ExlService Holdings (EXLS, $4.5B mkt cap) is an IT services firm that sold off 32.6% in February 2026 with the broader IT services cohort (CTSH, INFY, WIT, G, EPAM all -18% to -42%) as the market priced AI displacement of outsourced labor. The Q1 2026 10-Q, filed April 28, contains the first primary-source data point that addresses the bear thesis directly. The filing inverts it.

What the filing shows

Revenue: $570.4M, +13.8% YoY — same pace as full-year FY2025 (13.6%), against a 9-11% FY2026 guide. Q1 is the seasonally weakest quarter (India annual wage resets hit Q1-Q2). Annualized run rate already at midpoint of guide.

The thesis-resolving disclosure:

• Data and AI-led revenue: $341.6M, +27.5% YoY — now 59.9% of total revenue
• Digital Operations: $228.7M, -1.9% YoY — declining in absolute dollars

Traditional managed services is shrinking while AI-embedded workflows grow 27.5%. Progression: 53.5% AI-led (Q1 2025) → 56% (Q3 2025) → 59.9% (Q1 2026). Pace accelerating from ≈125 bps/qtr to ≈195 bps/qtr.

Segment quality: Healthcare +21% / gross margin +140 bps to 45.3%; Insurance +12.7% / margin +110 bps to 37.7%. These two = 60% of revenue, both expanding margins. Banking +8.2% with margin -40 bps; International Growth Markets +13.4% with margin -250 bps on lower existing-client volumes.

Capital allocation: Q1 buybacks $135.8M (3% of mkt cap in one quarter). Centerpiece is a $125M accelerated share repurchase entered March 16, 2026 at reference price $29.88 — funded by $120M draw on the revolving credit facility. Management borrowed money to buy stock at 29% below the 2025 average buyback price of $42.67. $368.5M of $500M authorization remains. Preceding the print, seven officers including CEO Kapoor and CFO Nicolelli purchased ≈$2.5M in shares Feb 17-20 around $28.75 — with full knowledge of Q4 results and FY2026 guide.

GAAP net income +0.8% masks operating income +17.1%; the $10.8M delta is a non-structural tax-rate headwind from $13.2M lower stock-based-comp excess tax benefits at vesting (stock was lower; fewer in-the-money exercises).

What the market thinks

Stock at $31.19, forward P/E ≈14.4x on guide midpoint $2.165. The cohort trades 12-14x; pre-disruption IT services historically trades 18-22x. Backing out implied probabilities: market is pricing roughly 30% permanent BPO impairment (12x), 50% stalled IT services (15x), 20% rerate to differentiated grower (18x+). Implied EV ~+5%.

Our scenario tree puts validates at 50%, mixed at 30%, stalls at 15%, breaks at 5% — implying ~+18% EV outright long, ~+20% pair (long EXLS / short CTSH). The implied edge on P(validates) is on the order of 25-35 percentage points.

Why the gap exists

1. Disclosure asymmetry. EXLS is the only peer disclosing AI revenue at material scale. INFY shows 5.5% (only at a Feb 2026 AI Day, never in earnings releases). EPAM 7% in Q4 2025. CTSH does not disclose. The 59.9% figure has no comparable peer benchmark — there is no like-for-like number for the market to anchor against.

2. Sector narrative stickiness. The Feb 2026 cohort selloff was sector-wide; consensus narratives unwind slowly, typically requiring 2-3 confirming prints. The Q1 10-Q is print one.

3. Cross-ticker not synthesized. CTSH announced Project Leap restructuring (third consecutive year of restructuring, $230-320M) the same day as their 10-Q, and earlier acquired 3Cloud + Astreya for $1.3B inorganic AI revenue. A $40B competitor paying restructuring charges and acquisition prices to chase what EXLS is doing organically is third-party validation. Not yet priced.

Risks (ranked by impact)

1. Q2 print disappoints. If AI-led decelerates to <20% or Digital Operations bounces to +5%, Q1 was timing, not trend. Highest information content; P≈25% in our scenario tree.
2. Idio % is 33%, well below the 75% target. Outright long is partly a sector bet — 67% of variance moves with the cohort. If the cohort goes to 10x, EXLS goes to ≈12x regardless of idio. The cleanest expression is the pair (long EXLS / short CTSH).
3. Banking + IGM softness spreads. Both verticals decelerated with margin compression in Q1. If structural across more verticals, the segment-quality thesis weakens.
4. Peer disclosure compresses the transparency premium. If ACN, CTSH, or G break out AI revenue at 25%+, EXLS's 59.9% lead becomes priced rather than novel.
5. India Labor Code further clarifications. ≈$5M/yr quantified; rules still being finalized.

Catalysts

• June 16, 2026 — ASR final settlement (mechanical; additional shares delivered if stock <$29.88)
• Late Jul / early Aug 2026 — EXLS Q2 2026 10-Q. Decision point.
• Same window — CTSH / G / INFY Q2 prints (cross-ticker dispersion test)
• Late Oct 2026 — EXLS Q3 print (likely first guide raise per FY2025 pattern)
• Feb 2027 — FY2026 10-K (resolves predictions: AI-led >65%, Digital Ops decline, $300M+ buyback deployed)

What would change our mind

• AI-led growth decelerating to <20% in Q2 — migration was timing, not trend
• Digital Operations bouncing back to +5% growth — cannibalization narrative was wrong
• Peer (G, ACN) discloses AI revenue at 30%+ — transparency premium evaporates
• Banking/IGM softness spreading to Healthcare or Insurance — vertical-quality thesis weakens
• CTSH Project Leap delivers cost reductions and growth reacceleration — pair-trade thesis compresses

DEMAND-type factor, 365d half-life, no catalyst urgency. The Q2 print is the single highest-information event.

Evidence