Business

Claude View

Know the Business

Genpact is a $5.1B people-and-process factory — 146,500 employees running the back-office plumbing of roughly a quarter of the Fortune Global 500. It was born inside GE, spun out in 2005, and still competes primarily on labor arbitrage dressed up as "process intelligence." The real investment question is not whether Genpact can win more business-process deals — it is whether the seats, which are the P&L, survive agentic AI. Management is loudly pivoting to AI ($1.2B "Advanced Technology Solutions" bucket, growing double digits), but 76% of revenue still comes from Core Business Services where AI is a customer-demanded productivity giveback, not a growth catalyst.

1. How This Business Actually Works

Genpact sells outcomes, but gets paid by the seat. Roughly 75% of its cost base is people — mostly India-based — and revenue lives and dies by billable headcount × utilization × bill rate. Contracts are long-dated (MSAs of 3–7 years, SOWs of 2–5), which creates recurring-revenue optics, but renewal is re-priced every few years and the client's in-house GCC option puts a ceiling on pricing power.

Three things actually drive incremental profit:

1. Pyramid steepness. Early in a contract, the supervisor-to-agent ratio is high and margins are thin. As processes stabilize, junior seats replace senior seats and margin expands — this is why management calls out "mix" every quarter.
2. Productivity give-back asymmetry. Most MSAs force Genpact to share productivity gains after year 1 or 2. If Genpact automates 20% of seats, clients often capture the majority. AI makes this lever both larger and scarier.
3. Revenue currency vs. cost currency. 75% of revenue is in USD; most costs are in INR, PHP, RON, MXN. A strong dollar boosts margin by 50–100 bps; a weak dollar compresses it. Genpact hedges, but imperfectly.

The three industry segments earn nearly identical adjusted margins (17–19%) — this is not an accident. Pricing is set by the labor pyramid and productivity commitments, not by vertical domain expertise. The "moat" Genpact claims in each industry is real on the sales side (it explains why it wins the deal) but does not translate into differentiated per-seat economics.

2. The Playing Field

Genpact sits in the awkward middle of IT services: too small to compete with Accenture's consulting-led model, too big and broad to be a pure-play BPM specialist like WNS. The peer table below is what actually matters.

What the peer set reveals:

• Accenture is the ceiling, not the competitor. ACN prints ~38% ROIC on a near-identical operating margin to Genpact (14.8% vs 14.8%) because its capital intensity is lower and its consulting deals rotate inventory faster. Genpact cannot get there — its model owns more delivery infrastructure and carries more receivables.
• EXLS is what "good" looks like for Genpact's scale tier. Similar model, half the employees, better growth (+13.6% CC in 2025 vs Genpact's +6.4%), higher ROIC, richer multiple. EXLS has leaned harder into analytics/AI and kept headcount growth below revenue growth (9.8% vs 13.6%) — textbook operating leverage.
• WNS is the fallen comp. Once the premium pure-play BPM, now growing ~2% with declining earnings. Capgemini's pending acquisition of WNS in 2025 signals that sub-scale standalone BPM is structurally challenged.
• Cognizant/Infosys are the India-heritage IT giants whose multiples pull Genpact's down. The market treats Genpact more like cheap IT services (11.7x P/E) than like a consulting/AI franchise.

Genpact trades closer to CTSH on multiples than to EXLS — the market is pricing it as a commoditized offshore labor arbitrage story, not an AI transformation play, despite management's messaging.

3. Is This Business Cyclical?

Yes, but the cycle hits differently than most IT services. Genpact's cycle is project pauses + pricing give-backs, not revenue cliffs. Managed-services contracts rarely terminate; they get paused, re-scoped downward, or forced into productivity concessions.

The FY2023 slowdown (+1% constant currency) is the most informative datapoint. Rising rates plus the 2023 regional-banking stress caused financial-services clients to defer discretionary work (Data-Tech-AI projects dropped sharply), while Digital Operations held steady. That is the cycle signature: advisory and project work is the first thing clients cut; managed seats are sticky because unwinding them is expensive. This is why Genpact's revenue is more resilient than, say, Accenture's consulting, but also why upside capture is muted when the cycle turns.

Secondary cycle exposures worth naming:

• FX. A 5% INR/USD move equals roughly 50–80 bps of operating margin. The INR sits at record lows vs USD in 2025, which has quietly helped margins.
• Wage inflation. India wage hikes run 8–10% annually. Genpact passes roughly half via pyramid re-design and half via re-pricing — meaning margins slowly compress unless automation absorbs the rest.
• Client GCC (in-house) decisions. When India-heritage enterprises decide to insource (large banks and retailers expanded GCCs 2023–2025), Genpact loses RFPs before they start. This is structural, not cyclical, but the decision cadence is lumpy.

4. The Metrics That Actually Matter

Skip gross margin. Skip headline revenue growth. These four metrics explain value creation in this business.

Genpact sits in the bottom third of its peer group on revenue per employee — by design, because it runs more managed-service seats than project work. The bull case requires this number to climb toward EXLS (~$46K) over 3–5 years. The bear case is that it stays flat: AI takes out seats and rates, net-neutral.

New bookings is the single most important leading indicator nobody puts enough weight on. Bookings were $5.7B in FY24 and $5.5B in FY25 — a 3% decline while revenue grew 6.6%. Either management is being more conservative in what it counts, or growth decelerates in FY26. Watch the Q1–Q2 FY26 prints closely.

5. What I'd Tell a Young Analyst

Five things to actually watch, in priority order:

1. New bookings trajectory. Not the absolute number — the direction. Two consecutive down years means structural, not lumpy.
2. Advanced Technology Solutions growth rate and mix. Management drew the line at 23.7% for FY25. If this is not 28%+ by FY27, the AI pivot is failing.
3. Revenue per employee. The cleanest test of whether AI is a net positive or a net giveback. If headcount grows faster than revenue for two years, the margin story breaks.
4. Top-20 client concentration (currently 34.4%). Stable or declining is fine. Rising concentration means new-logo wins are drying up.
5. Adjusted operating margin vs. management guidance. Genpact has a decade-long track record of hitting guidance within 50 bps. A miss here is a thesis-breaking event.

What the market is likely wrong about: the bear case on AI disintermediation is overstated in the near term (2026–2027) because Genpact's seats are buried inside regulated, domain-heavy workflows that LLMs cannot touch without human oversight. But the bull case on AI re-rating is also overstated — agentic solutions still decelerate as volumes compound, because clients capture most of the productivity gains on renewal. This is a mid-single-digit growth, mid-teens margin business trading at a mid-teens multiple. That is roughly fair. Do not own it for surprise; own it for dividend (~1.9%), buybacks (~3% of float annually), and optionality on the ATS mix shift. If you cannot articulate which KPI would make you sell, do not buy.