Phase Space AI

Valuation

ExlService Holdings [EXLS]

ExlService Holdings [EXLS] — Valuation

As of 2026-07-30 · spot $34.71 · Company state: A (mature-stable)

---

1. Base figures

TTM FCF margin is positive, so the interim-FCF correction moves conservatively; --fcf-margin 0.1229 was passed explicitly rather than omitted. Q1 free cash flow is negative in both 2025 and 2026 (−$9.7m, −$11.2m) on a recurring annual bonus-payment pattern; the annual sign is unambiguously positive and no forward sign inversion of the GOOGL type applies.

---

2. Terminal margin — 15.0%

Basis: EXLS's own demonstrated operating margin. Set flat at the FY2025 actual of 15.03% and the window peak, against a TTM of 14.90%.

The five-year record — this is a plateau, not a ramp

+1.14pp of range over five years while revenue nearly doubled. That is a services business with a stable cost structure, and it is the reason the terminal is set flat rather than expanding.

The initial 15.5% was wrong and steady_state_check said so

This memo first tested a terminal margin of 15.5%, on a mix-shift argument (Data-and-AI-led at 61% of revenue and 46.9% gross margin in Healthcare & Life Sciences). The check returned:

• TERMINAL_ABOVE_EVERY_OBSERVED_YEAR: 15.5% exceeds the window peak of 15.0%. Defensible only with a
  named mechanism — operating leverage on a fixed cost base is a reason; 'it has been rising' is not.

The flag was right and the assumption was revised, not defended. The mechanism failed its own evidence test: Q2 2026 revenue grew 15.6% and GAAP operating margin fell 1.1pp (15.8% → 14.7%), while adjusted margin was flat (+0.1pp). A mix shift that does not reach the operating line in the quarter it is largest is not a margin mechanism. Terminal set to 15.0%.

Opex bridge from 10-K lines (FY2025, % of $2,087.679m revenue)

Ties to the filed OperatingIncomeLoss of $313,750,000 and OperatingExpenses of $487,326,000. Constraint satisfied: m_EBIT,T 15.0% ≤ m_gross,T 38.4%.

Note the low gross margin. Per the brief's low-margin rule, EV/EBIT becomes an inadmissible instrument at 1–3% operating margins; EXLS runs 15%, comfortably clear of that trigger, and the gross margin of 38.4% is far above the <15% level that forced the MCK/FN switch to EV/gross profit. EV/EBIT is admissible here. Stated explicitly because it was checked.

steady_state_check.py --ticker EXLS --terminal-margin 0.15 --exit-multiple 16.20 --wacc 0.095

MARGIN RAMP
  2020-12-31 margin      11.5%
  latest margin          15.0%
  moved                  +3.5pp   latest is window peak: True
  terminal assumed       15.0%   (-0.0pp vs latest, -0.0pp vs peak)

RETURNS AND THE WARRANTED MULTIPLE
  ROIC measured                                  18.0%
  warranted multiple at steady ROIC              10.1x
  SCALING CONTAMINATION                          1.00x
  terminal reinvestment rate implied by g/ROIC  16.7%
  exit multiple used vs steady-state warranted   1.60x

FINDINGS
  • EXIT_MULTIPLE_UNWARRANTED_AT_STEADY_STATE: 16.2x is 1.6x the 10.1x warranted at a 18%
    steady-state ROIC.

What it flagged after the revision:

• TERMINAL_ABOVE_EVERY_OBSERVED_YEAR cleared once the terminal was moved 15.5% → 15.0%.
• No MARGIN_STILL_RAMPING flag, despite the latest being the window peak — because the move is only +3.5pp over five years and the last two prints go down. A flat terminal at a plateau is what the check is looking for.
• EXIT_MULTIPLE_UNWARRANTED_AT_STEADY_STATE remains, at 1.60x — the smallest gap of the three names in this cluster (PRGS 2.20x, PTC 1.66x, EXLS 1.60x), driven by the highest measured ROIC (18.0% vs 14.4% and 7.3%). EXLS is the only one of the three where the ROIC is comfortably above WACC and the multiple gap is correspondingly smallest.
• Reinvestment implied by g/ROIC is 16.7% of NOPAT. EXLS's capex is ~$52m/yr against ~$330m of NOPAT (~16%), plus small tuck-ins. The implied and observed reinvestment rates match almost exactly — no reinvestment gap. This is the only name in the cluster where that is true.

---

3. Implied-path test (primary output)

reverse_dcf.py --spot 34.71 --shares 152.782931 --net-cash -97.331 --revenue 2237.314 --terminal-margin 0.15 --exit-multiple 16.20 --fcf-margin 0.1229 --wacc 0.095

THE MARKET REQUIRES: revenue CAGR of 4.1% Held fixed: terminal margin 15.0%, exit multiple 16.20x EV/EBIT, WACC 9.5%, 5 years.

Required vs demonstrated — and here the bases agree

Unlike PRGS and PTC, the bases do not disagree in sign. Every consolidated measure — screened, guided, organic, reported, latest-quarter — sits between +13.5% and +15.6% against a required 4.1%. The margin is +9.4pp on the organic basis, and the screen's +8.9pp was, for once, roughly right.

Verdict: PASS. Not "pass with argument" — the required path sits materially below what the business has already demonstrated and below what the company has guided organically, with the acquisition contribution separately quantified at ~1.3 of 15 percentage points.

Sensitivity over the exit multiple

There is no flip point inside any plausible range. Even at the ROIC-warranted 10.1x — below the lowest multiple EXLS has traded at in the entire post-2024 window (12.1x) — the required ~11.5% still sits below the demonstrated 13.5%. This is the most robust of the three valuation tests in this cluster by a wide margin, and it is robust for the right reason: the required path is low because the demonstrated path is genuinely high, not because the exit multiple is doing the work.

Terminal value exceeds 60% of EV, so the reverse DCF is the primary output per criteria.md.

---

4. Regime-overlap check — EXLS is the most regime-stable name in the cluster

Daily series from Alpaca split-adjusted closes, 2020-07-27 → 2026-07-29, n = 1,509; current share count and net debt held fixed; TTM revenue lagged 60 days to filing availability. The 5-for-1 split of 2023-08-02 is handled by the split-adjusted feed — confirmed by querying SPLITS explicitly.

97.2% and 81.1% — both windows describe the same regime. Contrast ZS at 12.7% and PRGS's EV/Sales at 32.2%. EXLS has not re-rated and its earnings have not step-changed; the long series is usable, and the post-2024 window is used for the anchor as the more conservative of the two.

Window-B EV/EBIT deciles: [15.0, 18.9, 20.1, 20.5, 21.6, 22.9, 24.3, 26.7, 28.4] — a smooth, unimodal distribution with no denominator discontinuity. This is what a valid multiple history looks like, and it is exactly what PTC's is not.

---

5. Twelve-month target

Forward 12-month revenue and EBIT

FY2026 guidance: $2.390–2.415bn, midpoint $2.4025bn. Carrying forward at ~12% (organic guided 13–14%, moderated for the law of large numbers and the −1.5% Digital operations drag) gives a rolling 12-month revenue one year out of approximately $2,690m. At the 15.0% terminal margin — which is also the current demonstrated margin, so no expansion is assumed — forward EBIT is $403.5m.

Target

Anchor: EV/EBIT 16.20x, the current multiple held flat, at the 18.4th percentile of the regime-valid 2024–26 window. No re-rating assumed.

12-month target: $42 (+21.4%), at 16.20x EV/EBIT = the 18.4th percentile of the 2024–26 own-history window, held flat.

Cross-check on a second instrument

Holding EV/Sales flat at the current 2.41x (18.0th percentile of the same window) on forward revenue of $2,690m: EV $6,483m, less $97m of net debt, over 152.783m shares = $41.80.

The two instruments agree to within 0.8% ($42.15 vs $41.80) because no margin expansion is assumed in either. That agreement is the check that the target is not being manufactured by the multiple.

---

6. Criteria scoring