Last Price
₹1,135
Jeff Research · Interactive Equity Deep Dive · Single-Name
A data-led deep dive: nine years of the client pyramid, the operational KPI machine, the FY26 margin walk, six years of vertical & geographic mix, a decade of capital return, and the FY26 annual-report strategy — with an interactive valuation sandbox you drive yourself. Charts are live: toggle series, hover for values.
Section 01
This is a bottom-up rebuild of Infosys from its own fact sheets and FY26 integrated annual report. Three facts dominate: (1) the client pyramid is still climbing — 41 $100M+ clients (24 in FY18), 88 $50M+ and 1,018 $1M+ — with concentration falling (top-10 at 20.2%); (2) the operational machine is improving — attrition 12.6% (vs 27.7% in FY22), revenue/employee at a record ~$61.4K, FCF conversion 113%, ROE 31.6%; (3) the order book is at a record — $14.9B TCV (+28% YoY, 55% net-new), 96 large deals. Against all of that, the stock trades at 15.5x trailing earnings — the bottom decile of its 10-year range — while management bought back ₹18,000 Cr at ₹1,800, a 59% premium to spot. The bear case (AI/GCC structural deflation) is real and unresolved, but the disclosures do not yet describe an impaired franchise. We rate it BUY / ACCUMULATE, conviction 3/5, with the structural question pinned to the Q1 FY27 print on 23 July.
Section 02
The single best refutation of "the franchise is breaking" is the client pyramid. Infosys discloses client counts by annual revenue tier every quarter. Every tier has expanded, and revenue concentration in the top clients has declined — the opposite of what you'd see if growth depended on a shrinking set of mega-accounts. Toggle the tiers; hover any bar for the value.
In FY26 Infosys added 2 $100M+, 3 $50M+, 19 $10M+ and 26 $1M+ clients while top-10 concentration fell 50bps to 20.2% and top-5 fell to 12.6%. Repeat business runs ~98%. The pyramid is widening at the base and rising at the apex simultaneously — the structural signature of an account-mining engine that is still compounding, not a melting ice cube. Marquee FY25–26 expansions: Citizens Financial Group (AI-led banking transformation), Lufthansa Systems (GCC build), Siemens (GenAI learning), Metro Bank (finance ops), Ralph Lauren (conversational-AI, cited at +12% revenue / +50% engagement).
Section 03
The cleanest way to see whether AI is eroding or augmenting Infosys is the operational KPI set over time. Toggle between the four levers. The story they tell together: headcount peaked in the FY22 over-hiring binge, then rationalised; attrition has collapsed from a 27.7% crisis to a healthy 12.6%; utilisation has rebuilt to ~84%; and revenue-per-employee has risen 29% off the FY22 trough — the first quantitative fingerprint of AI-assisted productivity.
Q4 FY26 headcount fell 8,440 sequentially even as full-year revenue rose to $20.16B — pushing revenue/employee to ~$61.4K, best-in-class (TCS ~$50K, HCL ~$48K) and +29% off the FY22 trough. Falling headcount + rising revenue/head is exactly what AI-as-augmentation looks like inside the delivery chain. The bear reading — that this is demand weakness, not productivity — is the open question; but management is explicitly building bench capacity (utilisation eased to 83% in Q4) "for future growth," not shedding to defend margins.
Section 04
Revenue compounded from $10.9B (FY18) to $20.16B (FY26) — an 8% USD CAGR — but the post-pandemic deceleration is stark: +19.7% CC in FY22 collapsed to +1.4% in FY24, recovering to +3.1% in FY26. Underneath the flattish top line, the book has quietly re-mixed: away from BFSI and North America, toward Manufacturing and Europe.
BFSI (Infosys's largest vertical) fell from 32.4% (FY21) to 28.0% (FY26) as banking tech-spend compressed, then began recovering — 18 of the top-20 FS clients now name Infosys their "AI partner of choice," and management guides BFSI to accelerate in FY27. Manufacturing nearly doubled to 15.9%, now the swing vertical (and the source of the 75–100bps FY27 drag from one European auto client winding down). Geographically, North America revenue is near-flat in absolute terms (~$11.3B) while Europe added ~$3.2B over five years (24%→32% of mix) on large cost-takeout deals — a useful diversification away from the most AI-disruption-exposed US discretionary spend.
Section 05
FY26 adjusted EBIT margin held flat at 21.0% (reported 20.3%, the 70bps gap a one-time ₹/$143M Labour-Codes provision). The flatness is deceptive: real tailwinds were earned and then deliberately reinvested. The Q4 FY26 sequential bridge is the only formal walk Infosys publishes — hover each step.
Across FY26, ~30bps from Project Maximus (value-based selling + lean/automation + portfolio optimisation) and ~40bps from INR depreciation were not dropped to the bottom line — they were reinvested into AI talent, sales & marketing (+40bps) and ecosystem partnerships. Third-party/subcontractor cost fell ~1pp of revenue (a structural margin tailwind as FY21–23 mega-deal pass-throughs roll off). The FY27 guide of 20–22% absorbs a 60–70bps acquisition-amortisation drag (Stratus/Optimum) plus an undecided wage hike, offset by continued Maximus savings. Translation: margins are being held flat by choice while the company funds the AI pivot — optionality, not weakness.
Section 06
Infosys's answer to the disruption thesis is to be the deployment layer, not the disrupted layer. The FY26 annual report frames a "USD 300–400B incremental AI-services TAM by 2030" across six areas: AI strategy & engineering, Data, Process, Agentic legacy modernisation, Physical AI, and AI Trust.
A "composable stack of AI agents, services and models." ~600 purpose-built agents (50+ for IT operations), 9 out-of-the-box enterprise integrations, and — critically — agents that run inside major coding-assistant and enterprise-workflow platforms. Ecosystem partners: OpenAI, Google Gemini, NVIDIA, Microsoft, AWS, Intel, Cursor, Cognition. AI services exceeded 5.5% of revenue by Q3 FY26 (Q4 "much more"), and management states AI deals "come with better pricing and margins."
The frontier-AI thesis — capability gains compounding, "unhobbling" to autonomous agents — is the exact mechanism behind the bear case. But it also implies trillions in enterprise AI deployment that someone must integrate, govern and run inside regulated BFSI/healthcare/government estates. Topaz Fabric is Infosys's bet to own that toll-booth. 84% of staff are "AI-aware," 275,000+ AI-trained. The investment question: does Infosys end up the typist or the plumber of the build-out? At 15.5x, the market has priced the first and given you the second for free.
Section 07
Infosys returns ~85% of FCF over rolling 5-year windows. Five buybacks since 2017 (₹57,760 Cr cumulative) plus a steadily rising dividend. The FY26 ₹18,000 Cr tender at ₹1,800 — the largest ever, 59% above today's price — is the loudest insider signal in the story.
DPS rose from a ₹17.5 trough (FY20) to ₹48 (FY26); the policy now commits to progressively increasing the regular dividend. FY26 returned >₹37,500 Cr (₹18,000 Cr buyback + ~₹19,500 Cr dividend) — ~7.9% of market cap, among the highest shareholder yields in Indian large-caps. A board that extinguishes 10 crore shares at ₹1,800 is not one that believes its franchise is structurally broken.
Section 08 · Peak → Trough → Now
Section 09 · Interactive
The whole debate reduces to three numbers. Set your view on the next-3-year constant-currency revenue CAGR, the terminal EBIT margin, and the exit P/E the market will pay. The 12-month target recalculates live off the FY26 base (revenue ₹1,78,650 Cr, ~408 Cr shares, net margin ≈ 0.81 × EBIT margin) against the ₹1,135 spot.
Presets: Bear 1.5% / 19.0% / 12x → ~₹820 · Base 4.5% / 20.5% / 19x → ~₹1,450 · Bull 7.0% / 21.5% / 24x → ~₹1,920. The ₹1,800 buyback price corresponds to roughly the bull settings — management's own implied multiple.
Section 10
(0.20 × ₹820) + (0.55 × ₹1,480) + (0.25 × ₹1,920) = ₹1,458 (+29%). Right-skewed: a 20% chance of −28% against an 80% chance of flat-to-+69%, with a 7.9% yield paying you to wait for the 23-July diagnostic.
Section 11
Section 12