G › analyze
G · Analyze
2026-08-11 · $34.09 · Mkt cap $5.73B · EV $6.63B · Technology / IT & Business Process Services
Segment rename in effect: "Data-Tech-AI" is now Advanced Technology Solutions (ATS); "Digital Operations" is now Core Business Services (CBS).
Verdict up front
ACCUMULATE — conviction 6.5 / 10. The best risk/reward of the seven names analysed today.
Genpact is priced for the terminal decline of the seat-based BPO model at ~7.9× forward earnings and 8.0× free cash flow, and it is the only one of the three services businesses in this batch producing hard evidence that it is monetising AI rather than being eaten by it: Q2 revenue +7.1% on average headcount −2.6%. That divergence has held for two quarters — short, but it is the real thing, and Infosys cannot show it. The bear case is contractual, not merely narrative: some client agreements require passing productivity savings back. The bull case has $700M+ of free cash flow, 12.6% FCF yield, nine straight dividend raises, a 5% buyback and Graham valuing it at exactly the market price.
The single ugliest fact: every insider is a seller and there have been zero open-market buys.
1. Fundamentals
| $M | FY22 | FY23 | FY24 | FY25 |
|---|---|---|---|---|
| Revenue | 4,370 | 4,480 | 4,770 | 5,080 |
| Gross profit | 1,540 | 1,570 | 1,690 | 1,830 |
| Operating income | 556 | 627 | 702 | 760 |
| Net income | 353 | 631 | 514 | 552 |
| EPS (dil) | 1.88 | 3.41 | 2.85 | 3.13 |
| OCF | 444 | 491 | 615 | 813 |
| Capex | (54) | (59) | (85) | (89) |
| FCF | 389 | 432 | 530 | 724 |
| Buybacks | (214) | (226) | (253) | (283) |
| Dividends | (92) | (100) | (108) | (118) |
| Diluted shares (M) | 188.1 | 185.1 | 180.4 | 176.7 |
CAGRs (3yr): Revenue +5.1% · Net income +16.1% · FCF +23.0% · OCF +22.4% · Diluted shares −2.1%/yr
This is a genuinely good three years, and unlike Akamai's or Dropbox's it is not an accounting artefact. Operating income grew 36.7% on 16.2% revenue growth — real operating leverage. Gross margin went 35.2% → 36.0% and Q2 2026 printed 36.5% (+60bps). Free cash flow grew 86% while revenue grew 16%, driven by working-capital discipline and OCF conversion improving from 81% to 107% of operating income.
⚠️ One distortion to note. FY23 net income of $631M is above the FY23 and FY24 trend and
sits oddly against operating income of $627M — a net margin approaching the operating margin.
That trips Knowledge/Playbook/pattern-net-margin-above-operating-margin-is-a-tripwire. The
likely cause is a tax item (a valuation-allowance release or a one-off benefit). It inflates
the FY22→FY23 growth and makes FY24's "decline" look worse than it was. Use operating income
and FCF, not net income, for the trend on this name. Both are clean and both are rising.
Per-share (FY25): Revenue/share $28.76 · FCF/share $4.25 → 8.0× P/FCF, FCF yield 12.6% · BVPS $15.50 → P/B 2.20×
Balance sheet. Assets $5.84B, debt $1.76B, cash $854M → net debt ~$688M, which is 0.95× a single year's free cash flow. Debt/assets 30.2%; current ratio 2.00; interest expense $73.5M against $760M of operating income = 10.3× coverage. $350M of 1.75% notes matured April 2026; $350M of 4.95% notes issued November 2025 — the refinancing is done and the rate step-up is small in absolute terms. Goodwill $1.78B (30% of assets) is the legacy of old acquisitions and has not been written down. This is a comfortably financed business.
Capital allocation — disciplined and shareholder-friendly. - Buybacks: FY25 6.13M shares at an average $46.16 = $282.9M. $364.1M remains of a $2.75B authorization. Critically: Jan 1 – Feb 25, 2026 bought 1.20M shares at $38.15 — they are buying materially cheaper as the stock falls, which is the behaviour you want and the opposite of what most boards do. - Dividend: FY25 $0.68 (+11%), FY26 $0.75 (+10%). Nine consecutive raises since 2017, ~13%/yr over nine years. - Combined: ~$400M returned against $724M of FCF = 55% payout, leaving real capacity. - M&A is small and infrequent ($80M in FY25).
Total shareholder yield: 2.2% dividend + ~5% buyback = ~7.2%, before growth.
Scorecard: cash generation A · operating leverage A− · balance sheet A− · capital allocation A · revenue growth C · earnings quality B− (the FY23 tax item). Flagged gaps: FY21 statements absent (3yr CAGRs only, not the framework's 5–8yr); no disclosed contract renewal rate; the ">70% annuitized ATS" claim is management assertion, unverifiable from filings.
2. Q2 2026 — and the two-quarter inflection
Reported 2026-08-06.
| Q2 2026 | |
|---|---|
| Revenue | $1.343B, +7.1% YoY |
| ATS | $363M, +24.1% (from +17.3% a year ago) — 27% of revenue |
| Core Business Services | $980M, +1.9% (from +3.8%) — 73% of revenue |
| Gross margin | 36.5% (+60bps) |
| Adjusted operating margin | 17.4% (+10bps) |
| Adjusted diluted EPS | $1.00, +13.6% |
Bookings: largest-ever quarterly bookings; six $50M+ deals in the quarter, 12 year to date versus 6 in H1 2025. Agentic TCV exceeded $200M in 2025 and is targeted above $1B in 2026 (~5×), with more than half of cumulative agentic ACV coming from new clients. CFO Weiner conceded that very little current ATS revenue is agentic — this is a bookings story, not yet a revenue story.
FY26 guidance: revenue ≥7% reaffirmed, absorbing a ~2pt headwind from deliberately exiting non-strategic work; ATS raised to ≥25% (from mid/high-teens); gross margin 36.5%; adjusted operating margin ~17.7% (+25bps); adjusted EPS ≥12%. Q3 $1.369–1.382B, EPS $1.04–1.05.
The inflection — the one datapoint that separates Genpact from Infosys.
Q2 2026: average headcount fell 2.6% YoY while revenue grew 7.1%. Management's "productivity index" reads 1.04.
Through FY2025 the trend ran the other way: revenue per employee declined from ~$36.3k (2023) to ~$35.2k (2024) to ~$34.8k (2025), with headcount growing 7.8% against 6.6% revenue growth to 146,500. Genpact has now held revenue while shrinking the seat count for exactly two quarters. That is the entire bull case in one line — and it is two quarters old, which is why this is ACCUMULATE and not BUY.
Compare directly: Infosys in the same period grew 2.4% with headcount flat and cut guidance. Genpact grew 7.1% with headcount down and raised its ATS target.
The market's objection is also visible in that table, and it is fair: 24% growth in 27% of the business is buying only 25bps of full-year margin expansion, and the other 73% is growing 1.9% and decelerating from 3.8%. If agentic work is not structurally higher-margin, the mix shift is a treadmill.
3. Moat
Quantitative base. ROE 22.4%. ROIC = NOPAT ($760M × 0.75 ≈ $570M) / invested capital $4.09B = 13.9% — above cost of capital, stable, and materially better than Akamai's 5.5% or KT's ~5%, though well below Infosys's 31% (Genpact carries $1.78B of acquisition goodwill that Infosys does not). Gross margin rising 35.2% → 36.5%. Margin direction is the one quantitative signal that separates moat erosion from moat intact, and here it points the right way.
| Moat source | Strength | Direction |
|---|---|---|
| Switching costs (embedded process operations) | Strong | Stable, but see below |
| Intangibles — domain depth in F&A, supply chain, insurance | Strong | Stable |
| Cost advantage (India delivery) | Moderate | Eroding — same AI deflation as Infosys |
| Client diversification | Strong | Improving |
| Network effects / efficient scale | None | — |
Client concentration is far better than the reputation suggests. Top-5 clients 15.2% of 2025 revenue, top-10 23.4%, top-20 34.4%. No individual client is broken out at all — GE has no revenue disclosure in the FY2025 10-K (it was ~9% in 2021; the exact year it stopped being disclosed is unverified). No client exceeds 10% of receivables. The "GE captive" framing is a decade out of date, and this is a genuine, checkable strength.
Contracts: MSAs generally 3–7 years, SOWs 2–5 years. But the 10-K states MSAs "in most cases do not specify pricing terms or obligate the client to purchase a particular amount" — contract length overstates lock-in. No renewal rate is disclosed.
The bear case has a contractual mechanism, and the 10-K says so plainly. MSAs let Genpact retain a portion of productivity benefits, but "some of our MSAs and SOWs require certain minimum productivity benefits to be passed on to our clients." The risk factors are unusually candid: services "will be replaced, in whole or in part, by AI… clients may not accept new pricing or commercial models," with explicit flags on AI-native competitors and clients "who may develop their own AI-related capabilities." Management is not hiding from this — which is worth something, and is a favourable contrast with Infosys declining to quantify its own compression.
Adversarial stress-test. 1. The client's own India GCC. Global Capability Centres are booming and are Genpact's direct substitute — the same delivery model, in-house, without the margin. This is the most underrated threat because it needs no innovation at all. 2. An AI-native entrant with no seat revenue to cannibalise, pricing per-outcome from day one, unencumbered by 141,000 employees. 3. Accenture / Capgemini-WNS bundling agentic operations into transformation deals Genpact cannot lead. Accenture booked $5.9B of agentic AI in its last fiscal year.
The defence, and its weakness. Genpact's claim is that "context-rich process intelligence" — the last-mile knowledge of how work actually breaks in a specific client's insurance claims process — beats models. That is plausible and it is what an agentic system most needs. But it is an assertion, and it contains its own refutation: every process Genpact digitises makes that context more portable. They are converting tacit knowledge into documented, transferable form on the client's premises.
Evergreen rating: 5.5/10. Enterprises will always outsource process operations. That the provider earns 17.4% margins on 141,000 people is the open question — the same one Infosys faces, but Genpact has two quarters of evidence and Infosys has none.
4. Sentiment
The industry backdrop is genuinely bad, and the market is not wrong about it. - BPO annual contract value fell 14% in 2025, the lowest since 2020, as enterprises froze commitments pending AI clarity. - WNS was acquired by Capgemini — $76.50/share cash, $3.3B, closed 2025-10-17. - TCS's CFO explicitly warned of "revenue deflation" absent a move from effort-based to outcome-based pricing. Infosys cut FY27 guidance citing AI-led deflation (see the companion INFY report of the same date). - Concentrix Q2 2026: revenue +1.9%, margins contracting, full-year guidance cut. - TaskUs: ⚠️ contested — a Blackstone/co-founder $2B take-private at $16.50/share was announced May 2025; sources conflict on whether it completed or shareholders voted it down in Oct 2025. Do not rely on this datapoint.
Analysts — uniformly negative over six months. JPMorgan cut to Underweight; Citi cut to Neutral; Mizuho $49→$39; Baird $54→$45; Needham $55→$50. The stock is ~−30% YTD and −19% over twelve months. A prior print produced a ~19%-scale single-day drawdown tied to management refusing to give specific revenue guidance — a credibility wound that has not fully healed. Yahoo's consensus is nonetheless buy with a mean target of $42.18, ~24% above the tape.
Insiders: all sellers, zero open-market buys. CEO Kalra, CFO Weiner, CLO White and two SVPs, roughly $16M of net selling over twelve months. Kalra has 13 sells and 0 buys in five years. ⚠️ Aggregator data, not primary Form 4s. This is the loudest negative signal in the file and the hardest to explain away: if management genuinely believed the agentic pivot was being mispriced at 7.9× forward earnings, someone would be buying.
The offsetting ownership signal. Nalanda Capital (Pulak Prasad) added 1.02M shares in February 2026. Nalanda is a concentrated, quality-focused, extremely long-holding-period investor — not an activist and not a momentum buyer. That is a meaningful independent vote. Vanguard holds 6.14%.
⚠️ No take-private interest and no activist stake found. Despite the obvious setup — a services business at ~8× FCF throwing off $724M — zero reported sponsor interest. Treat "PE bait" as an untested thesis, not a catalyst. That said, WNS at $3.3B and the TaskUs attempt at $2B establish that sponsors and strategics are actively transacting in exactly this space, so it is not a fantasy either.
CEO. BK Kalra, CEO since 2024-02-09 (~2.5 years), Genpact's third CEO, successor to Tiger Tyagarajan. At the company since 1999 — a founding-team member from the GE Capital ISC days. The agentic pivot and the rebrand are his. Deep insider knowledge; also, his tenure is short enough that the strategy is unproven.
5. Valuation
Company type: profitable, moderately-growing, dividend-growing services business with real earnings and modest leverage. All four models apply — this is the one name in today's batch where the full framework is usable.
Multiples: P/E(ttm) 10.2× · P/E(fwd) ~7.9× · P/FCF 8.0× · FCF yield 12.6% · EV/EBITDA 7.5× · EV/Revenue 1.26× · P/B 2.20× · Dividend yield 2.20%
Graham IV — √(22.5 × 3.34 × 15.50) = $34.13 against a price of $34.09. A four-cent gap. Of the seven names analysed today this is the only one where Graham lands on the market price, and it is the model's natural habitat: a profitable company with real book value trading at a modest multiple. Graham says fairly valued on trailing earnings and assets, with no premium being paid for the future at all. Everything above $34 must be earned by growth, buybacks, or a re-rating.
Dividend Yield Theory — current 2.20% against a 5-year average of 1.33%. The yield is
65% above its own band, the widest in the company's dividend history. Decomposing per
pitfall-dyt-inverts-when-price-caused-the-yield: the dividend grew 10–11% and the price fell
~30% YTD, so price did most of the work — a question, not a signal. But note the dividend
has never been cut and has risen nine consecutive years, so unlike a distressed high-yielder
the numerator is not at risk. Reversion to a 1.33% yield implies ~$56, which is the optimistic
bookend and should not be taken as a target.
Dividend coverage. Correct denominator for an ordinary equity is FCF. FY26 dividends ~$127M (at $0.75) against FCF $724M = 17.5% FCF payout, 21% of EPS. Stress it: the dividend survives a 75% cut to free cash flow. Verdict: covered with an enormous margin. This is not a yield story — it is a growth-of-income story with a 13% nine-year CAGR and essentially no coverage risk.
Bogle expected return: 2.2% dividend + ~5% buyback yield + ~7% revenue growth (or ~12% adjusted EPS growth as guided) = low-to-mid teens before any multiple change. Re-rating to 10× forward adds ~27%; de-rating to 6× costs ~24%. The multiple is the swing factor, as it always is on a value name.
Reverse-DCF — what is priced in? At EV $6.63B against ~$724M of FCF, a 10% discount rate implies the market expects free cash flow to decline roughly 2–3% per year in perpetuity. That is an explicit terminal-decline assumption. Genpact is currently growing FCF at 23% and revenue at 7%. The gap between what is priced and what is happening is the investment case.
Fair value range: $39.00 – $48.00, centre ~$43. - Bear $30 (outside the range, stated for honesty) — 6× forward; CBS turns negative, give-back clauses bite, ATS margin never materialises. - Base $43 — 10× FY26 adjusted EPS ~$4.32; the two-quarter productivity divergence holds, CBS stays barely positive, margins expand as guided. - Bull $48 — 11× on $4.35 plus a partial re-rating toward services-sector norms; agentic TCV exceeds $1B and converts.
At $34.09 the stock trades at Graham fair value and ~21% below the base case, with a 12.6% FCF yield and a 7.2% shareholder yield paid while you wait.
6. Tensions and the debate round
Fundamentals vs. Moat. Fundamentals reports 23% FCF CAGR, rising gross margin, and a headcount/revenue divergence. Moat replies that the 10-K contains a clause requiring productivity give-backs on some contracts, and that every digitised process makes the "context moat" more portable. Resolution: split, and this is a genuine tension, not a resolvable one. Fundamentals owns the last eight quarters; Moat owns the terminal question. The correct response is not to pick a side but to size the position for a real chance of being wrong — and to note that at 7.9× forward you are being paid a great deal to accept that uncertainty.
Valuation vs. Sentiment. Valuation says Graham lands within four cents, the FCF yield is 12.6%, and the reverse-DCF implies perpetual decline against a business growing 7%. Sentiment replies that JPMorgan says Underweight, Citi says Neutral, industry ACV fell 14%, and every insider is selling with zero buys. Resolution: Valuation wins, but the insider signal caps conviction. Analyst downgrades on a sector-wide thesis are a lagging, crowded view — the same houses cut Infosys, and Infosys deserved it while Genpact's numbers are pointing the other way. Insider selling is harder to dismiss. The genuine counterweight is Nalanda adding 1.02M shares, which is the more informed buyer of the two groups.
Genpact vs. Infosys — the direct comparison, since both are in today's batch.
| Genpact (G) | Infosys (INFY) | |
|---|---|---|
| Latest revenue growth | +7.1% | +2.4% cc |
| Guidance action | Raised ATS target | Cut FY27 range |
| Headcount vs revenue | −2.6% vs +7.1% | flat vs +2.4% |
| Gross margin trend | +60bps | flat for 4 years |
| Forward P/E | 7.9× | 14.8× |
| FCF yield | 12.6% | 7.3% |
| ROIC | 13.9% | 31% |
| Balance sheet | Net debt 0.95× FCF | Net cash |
| AI compression | Disclosed in the 10-K | Refuses to quantify |
Infosys is the better business. Genpact is by a clear margin the better investment at today's prices — cheaper on every cash metric, growing three times faster, and more honest about the risk. If only one of these two belongs in a portfolio, it is G.
7. Risks
- Productivity give-backs are contractual — some SOWs require passing minimum savings to clients. AI efficiency can be revenue-negative by contract, not merely by competition.
- Core Business Services is 73% of revenue growing 1.9% and decelerating. ATS cannot carry the company if the core turns negative.
- Margin has not confirmed the mix shift — only 25bps of FY26 expansion with ATS at 27% of revenue. If agentic work is not structurally higher-margin, the thesis fails quietly.
- Client in-sourcing via India GCCs — Genpact's own delivery model, rebuilt in-house.
- Accenture / Capgemini-WNS bundling squeezing Genpact out of lead transformation roles.
- Industry demand freeze — BPO ACV −14% in 2025.
- Guidance credibility — management already punished once for withholding specifics.
- Insider selling with zero open-market buys at a claimed-cheap multiple.
- Two-quarter track record on the headcount/revenue divergence — too short to call a trend.
- Bermuda domicile with India-heavy delivery — tax and immigration policy exposure.
- Value-trap risk generally — per §0 of the framework, a poor business at a cheap price is a trap. The whole verdict rests on Genpact being a good business, and the evidence for that is eight quarters deep, not eight years.
A portfolio-specific passage was removed from the public build.
9. Verdict
ACCUMULATE — conviction 6.5 / 10. Fair value $39–48, centre $43. Entry $30–36 (the stock is in the zone now). Trim at 12× forward earnings.
Return to the framework's first principle: is this a good business, and has the market priced it? On the first question the evidence is favourable but young — rising gross margin, 13.9% ROIC, 23% FCF CAGR, low client concentration, and two quarters of growing revenue on falling headcount, which is the specific proof the whole sector is being marked down for lacking. On the second, a reverse-DCF says the market is pricing perpetual 2–3% FCF decline in a business currently compounding cash flow at 23%. Graham lands within four cents of the tape, meaning nothing at all is being paid for the future.
That is the configuration the framework calls value rather than value trap — provided the business is genuinely good. The honest statement is that this is a close call on business quality and a clear call on price. Two quarters of evidence is thin, the give-back clauses are real and contractual, and every insider is selling. Those are why this is ACCUMULATE at 6.5 and not BUY at 8.
Size it as one position in the AI-displacement sleeve rather than adding INFY, DBX or AKAM alongside it. Scale in; do not take a full position before the November print.
What would change the rating: a third and fourth quarter of revenue growth on falling headcount → upgrade toward 7.5 · adjusted operating margin beating 17.7% → upgrade · CBS growth turning negative → downgrade sharply · any insider open-market purchase → meaningful upgrade · a take-private bid → the thesis resolves at a premium.
Next check: Q3 2026 print, early November.