EPAM Doubled in 90 Days. The Print Says "Not in 2027."

A 0.7x-revenue IT services name re-rated 94% into a quarter of +2.5% organic growth, three signed mega-deals, and an explicit admission that next year's gross margin may go down

EPAM · Earnings Call · 2026-11-05

The setup: a re-rating, not a recovery

EPAM's tape is a study in repair. The stock sits 76.9% below its November 2021 peak of $717.49 — and yet the last 90 trading days read simply: up 17 weeks, +94.4%. That is not a business inflecting on the tape; that is a multiple un-collapsing. The company still trades at 0.7x revenue, down 62% year-over-year and 93% from the 10.1x it commanded in Q4 2021. So the question into this print was never "is EPAM cheap?" — it is "does the evidence support the second derivative the tape already bought?" On that measure the quarter was good and the outlook was honest to the point of being deflating.

The quarter that was good, and the growth that wasn't

EPAM beat the high end of its own range on revenue, profitability and EPS. “Revenue was $1.426 billion, delivering year-over-year growth of 2.3%. On an organic constant currency basis, revenue grew 2.5%.” — Jason Peterson, Chief Financial Officer · 2026-11-05 The organic constant currency line is the first-ranked keyword of the quarter for a reason: with FX turning to a roughly $11 million sequential headwind in Q4, it is the only growth number management wants the street to anchor on. The genuine engine is AI-native revenue: more than $185 million in Q3, about 13% of total, up from roughly 11% the prior quarter, with the full-year target raised from $600 million to $650 million — one of the few upward revisions in this name's recent history. The difficult arithmetic sitting next to it is that everything else is shrinking. Software and high tech declined 6.1%, business information and media 5.5%, consumer goods, retail and travel 2.5%, and the Americas — 57% of revenue — managed +0.2%. Asked directly about a >6% decline in the non-AI book, the CEO did not flinch: “you are right mathematically, non-AI revenue is not growing with the same speed.” — Balázs Fejes · 2026-11-05 That is not new either. In August he had already described the mechanism: clients “prioritizing AI budgets, reprioritizing away from task-based services... This shift is happening faster than the replacement work is ramping, right, creating a growth gap for us.” — Balazs Fejes, President and CEO · 2026-08-06 Two quarters later the gap is being funded by a much smaller base: AI-native is compounding off ~$185M a quarter while the legacy book bleeds. And Q4 guidance decelerates — organic constant currency of 0.5% to 1.6% — before any 2027 help arrives. Which brings us to the sentence that matters most.

The most expensive three words on the call

Asked point-blank whether EPAM returns to the double-digit growth it pitched at last year's analyst day, the answer was a flat “Jamie, at a very high level, definitely not in 2027. I do foresee that we can actually get to double-digit growth.” — Balázs Fejes · 2026-11-05 The large deals ramp in the first half of 2027, with "greater contributions in Q2." That is a 2028 story being sold into a stock that has already rallied 94%. Worse for the margin half of the thesis:

due to the impact of the reduced R&D benefit and expected normalization in variable compensation expense, we are not expecting a further increase in gross margin in the coming year, and we may operate at modestly lower levels of gross margin in 2027.
Jason Peterson, Chief Financial Officer · 2026-11-05

The Poland R&D incentive benefit steps down from roughly $40 million in 2026 to about $20 million in 2027. Note the contrast with August, when the CFO was still promising “gross margin in excess of 32% for each of the quarters, Q3 and Q4” — Jason Peterson, Chief Financial Officer (CFO) · 2026-08-06 — delivered in Q3 at 32.8% non-GAAP — and the whole 2026 story was margin expansion while growing. The margin walk has now been formally capped for a year, at 30.4% GAAP in the last reported quarter against a 39.0% all-time peak.

The keyword that quietly fell out of the deck

Here is the tell. Through the May and August calls, management's central narrative device was the big deals pipeline — a phrase that ranked near the top of the company's own keyword set in the prior quarter. In this quarter's set it appears as one of the largest decliners in momentum, displaced by the softer vocabulary of new deals and strategic deals. The deals did not disappear — three closed — the framing did. And the CEO explained why, with unusual candor: “This is the third earnings call where we're addressing it. So it took some time for us to really close these, and probably this shows our relative inexperience in this space.” — Balázs Fejes · 2026-11-05 Compare with May, when the pitch was breathless: “So it's no longer just 3 or 4. We're actually talking about close to 10 opportunities at this point of time.” — Balazs Fejes, Co-CEO · 2026-05-07 Three closed out of a pipeline still described as "tens of deals," TCV from the high tens of millions to over $100 million, ramping into 2027 with startup costs in the first couple of quarters. The win ratio is unknown to management itself. That is a real option, not yet a forecast.

Frontier Labs and cyber: company-unique, tape-hostile

The genuinely new material is the partner-credential build and the Frontier AI Service launch — 25,000 certifications in six months across Anthropic, OpenAI, Google Gemini Enterprise and Microsoft; 8,300 Claude architects against a 10,000 target; 5,800 forward-deployed engineers against 10,000. There are over 200 tracked opportunities across the AI-native partner cohort. This is a differentiated asset that no peer on today's tape has articulated at this specificity. But the market is currently hostile to that adjacency: Frontier AI labs sits among the 30-day decliners in the global theme tape. EPAM is leaning into the exact cluster the tape is de-rating, which is either early or wrong. The counterweight is cyber resilience, where the story is concrete: AI-native remediation demand emerged only from late August/early September, with clients reallocating budgets to remediate "sometimes tens of thousands of vulnerabilities" — engineering-heavy work that plays to EPAM's core. Professional services firms as a group are working in the 90-day tape; that is the one alignment EPAM has going for it.

Capital return is the actual change

Strip out the AI narrative and the most concrete news is financial engineering. A $500 million accelerated share repurchase next week; a commitment to return at least 50% of free cash flow annually across 2027–2029; a $1.8 billion authorization through 2029, $1.5 billion of it incremental. Against an $8.5 billion market cap, that is a serious statement. It also partly explains the 94%: effective net cash of $769 million, down 25% year-over-year and 63% from the $2.1 billion peak is a shrinking cushion, and the latest reported quarter showed negative free cash flow after stock comp of $64 million. Repurchases had already slowed sharply, to $85 million in the latest quarter from a $324 million peak in Q1.

What the tape has already priced

At 7.2x operating income for a business with $769 million of net cash, 10.8% operating margins and a 13%-and-rising AI-native mix, the re-rating case is not absurd. But the company has now told us the two things the rally most needed to hear differently: no double-digit growth in 2027, and no further gross margin expansion in 2027. What is left is a ramp story with an unusually candid CEO, a shrinking non-AI core, and a buyback. February's 2027 guide is the event that decides whether the last 90 days were the beginning of a re-rating or a 94% head-fake.