Five businesses. One thin margin.

The question. Redington is India's and MEA's largest IT distributor: it moves phones, PCs, enterprise hardware, software and cloud services through channels that pay it a thin margin on enormous revenue. The underwriting question this research asks is not whether revenue grows — it does — but which mechanism pays in each segment: hardware volume times a spread on a partly fixed cost base, a financing commission on working capital, a software lifecycle annuity still paying for its own capability build, or large data-centre deals that add revenue faster than margin. Group growth alone does not identify which of these is working, and the five businesses do not answer as one.

No investment view is formed. What follows is a segment-and-regime account of the evidence, its limits stated throughout, and an exploratory valuation model whose assumptions remain unreviewed by a human and whose statement lines remain relayed from research notes rather than read from primary filings. The latest evidence synthesis underlying this article is dated 8–10 September 2026; it is not a new financial-results update.

What business are we underwriting?

Redington's economic unit is the segment and the transaction. Endpoint (ESG) has price, component and replacement cycles; Mobility (MSG) turns inventory quickly at low margin; Technology Solutions (TSG) books lumpy enterprise and data-centre deals; Software Solutions (SSG) carries recurring and lifecycle economics; Arena, the listed Turkish company (49.4%-owned and consolidated), has its own credit, currency and strategic risks.

The unresolved contest: operating leverage or a thin financing commission?

Two economic models of this company compete. The bull model — stated in this research in 2026 and elsewhere — is volume × spread on a partly fixed cost base, assisted by holding gains during product-price inflation and by software/cloud mix. The 2021–2023 practitioner account instead describes a distributor paid a thin commission to finance and absorb OEM-channel risk: working capital scales with sales, and the discounts given on the last units of inventory decide profitability. An archived practitioner forecast that peak margins and trough working capital would reverse, together with a historical duopoly claim, remains a hypothesis awaiting financial or channel evidence — as does a privately supplied recommendation to treat the company as an AI/data-center proxy. These are source views, not this research's.

Both models look good during product-price inflation, and FY26–FY27 was such a period. The useful tests are EBITDA margin against revenue growth outside inflation quarters, cash conversion, provisions and financing cost. The earnings-call record strengthens mix, deal-timing and business-model explanations; it does not establish stable whole-company operating leverage. Nor does the practitioner model become fact by repetition.

Competitive position must show up in retained economics

Management's software-moat claim is lifecycle execution plus local presence, not a platform alone. It acknowledged that OEMs engage large enterprises directly but argued they increasingly rely on distributors for fulfilment, lifecycle management, services and solution breadth. It said non-renewal most often follows weak customer usage, and admitted some distributors occasionally surrender the available 5.5%–6% software margin to win share.

The top six vendors contribute "about 70% plus" of SSG revenue. Customer and transaction counts were described as roughly 70:30 toward smaller enterprises, while revenue mix may reverse because enterprise deals are much larger. These stay as quoted bounds rather than being converted into false precise points. An AWS executive, preserved as an external claimant at a company-hosted June 2026 investor event, described that relationship as scaling from one customer and one partner in India in 2015 to about 7,000 customers and more than 850 transacting partners in 2025; a described five-year commitment "in the billions of dollars" was not converted into a company KPI.

Vendor concentration became measurable only when the sell-side arrived. A note dated 9 September 2026 put Apple at about 31% of revenue, unsourced on its page; the initiation of 10 September 2026 carried a series sourced to the company through the broker: Apple 30%, 31%, 30%, 32% and 31% of revenue from FY23 to 1QFY27; the top five vendors (Apple, HP, Lenovo, Dell, Samsung) about 60%; vendors outside the top five rising from 37% to 43%. No filing has been read for any of this. The open questions are vendor dependence, discounting to win share, renewal cohorts, channel financing and actual services capture: measure the available margin and the capital required before inferring bargaining power from scale.

The reported trajectory across five quarters

These are transcript-labelled figures, with reporting and perimeter differences retained. The FY26 backfill improves the time series; it does not turn every quarter into a comparable accounting observation.

FY26 revenue was stated at ₹119,347 crore, up 20%; PAT growth excluding exceptional items was 17% at a 1.3% margin. Q4's ₹467 crore PAT excluding exceptionals and ₹391 crore including the Arena impairment are different scopes, not a contradiction. The Q1 FY27 transcript's ₹34,966 crore "revenue" is close to official total income rather than official revenue from operations; the wording is preserved rather than silently normalized. An earlier transcript-implied EBITDA-margin comparison (2.02% versus 1.55%) was corrected: Q1 FY27 figures imply about 2.15%, and a like-for-like comparator still needs rebuilding from primary statements.

Segment growth rotates; the software perimeter changes

This comparison is useful but not mechanically like-for-like: Q2 FY26 was the first quarter under a new software perimeter (software, cloud, security and professional-services components moved across the former TSG and ESG definitions), and management's remembered historical software aggregations vary by scope and currency. The pattern is a rotation in growth, not a single stable engine: Endpoint accelerated from 3% to 35%; Technology fell to −7% before rebounding to 34% and 50%; Software stayed high but volatile; Mobility remained the largest, steadier low-margin volume pool. Mix changes both the quality and the percentage margin of group growth.

Margin and operating-leverage tests

Q1 FY26 gross margin fell to 5.1% from 5.7%, attributed to higher Mobility mix, lower-margin large TSG deals, market pressure, PayNet and Arena. Management reported a 2.09% operating-profit margin that quarter with regular opex growth of 6%, rising to 19% including a receivable provision; Q2 opex grew 9%; by FY26 management said opex-to-revenue had fallen 17 basis points, but Q4 carried war-related freight and insurance, FX, one-off receivable provisions and continued capability spending.

Management's FY27 guide is roughly 2.2%–2.4% EBITDA margin excluding Arena, sensitive to large-deal composition; earlier it cited a 2.3%–2.5% operating-profit objective and PAT above 1.3% in an ex-Arena discussion. These are scorecard ranges, not established steady-state margins. The central underwriting tension is whether gross-profit growth and mix improvement can outrun software capability opex, Arena drag, large-deal dilution and financing costs.

Segment economics, and the next observation each needs

Endpoint: separate price inflation from unit growth

Endpoint's 3% → 11% → 21% → 28% → 35% growth progression looks increasingly price-, supply- and deal-assisted rather than a clean volume acceleration. Management first expected a Windows-refresh cycle, then said the Windows 10-to-11 migration had not produced a large demand uptick. Q3 brought component shortages and higher prices without an immediate margin benefit; Q4 described strong demand alongside smartphone, PC and GPU supply gaps. By Q1 FY27, management said PC units grew only single digits while prices rose 25%–50% over two or three quarters and channel inventory stood at two to three months. Management expected further price increases to pass through, while warning that higher PC and smartphone prices could dampen demand.

The sell-side relayed the external side of the same picture in the 10 September 2026 initiation: industry trackers' memory costs up 130% by end-2026 and PC prices up at least 17%; commercial laptops already 10–15% dearer; enterprises replacing 20–30% of planned refresh volumes; a front-loaded +32% PC market in 1QCY26; and a tracker estimate of −5.3% shipments for 2026 with recovery only in 2028. The initiation's own Endpoint build is ASP-led in FY27E (about 19%) followed by 8% and 8%, and it names memory-driven supply shortfalls as a risk to its own call. That is the normalization test above, stated by a bull. The test is not whether Endpoint revenue grows; it is where normalization first appears — price, units, inventory, demand, gross margin, working capital or provisions.

Mobility: fast turnover changes the group's capital and margin mix

Mobility remained the largest revenue pool — 35% share in Q2 and Q3 FY26, 33% in Q4, 35% in Q1 FY27, with growth of 18%, 15%, 19% and 21%. It is described as low-stock and high-velocity, helping group working-capital days while diluting percentage margin. The initiation relays: Apple 28% and Samsung 22% of India smartphone value in 2025; the $400–800-plus tier rising from 13% to 20% of volume between 3QCY24 and 1QCY26; India smartphone sales −10% year-on-year in 2QCY26 with the sub-₹15,000 tier −45%; Apple's 20–40% price rises on existing iPhones announced on 9 September 2026; and Android rising to 10.2% of Redington's revenue from 1.3% five years earlier through a 5,000-store direct-to-retail model. The initiation builds Mobility at 19% a year to FY29E and treats the memory shortage as a Mobility tailwind because Apple's localized supply is unconstrained — the same price inflation this research flags as a demand risk elsewhere. Open items: distinguish vendor and geography mix from genuine unit demand; watch smartphone supply and price-related demand loss. A better group working-capital ratio driven by this mix is not proof that every segment improved.

Technology Solutions: low-margin large deals can still earn acceptable returns

The 21% → 9% → −7% → 34% → 50% sequence is timing-sensitive. Management repeatedly linked it to selective large enterprise and data-centre deals, some sliding from Q3 into Q4. In Q1 FY27 it described nearly ₹1,000 crore of large-deal revenue, mainly India data-centre activity, with a backlog multiple times the quarter and no taper expected for four to six quarters — while stating that large-deal profitability is below normal and dilutes weighted margin.

The stated discipline is deal-level ROCE: the company structures back-to-back terms or factoring and says it will not breach return thresholds except for rare strategic reasons. Revenue growth can therefore coexist with percentage-margin dilution and higher finance needs. The model must evaluate incremental gross profit, working-capital duration, credit support and financing cost per deal, not reject or celebrate low-margin volume mechanically. The initiation builds the segment at 31% a year (+45% in FY27E, then 29% and 21%) and relays an external tracker's India data-centre capacity path of 2.2 GW in 2025 to 12 GW by 2030, naming larger-than-expected margin compression from these deals as a risk to its call.

Software Solutions: a lifecycle model with a measurable capability bill

Software's reported share moved from about 15% in Q1 FY26 to 16%, 18% and 17% through FY26, growing 24%, 48%, 40% and 31%; Q1 FY27 then grew 52% at a 17% share. Management put FY26 software revenue at US$2.2 billion and explicitly distinguished 29% growth on the presentation's dollar basis from 37% in rupee reporting, where currency depreciation helped translation. Software was 17% of group revenue versus about 12% two years earlier and was said to have doubled over three years. The FY26 mix was roughly 40% software, 33% cloud and 27% security, with services separately described as below 1%; rounding and boundary differences prevent exact reconciliation.

Management's three-year case is a scenario, not an adopted forecast: approximately US$5 billion of revenue, 5.5%–6% gross margin with professional-services upside, PAT margin near 2%, and software around 25% of Redington. Conditional illustrations of 15%–20% hardware growth and 30%–35% software growth remain assumptions. Q4 management described current software gross margin around 5.5% and reiterated the 5.5%–6% intent. The operating model is a shift from transactions to lifecycle economics — own the platform, attach services, manage onboarding through adoption and renewal — and the investments behind it (a cloud platform, presales, certifications, an academy, partner capability) can suppress near-term operating leverage before higher-margin mix appears. The initiation's exhibit 31 shows software gross-profit margin on net revenue of 6.4% in FY25 falling to 5.7% in FY26 — a decline the June event did not present; the level may be a basis effect, but the direction is a new test item.

Revenue recognition complicates the top line. For renewals, subscriptions and some vendor-led services, Redington may recognize only the spread rather than the invoice value. Management called this roughly 5% of software and also said invoiced revenue was about US$150 million above the reported US$2.2 billion; those approximations imply different percentages and remain unreconciled. Recurring revenue was stated at 69% in FY23, 72% in FY25 and 74% in FY26, with an 85% three-year target; current renewal was around 85%. On the renewal target itself, one executive said 90% and another later said above 95% — unresolved. "Banked revenue to the next year" was given as 63.5%, rising to 72% in three years; the arithmetic does not yet close (74% × 85% is 62.9%, not 63.5%, and 85% × 90% is 76.5%, not 72%), so definitions need clarification. Recurring revenue and ARR are used around the same 74% figure; they are not assumed identical here.

Low inventory does not mean low total working capital. Software inventory is very low, but receivable days are longer, leaving total working-capital deployment broadly similar to hardware; higher profitability, rather than radically lower capital, supports its stronger claimed returns. An estimated 80%–90% of capability investment is expensed through opex. Management targets a three-to-five-year payback but admitted historical investment was not tracked meticulously and displayed figures were indicative.

Professional services remains small relative to the ambition. Management described it as very small and above 20% gross margin, yet short of solution architects, practice leads and domain people; maturity was self-rated 2–3 on a stated 1-to-10 scale against a desired 8–9, with an explicit warning that the company was not ready. The intended model is productized AI, cloud and security services delivered with partners, not bespoke IT work. A stated 7×–10x growth aspiration has no clean baseline or endpoint.

Arena: continuation, remediation and exit are separate branches

Arena is improving operationally in places but remains unresolved. Management described an approximately US$8 million receivable provision in Q1 FY26 against roughly US$20 million of delayed collections, with low credit-insurance coverage in some markets. Redington-attributable Arena loss was ₹37 crore in Q2, ₹22 crore in Q3 and Q4, then ₹31 crore in Q1 FY27; Q4 also carried a separate ₹75.2 crore impairment impact on group PAT after minority interest. The sequence 37 → 22 → 22 → 31 means the apparent improvement was not yet durable.

Management is exiting the Connect/Vodafone business and selected Turkish-lira exposure, tightening credit and working capital, and considering structural or strategic outcomes while retaining profitable software/cloud activity. It expected a smaller continuing revenue base and a "new normal" for factoring, but made no firm dated commitment; collections under a concordat may take 18–24 months; no new capital infusion is planned. The initiation fills in the path: Paynet sold in FY25 with US$74 million of net proceeds used to repay Arena debt; the Vodafone distribution exited in FY26; a US$17.2 million trade-name impairment; Arena debt US$160 million (FY24) → US$90 million (1QFY27); Arena working capital US$179 million → US$129 million by 3QFY26; the 100%-share loss US$9.5 million (1QFY26) → US$4.9 million (3QFY26) → US$6.8 million, about ₹640 million, in 1QFY27 — which at Redington's 49.4% is about ₹31.6 crore and matches the attributable loss above; Arena's market value US$45 million; and a company expectation of break-even in FY28, relayed and not dated in any transcript on file. FY28 is now a stated date to score. The valuation must carry explicit continuation, remediation and exit branches with debt, impairment and minority-interest treatment; no unannounced turnaround or disposal is capitalized.

Cash conversion and West Asia can absorb the growth

Working-capital days moved 37 → 31 → 28 → 30 → 32 across the five quarters, and reported ROCE ran at 22%. Management described current ROCE as close to 20%, an above-18% FY27 guardrail and a 16% internal floor. This is strong aggregate evidence, but management also says 35–40 days can be a normal band, and Mobility's fast-turn profile improves the mix. Aggregate days cannot replace receivable, inventory, payable, vendor-credit and factoring schedules.

The financing points so far, on management's corrected figures: Q1 FY27 consolidated interest plus factoring fell to ₹122 crore from ₹148 crore (Arena ₹69 crore from ₹90 crore; consolidated factoring corrected to ₹31 crore from ₹68 crore). Q2 FY26 total interest plus factoring rose to ₹161 crore from ₹117 crore, with Arena corrected to ₹101 crore and the rest of the business ₹60 crore. Later calls expect benefit from reducing Turkish-lira exposure, the Connect disposal and lower factoring, but stop short of a committed run rate. A primary debt, cash and finance-cost bridge remains necessary. On definitions, the two research notes supply a bridge this research has not reproduced from primary statements: management's 22% is a net RoCE (cash netted out of capital employed); the notes' 16.3% and 16.5% are on gross capital employed; a post-tax 12.3% also exists. Three labelled bases, not two incompatible figures — and a calculation from primary statements is still missing. The initiation's full-year FY26 net RoCE is 19.8%, so the 22% quarterly figures are annualised quarters; its FY26 working-capital days of 34 (receivable 66, inventory 26, payable 58) are an annual ratio against management's 30 at quarter-end.

The FY26 calls also connect the West Asia conflict to war-risk insurance withdrawal, freight premiums, warehouse mitigation, inventory gaps and delayed receivables. Q1 FY27 management said existing projects continued while new ones were postponed; about 85% of pre-war sea shipments shifted to air; delays could be around 30 days; and transport-plus-insurance cost was about US$3 million, much but not all recovered. Q4 demand visibility was strong in India and Africa, conditional in the Middle East, softer in Saudi Arabia. These are not generic narrative risks: they affect availability, pass-through timing, gross margin, provisions, working capital and demand.

What remains unresolved before the valuation can be trusted

Definitions, corrections and track record:

• Software perimeter and growth: Q2 FY26 starts the new form; historical aggregates and currency bases differ. Do not score a clean series without restatement.
• Renewal target: 90% versus above 95% within the same June 2026 event remains unresolved.
• Cloud platform release: June says version 2.0 was released early in the year and names later modules; July describes migration from 1.0 to 2.0. Release versus migration scope is unclear.
• AI agent catalogue: June says more than about 450 curated agents; July says over 250. Definition, deployment readiness and date unclear.
• Professional-services growth: June's awkward "100% rate" wording was never quantified; July says both "triple digit" and "almost doubled". Audio or later disclosure is needed.
• Recurring terminology: recurring revenue and ARR are used around the same 74% figure; definitions may differ.
• Arena and finance-cost corrections: corrected values are retained where calls corrected themselves; irreconcilable Q3 units and arithmetic are excluded.
• Profitability scope: reported, ex-Arena and ex-exceptional margins must never be collapsed.

The data gaps have a concrete shape:

Aggregate margins or a single inflation quarter cannot settle these. The older "no primary filing read" position has been overtaken by transcript capture and a narrow official-results reconciliation; the remaining gap is a complete comparable financial history, not an absence of all primary material.

Scorecard for future results: (1) group and ex-Arena EBITDA/PAT margins versus the cited ranges; (2) Endpoint price, unit, inventory and gross-margin normalization; (3) Technology deal timing, incremental gross profit, working capital and ROCE; (4) software growth, the 5.5%–6% gross-margin intent, recurring share, renewal, services mix and capability opex; (5) working-capital components, finance cost and maintained ROCE; (6) Arena loss, collections, factoring, continuing-revenue scale and strategic outcome; (7) West Asia supply, freight, insurance and receivable normalization.

Two sell-side reference points, labelled

Research-firm names are withheld on this site; both documents are held privately and the labels below are consistent across the model page. Their historical figures are relays of company data and their forecasts are their own; none of it is independently verified here, and no filing has been read for the historical years.

Research firm A — note dated 9 September 2026 (first broker view; page 1 only). BUY, 12-month target ₹500 at 16× FY28E EPS of 30.9 against a then-market price of ₹378. Its path is the operating-leverage case made explicit: net sales compounding 22.5% a year from FY26 to FY28E (₹119,162 → 178,922 crore), EBITDA margin rising from 1.9% to 2.1%, adjusted PAT from ₹1,565 crore to ₹2,423 crore, RoCE from 16.5% to 22.5% on its own basis. On that page it does not address Arena, working-capital components, financing cost or units against prices; its premiumisation argument treats the very ASP inflation management has warned will dampen demand as a positive; and its FY27E EBITDA margin of 2.0% sits below management's 2.2%–2.4% ex-Arena guide. The page carries an arithmetic error (a printed 6.0% two-year net-profit CAGR that does not follow from its own row) and an unsourced Apple share of about 31% of revenue. Score at FY27 results: net sales ₹149,837 crore; EBITDA ₹3,036 crore; adjusted PAT ₹2,168 crore; EPS 27.7.

Research firm B — initiation dated 10 September 2026 (46 pages, read in full). BUY, 12-month target ₹470 on a DCF (13% cost of equity, 6% terminal growth) against the same ₹378 price; 12× FY28E EPS at the price, 15× at the target. Revenue ₹119,162 crore (FY26) → ₹151,431 crore (FY27E) → ₹181,550 crore (FY28E) → ₹214,639 crore (FY29E); EBITDA margin 1.87% → 2.03% → 2.15% → 2.18%; PAT ₹1,284 crore (FY26, after a ₹152 crore exceptional charge) → ₹1,939 → ₹2,375 → ₹2,777 crore; diluted EPS 19.1, 26, 31, 35; working-capital days 34 → 37; net RoCE 19.8% → 23.6%; net debt/equity 0.15 → 0.43. Segment CAGRs to FY29E: software 34%, technology 31%, mobility 19%, endpoint 11%.

Where they differ: the initiation consolidates Arena and carries the exceptional charge (FY27E PAT ₹1,939 crore against the note's adjusted ₹2,168 crore); it treats memory inflation as an Endpoint risk rather than only a premiumisation tailwind; it discloses its cash-flow build, in which the growth is debt-funded; and it supplies the RoCE definitions. Where both agree: about 22% revenue growth through FY28E, a 2.0–2.1% EBITDA margin by FY28E, and Apple at about 31% of revenue. Score the initiation at FY27 results: revenue ₹151,431 crore; EBITDA ₹3,079 crore (2.03%); PAT ₹1,939 crore; EPS 26; RoCE 19.7% gross and 22.4% net; working-capital days 35; operating cash flow about −₹200 crore. Its own inconsistencies (net RoCE FY29E 23.3% in prose against 23.6% in its table; 30 against 31 bps of margin improvement) are recorded as printed.

FY26 profit has three scopes — management's 17% ex-exceptional growth at a 1.3% margin; the note's adjusted ₹1,565 crore; the initiation's reported ₹1,284 crore after the ₹152 crore exceptional charge (its FY25 ₹569 crore sits after a ₹626 crore charge, against the note's adjusted ₹1,340 crore). RoCE has three labelled bases — net, gross and post-tax — as set out under cash conversion. Neither discrepancy is a contradiction; each is a scope. Both notes' price references date from 9–10 September 2026 and are not live quotes; no external target has been reproduced by this research's own model.

Accounting quality, contingent liabilities and governance: a first external review, relayed

The initiation's forensic screen rates Redington D4 ("Zone of Safety"), with red flags only on cash conversion (pre-tax operating cash flow / EBITDA 33% in FY26; free cash flow / revenue nil) and amber flags on contingent liabilities and group structure. Contingent liabilities are ₹5.2 billion, 5% of net worth (2.1% over FY17–22), mainly disputed tax: almost all the direct-tax exposure is a permanent-establishment dispute at the Singapore distribution entity, now disclosed and under the Mutual Agreement Procedure; separately, a customer claim of about ₹1.5 billion over an AWS data-deletion dispute carries an interim stay. The group has 46 subsidiaries, six holding 57% of assets and 29 with negligible or negative profit; assets and revenue outside the main auditor's scope fell from 54–58% (FY23) to 11–14% (FY26), with Deloitte in its fourth year.

Governance, as relayed: no promoter since July 2019; Synnex Technology holds 24.1% without promoter status; institutional holding about 78%; a non-executive, non-independent chair and 44% independent directors. In September 2023 the board made V.S. Hariharan Group CEO within ten days of his resigning as an independent director, over proxy-advisor objections based on the regulator's one-year cooling-off period, after Rajiv Srivastava's abrupt exit the month before — the second MD change in three years; dissent on the February 2025 MD resolution was about 3%. Ten-year cash conversion: 43% operating cash flow / EBITDA and 54% free cash flow / PAT.

Every figure in this section is the initiation's reading of filings this research has not opened, including its two hyperlinked sources, which are unread. Consequences for valuation: the tax and customer contingencies belong in the downside branch with Arena; the cash-conversion flags are the same fact as the negative operating cash flow in the initiation's own forecast; the 2023 governance episode is history to weigh against the current board, not a live event.

The exploratory DCF: what exists, and what it cannot support yet

A working whole-company model now exists on the interactive valuation page — rebuilt data-first on 10 September 2026, with seventy-three exposed assumptions, each carrying its basis (actual, guidance, estimate, extrapolation or assumption), its range and its source note: eighteen are extrapolations from the company's own statements, eighteen are the initiation's stated inputs held for the comparison scenario, twelve are figures the sources state directly, and twenty-five are the model author's, each with its range. Primary financial reconciliation and human assumption review remain incomplete, so nothing below is an endorsed value: the model's statement lines are the initiation's relay of company figures, and no person has reviewed the assumptions. Under the desk defaults it values the equity at ₹351.24 per share — a 13% discount rate (the initiation's cost of equity applied to firm cash flows), 6% terminal growth, an FY40 horizon, end-year discounting and an FCFF-growth terminal, on the Arena continuation branch. Enterprise value ₹29,246 crore, of which the terminal value is 64%; net debt ₹1,516 crore (borrowings less cash, FY26) and minority interest ₹263 crore at book.

What ties, and what does not. The computed FY25 and FY26 columns tie to the relayed printed figures within ₹2 million on EBITDA, PBT and PAT. FY26 EPS ties within ₹0.03 once the minority's 50.6% share of Arena's loss is removed from consolidated PAT. FY25 EPS does not tie (about ₹3.0 off) because the minority's share of the Paynet gain is not the Arena share. Other comprehensive income — ₹474 crore of the FY26 equity movement — and a ₹333 crore gap between the cash-flow and balance-sheet working-capital changes are unexplained. The implied core debt rate in FY26 is 12.8% after Arena's interest, above the initiation's forward rate of about 10.5%, and the Arena interest split rests on three quarters. These are the model's honest seams, printed as diagnostics rather than smoothed away.

The shape of the desk case, FY27–FY29: revenue ₹151,980 → 171,881 → 196,437 crore; EBITDA margin 2.04% → 2.15% → 2.18%, inside management's ex-Arena guide once Arena's drag is added back; PAT ₹1,983 → 2,337 → 2,741 crore against the initiation's ₹1,939 → 2,375 → 2,777 crore; EPS 26.2, 29.9, 35.1; net RoCE 23.1% → 23.4% → 24.0% against management's 18–20% target and 16% floor. Borrowings rise every year to FY33 under the defaults, because working capital grows faster than retained profit funds it — the debt-funded-growth signature of the initiation's own cash-flow statement, now visible as a line in this model. Working capital is held at FY26's component days (receivable 66.1, inventory 26.3, payable 58.1 on their own bases; 35.8 days of revenue) with the geography mix shaving about 0.29 days per unit of SISA share — SISA is the India-led geography cut in the sources, 97% India; the eleven-year range on file runs from 14 to 50 days. Opex is not a free assumption: the two statement years fix the variable ratio at 2.20% of revenue and the fixed base at ₹1,004 crore, leaving only fixed-base inflation to choose. Gross margin moves with the geography mix (a 1.3-point SISA-to-rest-of-world differential), a 0.85-point software premium and 2 points of dilution per unit of large-deal share.

Reverse questions at the dated ₹378 price (10 September 2026; not a live quote): the price implies terminal growth of 6.66%, or 1.063 times the FY27–FY29 segment-growth defaults, or 34.1 working-capital days. Timing and terminal conventions move the answer without touching a business driver: mid-year discounting gives ₹375; an FY36 horizon gives ₹251; a value-driver terminal at a 20% return on new capital gives ₹390. On the Arena branches: remediation ₹343.41; exit ₹349.35 (FY28 exit year) — a few rupees, because the loss and the minority largely offset.

The comparison scenario. Run under the initiation's own stated inputs and printed net debt, the same engine reproduces its FY40 revenue and EBIT within 2% but gives ₹363 against its ₹470 target with end-year discounting and an FCFF-growth terminal; ₹389 with mid-year timing; ₹415 with a value-driver terminal at a 25% return on new capital; ₹445 with both. The unexplained residual — ₹25 under the most favourable standard reading, ₹107 under the plain one — sits in a terminal construction the initiation does not disclose. It is reported, not tuned away; no target has been chased and no probabilities have been invented. None of this is a view: the defaults are the extrapolations and assumptions in the model's own sheets, and the calibration is a test of the initiation's arithmetic, not of the business.

What the model cannot support yet: the model statement inputs have not been reconciled to a complete primary financial filing; segment-level gross margins and days are not disclosed anywhere, so economics run through geography differentials whose levels miss the group by 12–17 bps; unit volumes, prices and headcount do not exist in any source; and the desk case has no test beyond the tie-outs and the calibration above. Existing model plans already own primary-statement capture, convention review and later-evidence tests; publication of this model adopts nothing.

How to interpret the valuation

The market price on 10 September 2026 — ₹378 — sat above the model's ₹351 desk default while both sell-side houses had targets of ₹470 and ₹500. The gap between the model's default and the initiation's own target is almost entirely terminal construction, not business judgment: timing and terminal-method choices span ₹351 to ₹441 under the desk inputs; applying the comparison inputs as well gives ₹445. The three levers that move the model most are the discount rate and terminal growth, the working-capital days (and who funds them — the debt line rises every year to FY33 at the defaults), and the FY27–FY29 segment growth arrays. Every figure on the valuation page is conditional on stated assumptions that no human has reviewed; the statement inputs are relayed; and neither the model's output nor either research firm's target is an endorsed or adopted view. What would change that fastest, in order: the FY27 primary statements (closing the EPS, OCI and cash-flow seams); a complete debt, cash and finance-cost bridge; and the Arena FY28 break-even score.

Evidence and source quality

The research draws on three layers, in descending reliability for this company.

1. Edited exchange-filing transcripts of six management events — the earnings calls of 30 July 2025, 6 November 2025, 5 February 2026, 14 May 2026 and 30 July 2026, plus the 17 June 2026 software investor event (management-curated and unusually promotional; its cloud-provider segment is still a company-hosted presentation). Each transcript says the exchange-hosted audio prevails where the edited text differs. Management statements are preserved with attribution and are not independently verified facts; in-call corrections supersede original values.
2. Two sell-side documents: page 1 of the 9 September 2026 note (an image, transcribed) and the complete 46-page initiation of 10 September 2026. Their historical figures are labelled company-sourced by the brokers themselves and remain unreconciled to filings; the initiation's own hyperlinked sources were not opened. Broker forecasts are recorded as claims to score, not endorsements.
3. An archived practitioner account (2021–2023) and privately supplied observations, kept as hypotheses and source views respectively — including the inventory-tail mechanism, the May 2022 reversal forecast and the AI/data-center proxy recommendation.

The selection is deliberately partial: reported results, forecasts, operating mechanics, causal explanations, explicit risks and falsifiable strategy are admitted; ceremony, repeated promotion, product catalogues, unsupported third-party statistics, analyst premises not adopted, and arithmetically inconsistent figures are not. Where a figure could not be published faithfully without an identifying detail, it is held rather than paraphrased into pseudo-fact.

About this edition. Published 26 September 2026 from research synthesised on 8–10 September 2026; prices quoted date from 9–10 September 2026 and are not live. Research-firm and individual names beyond Redington's own management are withheld under this site's editorial policy and replaced with consistent neutral labels; the underlying private edition preserves the exact sources. Figures attributed to transcripts or research notes are relayed claims, not audited facts read from primary filings — that gap is the largest single limit on everything above, and it is why the model on the valuation page is presented as exploratory.