Persistent Systems — a cash machine in the AI crosshairs
Persistent Systems Ltd
Snapshot
Persistent Systems builds software for other companies — it is a “digital engineering” shop, which means its 25,000-odd engineers design, build, modernise and now AI-enable the products and systems its clients sell. Market cap ₹76,177 Cr, share ₹4,829 (52-week range ₹6,599 / ₹4,449 — so it sits near the bottom, off about 27% from its high). It trades at a P/E of 39.4 (you pay ₹39 for every ₹1 of yearly profit) and 9.7× book value, earns a return on equity of 27.3% (profit per ₹100 of owners’ money) and a return on capital of 34.4%. In one phrase: a Great, asset-light cash machine — caught in the one storm Buffett always feared, “lots of technology.”
As of 2026-06-20, from screener snapshot.
The verdict in one box
| Lens | Result |
|---|---|
| QGLP score | 19.5 / 25 (Quality 9/12 · Growth 6/6 · Longevity 4/5 · Price 0.5/2) |
| Buffett rubric | 7 / 10 PASS |
| Business bucket | Great economics (asset-light, fountain of free cash) — with a predictability asterisk |
| Wealth-creator type | Enduring (so far) · Consistent (profit never fell >10% in 12 years) |
| Economic Profit | +₹1,199 Cr (RoE 27.3% − CoE 12% on ₹7,838 Cr net worth) — strongly creating value |
| Margin-of-safety price band | ₹3,000–₹3,600 (P/E 25–30× on FY26 EPS); strict PEG ≤ 1× wants ~₹2,200. CMP ₹4,829 is demanding |
A Great business and a genuine wealth creator — currently priced rich versus the quality, by a market that is half-afraid of artificial intelligence and half in love with it.
In plain English
Imagine a workshop that doesn’t sell its own product. Instead, the best software companies in the world — and increasingly banks and hospitals — hand it their hardest building problems and pay it to solve them. That is Persistent. It started in Pune in 1990 as an outsourced research-and-development lab for American software firms, and it has stayed truer to that “engineering-first” DNA than almost any Indian IT company. It does not dig mines or build factories. It rents brains and sells their output. So it needs very little capital, throws off enormous cash, carries almost no debt, and earns a return on capital — 34% — that most factory owners would sell a kidney for.
The numbers tell a beautiful story. Sales have gone from ₹2,312 crore in 2016 to ₹14,748 crore in 2026 — more than six-fold. Profit has climbed every single year for over a decade, from ₹277 crore to ₹1,865 crore, and never once fell by more than a rounding error. Twenty-four straight quarters of growth. A return on equity parked stubbornly in the mid-20s. This is what a wealth creator looks like on paper: it takes the rupees it keeps and reliably turns each one into several more. By the strict Buffett “one-dollar test” — does every rupee retained create at least a rupee of market value? — Persistent passes with room to spare.
Now the storm. Persistent’s whole industry was built on a simple trade: Indian engineers write code more cheaply than Western ones. Artificial intelligence — the coding assistants that now write a great deal of software themselves — attacks the very heart of that trade. If a machine can do in one hour what used to take a person ten, and you bill by the hour, your bill shrinks. The giants are already feeling it: Tata Consultancy actually shrank last year and let go of 12,000 people; the cautionary tale, EPAM, has lost roughly half its value. This is not a theory. It is happening, and the sector is openly calling it “AI deflation.”
Here is the twist, and it is the whole report. Persistent is, so far, on the winning side of this split. While the giants shrink, it grew 17% last year. It doesn’t defend billable hours; it sells its own AI tools (a platform called SASVA) that sit on top of Claude, Copilot and Codex — so it makes money helping clients adopt the very machines that threaten everyone else. The bear’s strongest point is that Persistent’s most exposed customers — software companies — are also its specialty, and its own CEO admits it may have to “cannibalise our own business.” The bull’s strongest point is that it is already doing exactly that and still growing faster than almost anyone. The tension is not about whether this is a good business. It plainly is. The tension is that the market knows it, and at 39× earnings you are paying a wonderful price for a wonderful business in a moment when the ground under the whole industry is shifting.
Sitting down with the management
If you spent an afternoon with the people who run Persistent, you would come away mostly impressed, and a little uneasy about one thing.
You would start with Dr. Anand Deshpande, who founded the company in 1990 with a few thousand dollars after a spell at Hewlett-Packard’s labs in Palo Alto, with a Ph.D. in computer science behind him. He is the rare Indian promoter who reads like an institution-builder rather than an empire-builder — a trustee of research endowments, a backer of engineering education, a man whose conduct over 35 years carries no scandal. He owns about 30% of the company, which is enough to care deeply but not enough to rule it like a fiefdom. And he did the hardest thing a founder can do: in 2020 he handed the steering wheel to a professional, Sandeep Kalra, and stepped back to chairman without blocking him or hovering. That clean transition is itself a mark of character. Under Kalra, revenue has roughly tripled to over $1.6 billion and the stock compounded furiously. The bench beneath them — a new CFO, vertical heads, a CTO dedicated entirely to “engineering hyper-productivity” — is real and professional, not a family roster.
How have they spent your money? Well. The acquisitions have been small, sensible tuck-ins — a cloud capability here, a Google-partnership shop there, a communications-automation business — never a bet-the-company “toad” that needed a magic kiss. (Note for the record: the famous ~$1.8bn IBM software carve-out went to HCL, not Persistent — a common confusion worth correcting.) Through all that buying and building, return on equity stayed in the mid-20s, which is the proof that the rupees were deployed well rather than just deployed. Dividends have risen every year, the payout sits around a healthy 34–37%, there is zero promoter pledging, no auditor drama, no related-party games, and reported profit is reliably backed by hard operating cash. On integrity — the gate that voids everything else if it fails — Persistent is clean.
The unease is about pay and dilution. Kalra’s FY25 remuneration came in around ₹148 crore, up ~93%, the bulk of it stock options he exercised — enough to top India Inc.’s CEO-pay charts. You can defend it as performance-aligned (he did create the value), but it is eye-watering in absolute terms, and it is the engine behind a quiet, steady creep in the share count and the gentle drift down in promoter holding (31% to 30.3% over a couple of years). The company returns cash through dividends, not buybacks, so it never mops up that dilution. None of this is a red flag in the forensic sense — it is a question of how generously the pie gets shared with the people running the kitchen versus the people who own it. The second, softer concern is succession below Kalra: there is no visible heir-apparent, so the key-man risk has moved from the founder to the CEO rather than vanished.
Would Buffett and Agrawal shake hands on this management? Yes — with a raised eyebrow at the pay packet. What would change their mind: the promoter drift accelerating into real selling, or the ESOP largesse widening while returns soften.
What’s on the horizon (live-issues tracker)
1 — The AI reckoning (the crux). Status: 🟡 winning so far, but the contest is live. Treated in full below — this is the whole ballgame.
2 — The “$2 billion by FY27” ambition. Status: 🟢 on track but loud. Management has planted a flag: $2bn in annual revenue by FY27, up from ~$1.65bn in FY26. They have earned the right to talk this way — 24 straight quarters of growth is not luck. But the talk is promotional, not conservative; this is a team that leads with streak-metrics and aspiration. The market has punished them before for a single soft outlook (down 20% in four days, back in an earlier cycle), so the bar is high. Watch the quarterly constant-currency growth rate: high-teens keeps the story; a slip toward single digits (as Q3 FY26 hinted at ~8%) is what rattled the stock down to its lows.
3 — Margins grinding upward. Status: 🟢 on track. Operating margin has climbed from ~17% to 19%, and the AI pitch is explicitly that doing “more work with fewer people and more technology” is margin-accretive, not dilutive. So far the numbers agree — EBIT margin expanded ~90bps in FY26. This is the optimistic reading of AI made concrete: if Persistent keeps the productivity gain rather than handing it all to clients, margins rise even as billed hours fall. The bear says competition forces them to give it away. The next four quarters of margin prints settle that argument.
The crux, interrogated
This investment works if and only if Persistent stays on the accelerate side of AI — monetising adoption faster than AI deflates its billed hours.
The mechanism, in plain English. The old IT-services deal was labour arbitrage: rent a cheaper engineer, bill per hour. AI coding assistants break that in two places at once. They cut the hours a job needs (fewer hours, smaller bill), and they hand clients a club to demand the savings back at renewal (lower price per hour). Think of a translation agency the year machine-translation got good: the work didn’t vanish, but the easy pages collapsed in price, and the agency only survived by moving up to the hard, sensitive, “you’d better get this exactly right” pages — and by selling the machine-translation service itself. That is precisely the move Persistent is making. The analogy holds because, like translation, the commodity layer (boilerplate code) is what AI eats first, while the judgement layer (architecture, integration, regulated systems, “make this actually work in a bank”) is what it protects.
Who can take the share — named.
| Threat | Posture / proof point | Read for Persistent |
|---|---|---|
| Tier-1 Indian IT (TCS, Infosys) | Already in deflation — TCS shrank ~0.5%, cut ~12,000 jobs; Infosys guides just 1.5–3.5% for FY27 | Persistent is taking share from them — growing 17% while they degrow |
| Accenture | Big GenAI bookings (~$2bn/quarter) but only 2–5% growth, then stopped disclosing the metric | Proves AI bookings are real but don’t yet offset core deflation at scale |
| AI tool vendors (OpenAI, Anthropic, Google) | OpenAI’s “Deployment Company” (May 2026, $4bn) pitches CIOs directly — IT shares fell 3–5% on the news | The genuine new disintermediation risk — but Persistent partners with them (SASVA runs on Claude/Codex) |
| Mid-cap peers (Coforge, EPAM, Globant) | Coforge grew 29% (an “AI-native” model that works); EPAM down ~49% (engineering-hours compression gone wrong) | The fork in the road: Persistent sits much closer to Coforge than to EPAM |
The precedent. European power exchanges, when their market was “coupled,” survived by commoditising the auction and differentiating on services — incumbents lived, but margins on the easy part fell. The closer precedent here is machine translation and, before it, the move from custom-coded websites to templates: in every case the commodity layer collapsed in price and the survivors climbed to higher-value work. Incumbents who climbed (the specialists) kept their economics; incumbents who clung to the commodity layer (the body-shops) did not. Persistent’s heritage is the specialist’s, which is the single most important fact in its favour.
The follow-on questions, answered. Is the damage to volume or to price? Both — but concentrated in commoditised, maintenance and BPO-style work, which is the giants’ problem more than Persistent’s. Which of Persistent’s segments is hit first? Its software/hi-tech clients (most AI-ready) — awkwardly, its specialty. Which is protected? Banking and healthcare, where regulation slows AI adoption; tellingly, management is steering FY27 growth toward exactly these protected verticals. Has anyone actually switched away yet? No visible client loss — the evidence so far is share gain, not loss. Who’s on the other side of the bet? A genuine technology shift, funded AI-tool vendors, and a market that has already re-rated the stock down ~27%.
Honest verdict: right side of the technology — but the price already assumes it. The mechanism, the precedent and the live numbers all point the same way: Persistent is a specialist monetising AI adoption, not a body-shop defending billed hours, and its 17% growth amid sector deflation is hard proof, not a story. This is not a “too hard” case on the business. It is a “too hard” case only on whether 39× earnings leaves any room if growth fades from high-teens toward the EPAM end of the fork.
The watch-list — what tells you the thesis is working or breaking:
- Constant-currency revenue growth holding mid-teens+ (a slide toward single digits, as Q3 FY26 flirted with, is the warning).
- EBIT margin continuing to grind up past ~16% — proof AI is accretive, not given away.
- TCV / order book staying above ~1.4× revenue (FY26 was $2.4bn, ~1.45×).
- Revenue per employee rising, not falling — the single cleanest read on whether AI helps or hurts the billing model.
- Software/hi-tech segment growth not rolling over while BFSI/healthcare carry the load.
- Any sign of an actual client switching to an AI-vendor “deployment” team instead of an integrator.
QGLP scorecard (the Motilal Oswal lens) — the receipts
| # | Question | Score | Evidence |
|---|---|---|---|
| 1 | Large opportunity? | 1 | Global digital-engineering + AI-modernisation TAM, multi-decade runway |
| 2 | Industry structured favourably? | 0.5 | Competitive, but Persistent’s product-engineering niche keeps OPM stable 14–21%, now 19% |
| 3 | Defensible moat? | 0.5 | Switching costs + IP; RoCE >18% for 9 of 10 yrs proves durable returns — but a services moat, not a pricing-power franchise, and AI contests it |
| 4 | RoE & RoCE >15% consistently? | 1 | RoE 27.3%, RoCE 34.4%; both >15% for a decade |
| 5 | Asset-light? | 1 | IT services; FCF ₹1,572 Cr vs OCF ₹1,767 Cr (89% conversion), RoCE 34% |
| 6 | Favourable terms of trade? | 0 | Banks its customers (debtor days 53); working-capital days rose 48→73 — structural for IT, but not negative-WC |
| 7 | Integrity? | 1 | Zero pledge, clean audits, profit backed by cash, no controversy |
| 8 | Proven execution? | 1 | 24 straight quarters of growth; revenue $481m → $1.65bn |
| 9 | Growth mindset & vision? | 1 | Heavy AI/IP reinvestment (120+ patents), $2bn FY27 ambition |
| 10 | Superior capital allocation? | 1 | Small accretive tuck-in M&A, RoE sustained through growth, no diworsification |
| 11 | Succession plan? | 0.5 | Clean founder→Kalra handover + pro bench, but no visible heir below the CEO |
| 12 | Minority interests protected? | 0.5 | Rising dividends, but outsized ESOP pay dilutes and no buybacks mop it up |
| 13 | Structural tailwind? | 1 | Digital/AI spend growing well above nominal GDP |
| 14 | Volume-led growth? | 1 | Historically deal/client-led, not price — though AI repricing is the forward risk |
| 15 | Operating leverage? | 1 | OPM 17% → 19% as revenue scaled |
| 16 | Manageable leverage? | 1 | Near debt-free; borrowings ₹477 Cr vs ₹7,838 Cr net worth |
| 17 | Market-share gain potential? | 1 | Growing 17% while Tier-1s degrow — taking share now |
| 18 | Earnings growth >15% CAGR? | 1 | 5Y PAT CAGR ~33%, 10Y ~21%; forward high-teens likely (AI is the caveat) |
| 19 | Relevant for 10–15 yrs (low disruption)? | 0.5 | The weak point — IT services is technology-fragile; AI repricing is real |
| 20 | Can extend its CAP? | 0.5 | Moat durability under AI pressure is genuinely uncertain |
| 21 | Can sustain its GAP? | 1 | Long AI-modernisation runway, large under-penetrated TAM |
| 22 | Diversification headroom? | 1 | Expanding verticals (BFSI, healthcare), geographies, IP platforms |
| 23 | Adaptive, resilient culture? | 1 | Engineering DNA; pivoting to AI tooling faster than most peers |
| 24 | Valuation reasonable (PEG)? | 0.5 | PEG ~1.2 on trailing 33% growth, but ~2.2 on a realistic ~18% forward |
| 25 | Margin of safety (PEG<1 / payback<1)? | 0 | 5-yr payback ~4.8×; no margin of safety at CMP |
| Total | 19.5/25 | Quality 9 · Growth 6 · Longevity 4 · Price 0.5 |
The pattern: Quality and Growth are the strength — this is a textbook compounder on the business. The two real gaps are Longevity (the AI-disruption cloud over a tech-fragile model) and Price (the only pillar that scores near zero). Strip out price and it’s a 19/23 business; the price is what stands between this and a clean buy-zone.
Buffett lens (the Berkshire-letters read)
| # | Test | Result | Evidence / Buffett line |
|---|---|---|---|
| 1 | Good boat? (business > management) | PASS | Great economics — asset-light, 34% RoCE, fountain of FCF. “A good managerial record is a function of what boat you get into.” |
| 2 | Moat + franchise + pricing power | PARTIAL | Switching costs + IP are real, RoE > CoE for 10 yrs — but limited pricing power, and AI is testing it |
| 3 | See’s test (high returns, little capital) | PASS | Capex << operating cash; FCF/PAT ~84%; grows without swallowing capital |
| 4 | One-dollar test (capital allocation) | PASS | Every retained rupee compounded book and market value; RoE held mid-20s through 6× growth |
| 5 | Owner-oriented, candid management | PARTIAL | Clean, delivers, zero pledge — but promotional tone and chart-topping ESOP pay |
| 6 | Integrity / forensic (no “credit P&L, debit B/S”) | PASS | OCF backs PAT; no balance-sheet bloat; clean auditor history |
| 7 | Circle of competence / predictability | PARTIAL | The weakest test — “if there’s lots of technology, we won’t understand it.” 10-yr predictability is genuinely clouded by AI |
| 8 | Mr. Market — gift or trap now? | PARTIAL | Off ~27% from high, P/E compressed 50→39 on AI fear — cheaper, but still not a fearful giveaway |
| 9 | Patience / compounding runway | PASS | Long AI-modernisation runway at ~27% RoE — a decade of compounding if the model holds |
| 10 | The honest red flag | (see below) | The bear case, in full |
Count: 5 PASS + 4 PARTIAL = 7/10. A real, high-quality business with two real gaps — exactly where it should land for a Great franchise facing a genuine disruption.
The See’s test, in numbers. See’s Candies needed only $32m of reinvestment over 35 years to throw off $1.35bn. Persistent is cut from the same cloth: in FY26 it generated ₹1,767 Cr of operating cash and ₹1,572 Cr of free cash after all capex — about 84% of profit converts straight to spendable cash. It funds its entire growth from its own pocket, carries no real debt, and still has cash left to pay a rising dividend. This is the asset-light dream Buffett spent his life chasing.
The one-dollar test, in numbers. Over the last decade Persistent retained the bulk of its earnings and grew net worth from ~₹1,400 Cr to ₹7,838 Cr — and over the same stretch the market value went from a few thousand crore to ₹76,000 Cr. Each rupee kept inside the business has created far more than a rupee of value outside it. By Buffett’s own arithmetic, retention here has been not just justified but richly rewarded. The only smudge is the share count creeping up via ESOPs — value created, but a slightly thinner slice reaching the owner.
The framework metrics
- Economic Profit = ₹7,838 Cr × (27.3% − 12%) = +₹1,199 Cr. Genuinely creating value far above the cost of owners’ money — top-quintile territory.
- Terms of Trade = debtors fund customers (53 debtor days, working-capital days 73) → >100%, unfavourable — but structural for IT services, not a quality flaw.
- 5-yr Payback = ₹76,177 Cr ÷ ~₹15,750 Cr projected cumulative 5-yr PAT (18% CAGR assumed) = ~4.8×. Far from the <1× multibagger signal — the price is full.
- PEG = 39.4 ÷ 18 (forward) = ~2.2 (or ~1.2 on the realised 33% 5-yr growth). Demanding on any forward-looking view.
- RoE − CoE spread = +15.3%; RoE > 15% in all of the last 10 years.
- Consistent / Volatile = Consistent — profit fell >10% zero times in 12 years; terminal profit 6.4× the initial. Value it on P/E, not P/B.
Peer comparison
| Company | Mcap (₹ Cr) | CMP (₹) | P/E | P/B | RoE | RoCE | OPM | FY26 Sales (₹ Cr) |
|---|---|---|---|---|---|---|---|---|
| Persistent | 76,177 | 4,829 | 39.4 | 9.7 | 27.3% | 34.4% | 19% | 14,748 |
| Coforge | 62,922 | 1,463 | 38.3 | 5.2 | 20.6% | 23.5% | 18% | 16,403 |
| Mphasis | 43,280 | 2,268 | 22.9 | 4.0 | 18.5% | 22.8% | 19% | 15,880 |
(LTIMindtree, a large-cap comparable, is excluded — the data fetch resolved to the wrong entity; treat it as a separate large-cap class anyway.)
The relative read sharpens the absolute one. Persistent earns the best returns in the set by a clear margin (RoE 27% / RoCE 34% vs low-20s for both peers) and has the most consistent growth record — which is exactly why it commands the top P/E. So its premium isn’t unjustified; you are paying up for genuinely superior economics. But it is not cheap even within its own asset class: it sits at the expensive end alongside the faster-growing Coforge, while Mphasis at 22.9× is the value option for an investor who will accept lower returns and slower growth. A sector-allocator’s verdict and a patient value-investor’s verdict diverge here: relatively, Persistent is “best-in-class, fairly-to-fully priced”; absolutely, it fails the QGLP price bar. Both are true — they answer different questions.
Latest quarter & what’s happening now
Q4 FY26, reported 21 April 2026. Revenue $1,654m for FY26, +17.4% YoY; Q4 up 16.2%; 24th consecutive quarter of growth; EBIT margin 15.6%, up ~90bps. Order book healthy — FY26 total contract value $2.4bn (~1.45× revenue). BFSI was the standout vertical (+28% to a ~$600m run-rate); healthcare steadier (+10%).
Two concall takeaways worth holding onto. First, the AI posture is concrete, not hand-waving: the SASVA platform now runs as a layer on top of Claude, GitHub Copilot and OpenAI Codex — Persistent monetises the tools rather than being replaced by them [HARD]. Second, CEO Sandeep Kalra’s candid money-quote: tech-segment revenue “may be compressed… we will also go and cannibalise our own business,” but “there’s enough and more business” [HARD]. That honesty is admirable and unsettling in equal measure — he is telling you the risk is real and betting expansion outruns it, with no FY27 guidance to lean on.
Where the two lenses agree — and disagree
They agree on the business: both call it Great economics, both pass capital allocation and integrity emphatically, both flag price as the binding constraint. QGLP 19.5/25 and Buffett 7/10 are consistent — strong quality, gaps in longevity and price.
The disagreement is the signal, and it lives in one test. QGLP’s checklist happily scores Growth a perfect 6/6 — the trailing numbers are spectacular. But Buffett’s predictability test (7) flags PARTIAL/FAIL, because the letters weight “can I see this business in ten years?” far more heavily than a checklist does, and the honest answer for any IT-services firm in 2026 is less clearly than two years ago. A checklist sees a 33% grower; Buffett sees a technology-fragile model whose pricing engine is under live attack. Trust the Buffett flag here — it is not predicting doom, it is pricing the uncertainty the growth columns hide. That single divergence is why this is a 19.5, not a 23, and why the margin of safety matters more than usual.
Margin-of-safety price band
Not a recommendation — the framework’s arithmetic. On FY26 EPS of ₹118:
- Strict QGLP (PEG ≤ 1× on ~18% forward growth): ~₹2,100–₹2,400. This is the deep-value, “everything has to go right and I still want a discount” price. You rarely get a 27%-RoE compounder this cheap outside a crash.
- Fair-for-the-quality (P/E 25–30×): ₹3,000–₹3,600. A 17–20% grower earning 27% RoE with a decade-long consistency record genuinely deserves a premium multiple; this band is where quality-versus-price gets interesting.
- CMP ₹4,829 (P/E 39×): demanding. It prices in continued high-teens growth and that AI proves a tailwind rather than a headwind. Buffett’s Mr. Market is neither fearful nor greedy here — he’s cautiously optimistic, and charging you for the optimism.
Plainly: a wonderful business at an unwonderful-to-fair-minus price. The 27% drop from the high has taken it from euphoric to merely expensive, not to cheap.
Conviction texture
The bull, in its strongest form: This is the best-returning mid-cap in Indian IT, run by clean, capable people, growing 17% while the giants shrink — because it is on the right side of AI, selling adoption rather than defending billed hours. The order book covers 1.45× of revenue, margins are rising, and the runway in AI-led modernisation is a decade long. Buy a Great compounder when a sector-wide fear knocks 27% off it, and time does the rest.
The bear, in its strongest form (Buffett’s test-10 red flag): The single strongest reason this is not a wealth creator from here is that its growth engine and its biggest risk are the same thing. Its specialty — software and hi-tech clients — is precisely the segment AI deflates first, and the CEO has openly said he’ll cannibalise it. The whole industry has confirmed AI deflation is real (TCS shrinking, EPAM halved). At 39× earnings, the price has already paid for the bull case; if constant-currency growth fades from high-teens toward single digits, the de-rating and the earnings miss compound against you at once. You are buying certainty of quality at a price that assumes certainty of outcome — and the outcome is the one genuinely uncertain thing.
What the numbers actually support: Great business, Enduring and Consistent on every historical measure, strongly value-creating (EP +₹1,199 Cr) — but priced for the disruption to break its way. The business question is largely answered; the price question is wide open. No buy/sell/hold here — the deliverable is: own the quality at ₹3,000–3,600, demand a real margin of safety nearer ₹2,200, and at ₹4,829 understand you are paying for a future that is more uncertain than the past it’s extrapolating from.
Sources
- Screener.in: PERSISTENT consolidated (snapshot fetched 2026-06-20), peers Coforge, Mphasis.
- Persistent Q4 FY26 concall, 21 Apr 2026 · FY26 press release.
- AI-disruption context: The Register — AI deflation in Indian IT, 28 Apr 2026 · ICICI Direct IT sector 2026 · Coforge FY26 results · EPAM decline.
- Management: Business Today — Sandeep Kalra deep-dive, Apr 2025 · Business Standard — CEO pay.
- Assumptions: Cost of equity 12%; forward PAT CAGR 18% for payback/PEG; EP on FY26 net worth ₹7,838 Cr and RoE 27.3%. No buy/sell recommendation.