EXL's AI-Led Revenue Hits 61% While Its Legacy Operations Line Shrinks

EXL's data and AI-led revenue reached $362.6 million in the second quarter of 2026, 61% of the company's $594.8 million total and up 30% year over year, while its legacy digital operations line fell 1.5% to $232.2 million in absolute dollars (EXL, GlobeNewswire, July 28, 2026).

That is not two lines growing at different speeds. It is one line taking revenue directly from the other, and substitution has different implications for a buyer than an add-on does.

Key Takeaways

• $362.6 million of data and AI-led revenue, 61% of the $594.8 million total and up 30%, while legacy digital operations fell 1.5% in absolute dollars to $232.2 million.
• The two categories moved in opposite dollar directions for the first time in the same release, which is the difference between a substitute and an add-on.
• Gross margin reached 38.0% overall and 34.6% in the insurance segment, movement a pure reclassification of the same work would not produce.
• Headcount grew 986 sequentially to 68,413 with attrition at 24.7%, so AI is raising output per employee rather than eliminating the analytical layer.
• iMerit at up to $310 million pushes EXL upstream into model training and evaluation, a layer few insurers have internal capability to build.

One Line Taking Revenue From the Other

Data and AI-led revenue is not a GAAP segment. It is an internal classification EXL discloses voluntarily, defined as work delivered through "data management, analytics, AI services and solutions," including AI-embedded operations such as payment integrity, set against digital operations, the "managed services that blend the company's deep domain expertise with industry-specific solutions" (EXL, Form 10-Q). Engagements move between buckets as AI displaces the labor content of a workflow.

The four-quarter trend is accelerating rather than plateauing: 18% year-over-year growth in the third quarter of 2025, 21% in the fourth, 28% in the first quarter of 2026 and 30.7% in the second (EXL Q2 2026 investor presentation). Total company revenue grew 15.6% over the same comparison, so the AI-led category is capturing growth from both new demand and the digital operations base.

Metric	Q2 2025	Q2 2026	Change
Total revenue	$514.5M	$594.8M	+15.6%
Data and AI-led revenue	~$277.4M	$362.6M	+30.7%
Digital operations revenue	~$235.7M	$232.2M	−1.5%
Insurance segment revenue	$172.2M	$197.8M	+14.9%

Total operations revenue, the sum of both categories, still rose 10%, which is the evidence EXL points to that carriers are expanding scope rather than relabeling existing work. The margin data supports that reading independently. Insurance segment gross margin was 34.6% and overall gross margin 38.0%, up from prior periods as the AI-led mix grew. A pure relabeling exercise would not move the blended margin.

The first quarter crossed 60% of $570.4 million. The second quarter is the first time the two categories moved in opposite dollar directions in the same release rather than merely at different rates.

What Buying the Substitute Does to In-House Capacity

The insurance segment, EXL's largest, grew 14.9% to $197.8 million, against 12.6% growth to $193.9 million in the first quarter. It covers claims processing, underwriting support, subrogation, regulatory reporting and customer operations: the functions sitting closest to a carrier's actuarial teams without being actuarial work product.

As those engagements migrate from headcount-priced digital operations to AI-led delivery, the labor content a carrier used to buy by the FTE arrives as a packaged analytics output instead. The review point moves earlier in the pipeline, from validating a large volume of manually processed transactions to validating the outputs and assumptions of a vendor's model. That is a narrower and more specialized skill set than the general reserving-analyst headcount insurers carried to absorb data-processing volume.

The labor backdrop makes that trade more attractive than it would have been a few years ago. Roughly 30% of insurers already outsource actuarial and risk functions, in a global actuarial services market valued at $25.94 billion in 2026 and projected to reach $38.69 billion by 2035, a 4.8% compound annual rate (Business Research Insights, July 2026). Actuarial, executive and analytics positions have ranked among the hardest to fill for five consecutive years of insurance labor surveys, with roughly half of current personnel expected to retire within 15 years.

Core reserving, pricing, capital and reinsurance work is not being outsourced at anything like that 30% rate. What is moving is the data preparation and analytics layer beneath those functions. EXL's own headcount says the same: the company added 986 employees sequentially to reach 68,413, with attrition at 24.7%.

A vendor genuinely automating headcount away at scale would show flat or shrinking employment against accelerating revenue. EXL is still growing headcount, just growing revenue faster, which is consistent with AI raising output per employee rather than removing the analytical layer.

Where the Substitution Concentrates

The adjacent lines show where it is moving fastest. Payment integrity, which reviews claims for improper payments, is growing quickly enough that related contract-asset receivables rose from $24.8 million at the end of 2025 to $36.9 million by the end of the second quarter, a 49% increase in six months. Healthcare and life sciences overall reached $158.0 million in the quarter and $309.9 million for the half, from $129.5 million a year earlier.

Data management, the discipline of structuring and governing the information feeding both AI analytics and traditional actuarial models, is described by management as accelerating from a smaller base. That line is arguably more consequential than the AI-led headline, because weak governance upstream produces poor reserving and pricing inputs regardless of how sophisticated the layer above it becomes.

A carrier outsourcing data management to the same vendor running its AI-led analytics concentrates the data quality function and the analytical function in one counterparty. The actuarial opinion attached to loss reserves or a rate filing does not move with them; it stays with a credentialed actuary at the carrier. What moves is the documentation burden, from validating manually keyed claims data to understanding how a third-party system classified, triaged or scored that data before it reached the reserving database.

That is not new in kind. Actuaries have long relied on vendor loss development factors, catastrophe models and rating algorithms without rebuilding them. It is new in scale, because a governance framework built for occasional vendor tools does not extend cleanly to a relationship where 61% of the vendor's revenue is AI-embedded delivery, and audit rights and model-change notification terms in the contract determine what the actuary of record can say about provenance.

The acquisition pushes the concentration further upstream. EXL agreed to buy iMerit, described by management as "a recognized leader in AI model training, evaluation, and reinforcement learning," for $170 million upfront plus up to $140 million in earnouts, a maximum $310 million, expected to close July 31, 2026 and to contribute $28.0 million to $32.0 million of revenue for the remainder of the year. Replicating EXL's stack in-house now means replicating a training and evaluation layer alongside an application one.

The margin guidance is the check on how finished that platform is. EXL raised full-year revenue guidance to $2.39 billion to $2.415 billion with adjusted diluted EPS of $2.25 to $2.29, and adjusted operating margin reached 19.7% in the quarter. Management also expects second-half margin below the first half as it funds sales capacity and AI capability build. A purely margin-accretive reclassification would not require that.

Further Reading on actuary.info

• Verisk's loss-cost data pricing power and the actuarial build-versus-buy problem - A parallel vendor-dependence story on the pricing-data side of the same build-versus-buy question.
• EXL Q1 2026: AI Revenue Hits 60% and Reshapes the Insurance Vendor Model - The first-quarter milestone this piece extends, including the vendor ecosystem comparison against Verisk, Guidewire, and Sapiens.
• EXL's $310 Million iMerit Acquisition Moves Insurance AI Into the Training Layer - A deeper look at the deal terms and what the training-layer push means for carrier governance.
• Insurtech's H1 2026 AI Funding Concentrates Around Build-vs-Buy - How capital allocation across the wider insurtech AI market mirrors EXL's own vertical-integration strategy.
• Bevaya and the Build-vs-Buy Data Moat in Insurance-Native AI - A carrier-side comparison for what building a proprietary data advantage in-house actually requires.
• McKinsey's $50B-$70B Insurance GenAI Revenue Thesis - The market-sizing context for why vendor AI-led revenue is scaling as fast as EXL's numbers show.