{
  "@context": "https://orbyd.app/schemas/dossier.v1.json",
  "ticker": "EPAM",
  "name": "EPAM SYSTEMS, INC.",
  "url": "https://orbyd.app/dossiers/EPAM/",
  "json_url": "https://orbyd.app/dossiers/EPAM.json",
  "status": "DORMANT",
  "current_conviction": "LOW",
  "graded_conviction": null,
  "archetype": {
    "code": "a4",
    "n": 4
  },
  "current_thesis": "AI-native revenue is compounding ($160M in Q2, >11% of total) but the task-based book is running off faster — FY26 organic constant-currency growth was cut to 2.0%–3.0% on 2026-08-06, Q3 guided to ~1.7% YoY, and the stock closed $97.43, 56% under its 52-week high. The compounder narrative is broken and nothing resolves before the ~November Q3 print.",
  "invalidation_trigger": "A weekly close below $92 breaks the post-guidance-cut shelf and confirms the de-rating leg is still running; secondarily, a Q3 report around 2026-11-05 guiding Q4 organic constant-currency growth below zero would end any stabilization case.",
  "catalyst_date": null,
  "outcome": "OPEN",
  "outcome_date": null,
  "invalidation_fired": null,
  "themes": [
    "ai-datacenter-infrastructure",
    "managed-care-health-services"
  ],
  "tags": [],
  "sources": [],
  "notes": [
    "EPAM reports on a calendar fiscal year; as of 2026-08-07 no Q3 2026 earnings date had been scheduled.",
    "GAAP and non-GAAP guidance diverge widely (FY26: $8.22–$8.38 vs $13.08–$13.24 diluted EPS) — check which basis a headline uses.",
    "No dividend. Capital return runs entirely through repurchases: $85M in Q2 2026, $409M year-to-date.",
    "Reported revenue growth exceeded organic constant-currency growth in Q2 2026 (4.5% vs 3.4%), so FX flatters the headline line."
  ],
  "body_markdown": "## Current Thesis\n\nThe leg an investor would have been buying here — EPAM as a secular digital-engineering compounder that converts AI into a bigger book — was repriced on 2026-08-06. Q2 revenue of $1.415B (+4.5% YoY, +3.4% organic constant currency) and non-GAAP EPS of $3.38 both cleared consensus, and the company still cut full-year revenue guidance to $5.632B–$5.686B from $5.675B–$5.813B, taking organic constant-currency growth to 2.0%–3.0%. Q3 was guided to $1.410B–$1.425B against a $1.445B consensus — about 1.7% YoY at the midpoint. CEO Fejes attributed it to clients moving budget out of task-based work faster than AI-led replacement work ramps, which he described on the call as \"creating a growth gap.\" That is the company itself describing structural substitution rather than a macro pause.\n\nLife-cycle: **DEAD**. The dating is specific — the 2026-08-06 guide cut, William Blair's downgrade to Market Perform the same day, and a 2026-08-07 close of $97.43, 56.0% below the $221.4 52-week high. A replacement narrative exists inside the numbers (AI-native revenue of $160M in Q2, more than 11% of total, a sixth consecutive quarter of double-digit sequential growth, tracked by management against roughly a $600M annual figure), but management put the meaningful revenue contribution from large AI-driven deals in early 2027. There is no dated proof point between now and the Q3 print.\n\n## Bull Case\n\n- AI-native revenue reached $160M in Q2 2026 — over 11% of total revenue and a sixth straight quarter of double-digit sequential growth (Q2 call, 2026-08-06). Management holds the roughly $600M annual trajectory despite cutting the top line.\n- Financial Services grew 11.5% YoY and Life Sciences & Healthcare 8% YoY in Q2 2026 (Q2 call, 2026-08-06) — where budgets exist, the AI-led modernization work is being bought.\n- The margin line did not break: non-GAAP income from operations was 16.4% of revenue, GAAP 10.8%, with non-GAAP diluted EPS of $3.38, up 22.0% YoY, against a $3.14 estimate (2026-08-06 release).\n- FY2026 non-GAAP EPS guidance was narrowed with the low end lifted — $12.98–$13.28 to $13.08–$13.24 against a $13.04 consensus — while revenue came down. Mix is moving toward higher-value work even as volume shrinks.\n- Balance sheet funds the transition internally: $789.4M cash, $85M of repurchases in Q2 and $409M year-to-date (2026-08-06 release). The company has been buying its own stock through the de-rating.\n- EPAM joined the OpenAI Partner Network as an Advanced Partner on 2026-07-28 — a credential and distribution datapoint, with no disclosed revenue attached.\n- On the published arithmetic: a $97.43 close against a guided FY2026 non-GAAP diluted EPS range of $13.08–$13.24 is roughly 7x that measure.\n\n## Bear Case\n\n- Full-year revenue guidance cut on 2026-08-06 to $5.632B–$5.686B (3.2%–4.2% growth) from $5.675B–$5.813B, with organic constant-currency growth of 2.0%–3.0%.\n- Q3 2026 guided to $1.410B–$1.425B versus $1.445B consensus — approximately 1.7% YoY at the midpoint, and organic constant currency of about 1.8%. The trajectory points at flat, not at a trough already passed.\n- Software & Hi-Tech revenue fell 1.3% YoY in Q2 2026 (Q2 call, 2026-08-06) on ramp-downs in non-AI services. The most AI-literate client cohort is cutting first, which is what a substitution thesis predicts.\n- Headcount ended Q2 at 62,850 with 56,650 delivery professionals, up 1.5% YoY, against 4.5% revenue growth. If billable-hour work is what agentic delivery compresses, this is the line under pressure.\n- CFO Peterson flagged client vendor-consolidation exercises where customers expect economic advantage — pricing pressure sits alongside the volume problem.\n- Sell-side marked down in one window: William Blair to Market Perform (2026-08-06); PT cuts to $120 from JP Morgan and TD Cowen, $110 from Truist and $146 from Susquehanna (2026-08-07), $120 from Needham (2026-08-06), $100 from Morgan Stanley (2026-08-04) and $100 from Citigroup (2026-07-27).\n- The re-acceleration is dated to early 2027 by management, which declined specific 2027 guidance. Two prints separate the story from its evidence.\n\n## Setup & Price Structure\n\n- Basis close $97.43 on 2026-08-07, 56.0% below the $221.4 52-week high, with a three-month return of -1.8%. Inference, not measurement: a roughly flat quarter that ends 56% below the high implies the tape had recovered ground into the print and the guide cut took it back.\n- No base is visible. One or two sessions of post-gap price is not a shelf, and there is no company-scheduled event to build one against before the Q3 report.\n- A 14-day RSI of 56.8 on the same 2026-08-07 basis is hard to square with a fresh guidance gap; the momentum reading looks like it lags the move and should not be read as strength.\n- Positioning observables, stated as observables: eight sell-side revisions in the 30 days to 2026-08-07, all downward, plus one rating downgrade. Post-cut targets still cluster at $110–$120 with a $100 floor and a $146 high — every published target sits above the $97.43 close, so the analyst complex has not marked to the tape. No earnings date falls inside the next 30 days. No insider transaction filings appear in the recent window. The company repurchased $409M year-to-date rather than issuing into strength.\n- What is absent matters: this is not a retail-clustered name. The coverage density is institutional downgrade flow, which typically produces grinding supply rather than a sharp squeeze.\n\n## Catalyst Calendar (next 30 days)\n\n- **None scheduled.** As of 2026-08-07, no company-dated event falls inside the window to ~2026-09-07. Flow in that period is sector read-across and residual estimate revisions.\n- **~2026-11-05 (est.)** — Q3 2026 results. Tests whether the ~1.7% guided YoY midpoint is the trough, whether AI-native revenue holds the sequential streak, and whether Financial Services' 11.5% growth rate persists.\n- **~2027-02-12 (est.)** — Q4/FY2026 results and initial FY2027 guidance. This is where management's \"significant revenue contribution from large AI-driven deals beginning in early 2027\" either appears as a number or does not.\n\n## What Would Change Our Mind\n\nThe question is arithmetic: whether the AI-native line ($160M in Q2, over 11% of revenue) grows faster in dollars than the task-based book runs off. Everything else is commentary on that ratio.\n\nConfirming the bear read: a Q4 guide at the November print implying a YoY revenue decline, or organic constant-currency growth below zero; Financial Services decelerating out of double digits; non-GAAP operating margin falling below the 16.4% Q2 level while headcount is flat. On price, a weekly close below $92 breaks the post-guidance-cut shelf and says the de-rating leg is still running.\n\nFlipping constructive would take more than a bounce: AI-native revenue clearing the roughly $600M annual pace with management quantifying an FY2027 contribution, plus a weekly close back above the $110–$120 target cluster to show the tape agrees. Absent both, this is a broken structure with a story dated to 2027, and the case for a fresh entry at $97.43 is weak.\n\n## Correlation Notes\n\n- EPAM sits in the IT-services and digital-engineering complex — Globant, Cognizant, Infosys, Wipro and Accenture are the read-across set. The mechanism is shared: genAI capability that bids up the model and compute layer is the same capability that compresses billable engineering hours downstream. Note the sign: this name can fall on the news that lifts AI infrastructure.\n- Within the group, Globant is the tightest comparison by revenue mix and client profile; Accenture's quarterly bookings commentary tends to set the sector's tone. No dated cross-sectional correlation data is presented here, so treat the peer linkage as a mechanism, not a measured beta.\n- Reported growth ran above organic constant currency in Q2 2026 (4.5% versus 3.4%), so a stronger dollar mechanically pressures the reported line independent of demand.\n- Client concentration by vertical creates its own factor exposure: with Financial Services the fastest-growing vertical at 11.5% YoY, bank and insurer technology budgets are now a larger share of the growth story than software and hi-tech, which declined 1.3% YoY.",
  "first_seen": "2026-07-30",
  "last_analyzed": "2026-08-08T07:44:00+00:00",
  "last_synthesized": "2026-08-08",
  "last_update_source": "watchlist_research",
  "license": "Content © orbyd. Cite the canonical URL."
}