08.2 · IT services, ER&D & software¶
Sector playbook. India's IT services industry — ~$250 bn of exports, five companies with 1.5 million employees — is the country's largest listed sector after financials and, in 2026, the one facing the sharpest structural question: what happens to a business model that bills for people when the people are being replaced by models? This page gives the metrics, the drivers and the AI debate as it stands in September 2026.
Time: ~70 min · Prerequisites: 05.1 Business models, 04.2 Margins
1. How the industry makes money¶
An IT services company sells engineers' time (or, increasingly, outcomes) to corporate clients, mostly in the US and Europe, delivering from India at a fraction of onshore cost. Revenue = billable people × utilisation × billing rate (time-and-materials) or fixed-price/managed-service contracts priced off the same cost base. Costs are ~55–65% people, plus subcontractors, travel, facilities and SG&A. The economics are a labour arbitrage plus process scale: EBIT margins of 15–25% for the large players, near-zero capital intensity, huge free cash flow, and returns on capital of 25–40%+ — which is why the sector trades at premium multiples despite single-digit growth.
Sub-sectors with different economics:
| Segment | What it sells | Growth driver | Margin | Examples (verify) |
|---|---|---|---|---|
| Large IT services | Application development/maintenance, infrastructure, BPO, consulting | Client tech budgets; large deals; vendor consolidation | 20–25% EBIT (TCS/Infosys), 15–18% (Wipro/TechM) | TCS, Infosys, HCLTech, Wipro, Tech Mahindra, LTIMindtree |
| Mid-caps | Same, focused verticals/geographies | Share gains from large players; niche skills | 14–20% | Persistent, Coforge, Mphasis, KPIT |
| ER&D / engineering services | Product engineering, embedded software, automotive, semiconductors | Outsourcing of R&D; EV/software-defined vehicles | 15–20% | L&T Technology, Tata Elxsi, KPIT, Cyient |
| Product / SaaS | Software licences and subscriptions | ARR growth, net retention | Negative to 25%+ | Zoho (unlisted), Freshworks (US-listed), Tanla, RateGain |
| GCCs (global capability centres) | Not listed — MNCs' captive centres in India | Compete with services firms for talent and work | — | ~1,700 GCCs employing ~1.9 mn (verify NASSCOM) |
2. KPIs¶
| KPI | Definition | Why it matters |
|---|---|---|
| Constant-currency (CC) revenue growth | Growth with FX held at prior-period rates | The clean volume-and-price signal; reported USD/INR growth is polluted by currency |
| Deal TCV / large-deal wins; net new % | Total contract value signed; share not from renewals | Leading indicator of revenue 2–4 quarters out; Infosys reported $4.8 bn large-deal TCV in a recent quarter with 57% net new (Infosys 6-K) |
| Utilisation (excl. trainees) | Billable hours ÷ available hours | 80–87% is the range; higher = margin, lower = bench for growth |
| Attrition (LTM) | Voluntary departures ÷ average headcount | 12–14% in 2026 (TCS 13.6%, Infosys ~13%); 20%+ in 2022 squeezed margins |
| Headcount and net adds | FY26 saw net reductions at the top five (TCS −23,000+ as it moved to an "AI-first" model; combined top-5 −6,981) (Entrepreneur India; TCS Q1 FY27 release) — revenue per employee is the new watch metric | |
| Pyramid | Share of junior vs senior staff | A fat base of freshers = lower cost per billable hour |
| Pricing / realisation | Revenue per person-month | Falling with AI-driven productivity commitments ("AI deflation": HCLTech flagged 3–5% revenue deflation in existing work — The Register) |
| Vertical/geography mix | BFSI, retail, manufacturing, healthcare, communications; US/Europe/RoW | BFSI ~30% for the majors; US 55–60% |
| Offshore/onsite mix | More offshore = higher margin | |
| DSO; unbilled revenue | Watch unbilled growing faster than revenue (fixed-price projects) | |
| Guidance | Revenue and margin bands (Infosys, HCL guide; TCS does not) | Say-do tracking |
3. What drives earnings and multiples¶
- Global tech spending (US and European corporate budgets; BFSI health; discretionary vs run-the-business): the demand cycle. 2023–26 was a low-growth stretch (Infosys ~3% CC growth in FY26; the sector's revenue growth was low-single-digit).
- Currency: a weaker rupee lifts INR revenue and margins (roughly 30–50 bps EBIT margin per 1% INR depreciation, net of hedges); the majors hedge 6–12 months out.
- Wage inflation and attrition: the 2022 spike cost 200–400 bps of margin; normalised since.
- Vendor consolidation and large deals: mega-deals ($1 bn+) shift share among the top players but often at lower margins and with transition costs.
- AI: the debate of 2025–26. Bear case: generative AI cuts the labour needed per unit of work by 30–50%, clients demand the savings, and revenue — billed on effort — deflates; headcount cuts confirm it. Bull case: AI creates new work (data, agents, modernisation), services firms capture it via "AI-first" delivery and outcome-based pricing, and margins hold as headcount falls faster than revenue. The stock market priced the bear case in 2026 (the IT index fell sharply in mid-2026 on AI-disruption fears — verify magnitude and dates). The analyst's task is to model revenue per employee, pricing deflation and the mix of new vs deflating work explicitly, and to treat "AI revenue" disclosures sceptically until they show up in growth.
- Multiples: the majors traded at 20–30x earnings through the 2010s–2020s on stability, cash returns (buybacks, 70–90% payouts) and RoE; de-rated in 2025–26 as growth stalled. Mid-caps command premiums when growing 15%+. Valuation is best done on P/E vs growth with a check on FCF yield (typically 3–5%) and on what the multiple implies for the AI transition.
4. Accounting quirks¶
Revenue recognition on fixed-price contracts (percentage-of-completion; unbilled revenue and contract assets); hedge accounting (cash-flow hedge reserve in OCI; realised hedge gains in revenue or other income depending on policy); SEZ tax holidays expiring (effective tax rates rising toward 25%); large cash piles and treasury income (10–15% of PBT — value separately); acquisitions and goodwill impairments (Wipro, TechM history); ESOP/RSU costs; share buybacks via tender offers (tax treatment changed in 2024 — 11.5).
5. Red flags¶
Unbilled revenue and DSO rising; margins held by cutting training/bench while attrition rises; "AI revenue" that is relabelled existing work; large-deal TCV rising while revenue growth falls (deals at lower prices/margins); guidance cuts mid-year; acquisitions to buy growth at high multiples; senior-management churn (CEO exits at several majors in 2023–25); client concentration (top-10 > 25%); dependence on a single vertical in stress (BFSI 2008, communications).
6. How to value¶
P/E on forward earnings vs growth and vs the company's own history; FCF yield and payout (most cash is returned); DCF with an explicit fade and an AI-deflation scenario (revenue −3 to −5% a year on the legacy base, partly offset by new work); EV/EBIT for cross-company comparison (cash piles distort P/E). Reverse-DCF the current price for implied growth (06.6) — in 2026 many majors' prices implied low-single-digit growth for a decade, which is the market's AI verdict.
7. Representative companies (for study; verify)¶
TCS, Infosys, HCLTech, Wipro, Tech Mahindra, LTIMindtree (large); Persistent, Coforge, Mphasis, KPIT, Tata Elxsi, L&T Technology Services (mid/ER&D); Tanla, Nazara, RateGain, Zaggle (products/platforms). Study TCS and Infosys annual reports for the disclosure standard, and KPIT for ER&D economics (08.6).
8. The 10-question sector checklist¶
- CC growth vs guidance and vs peers, and the mix of new deals vs renewals?
- What is revenue per employee doing, and is headcount falling faster or slower than revenue?
- Pricing: any disclosed deflation on existing contracts; share of outcome-based pricing?
- Utilisation, attrition and pyramid — margin levers used or exhausted?
- Vertical/geography exposure to the weakest end-markets?
- Unbilled revenue and DSO trend?
- Cash returns: payout, buyback timing and price?
- Acquisitions: price, integration, goodwill?
- Management stability and succession?
- What growth does the price imply, and does the AI scenario range bracket it?
← Previous: 08.1 Banks & lending · Module index · Next: 08.3 Consumer →