair.

【Fair Buy Price】 $80–95 is most ideal; $95–110 is acceptable for building a base position in tranches; if the price is above $130, I would be clearly more cautious. This range comes from the combined judgment of the conservative valuation of $95–110, the generally higher margin of safety U.S. services companies usually require, and the fact that EPAM's moat is not deep enough.

【Target Holding Period】 At least 5 years, and better suited to more than 10 years. In the short run it will likely keep being priced back and forth by industry sentiment, the tech investment cycle, and the AI narrative; only with long holding can valuation mean-reversion, per-share buybacks, and business recovery be realized.

【Expected Annualized Return】

• Conservative scenario: 6%–8%

• Base scenario: 10%–13%

• Optimistic scenario: 14%–17% These returns come not from "high-multiple expansion" but mainly from the owner-earnings yield at the current price, some growth, and the per-share value growth that buybacks may bring.

【Maximum Loss Risk】 If over the next two to three years the company proves it is in fact merely an ordinary IT services vendor with a lowered growth center, margins that cannot be lifted back, and AI still depressing demand, then the market may keep pricing it on low-single-digit growth and low-double-digit or even single-digit operating margins. In that worst case, a pullback toward around $70 or even lower is not unimaginable, with a permanent loss versus the current price of roughly 30%–40%. This risk comes not from bankruptcy but from "the disappearance of a high-quality illusion."

【Tracking Metrics】

• Revenue year-over-year growth, especially organic growth excluding M&A

• Operating margin and the gap between non-GAAP and GAAP margins

• Utilization and the quality of delivery-personnel growth

• The long-term relationship between free cash flow and net income

• Stock-based compensation as a share of revenue and of net income

• Whether the average buyback price is rational relative to intrinsic value

• Whether the client/profit synergies from the 2024 acquisitions materialize

• Changes in the Ukraine and Eastern European delivery share

• Whether AI-related business truly drives high-value-add projects

• Whether top-client concentration keeps declining

【Signals to Trigger Reassessment】

• Organic growth falling back to low single digits or even negative for several consecutive quarters

• Operating margin stuck at 8%–9% for the long run with no visible path to repair

• Free cash flow markedly below the conservative band of $500 million

• Management continuing to make large buybacks when the price is unattractive

• Poor M&A integration results, with accelerating goodwill/intangible impairment

• AI clearly cutting clients' outsourced engineering budgets, with EPAM unable to substitute with higher-value services

【Final Recommendation】 Put plainly, EPAM today looks more like a long-term investment with good odds, provided you acknowledge its business quality is not top-tier. For those who understand the IT services industry and are willing to track execution and valuation, I think it has entered a range worth studying and worth buying in tranches; for investors who only want to hold "super-moat consumer or platform stocks," I would still advise more restraint. The core reason to buy it is not "it is perfect" but "it is not priced for its true resilience"; the core reason not to buy it is equally clear: it is, after all, a services company that must keep proving its value.

Open questions / limitations This report has tried to prioritize EPAM's latest 10-K, 10-Q, company IR materials, and official/authoritative market data; but several limitations should be made explicit. First, EPAM does not separately disclose the full subsequent ROI of its acquisitions, so the ultimate capital return on NEORIS and First Derivative can only be moderately inferred. Second, the "comparable basis" for P/B, ROIC, and EV/EBITDA across peers is not fully consistent, and this report places more weight in relative valuation on P/E, P/FCF, EV/EBIT, and operating quality. Third, 2025 owner earnings are affected by Polish tax incentives, SBC, working capital, and other factors, so the valuation ranges should be viewed as a disciplined range rather than a precise point estimate.

This report is based on public information and does not constitute investment advice. Markets carry risk; invest with caution.