ExlService Holdings, Inc. (EXLS) stock intrinsic value fair value estimate & owner earnings analysis

WHAT THE BUSINESS IS ExlService (EXL) runs outsourced operations, data analytics and AI work for large companies, delivered largely from centers in India, the Philippines, South Africa and other lower-cost locations for clients based mainly in North America and the United Kingdom. It has four units: Insurance ($711M of 2025 revenue: claims, underwriting and subrogation support, including its Subrosource subrogation platform and its LifePRO life and annuity administration software), Healthcare and Life Sciences ($533M, serving US health insurers, providers, pharmacy benefit managers and life sciences companies), Banking, Capital Markets and Diversified Industries ($482M) and International Growth Markets ($362M). 2025 revenue was $2.09 billion, up 13.6%: 10.6 points came from existing clients buying more and 3.0 points from new clients, including the August 2024 purchase of ITI Data. Revenue has grown from $1.63 billion in 2023, helped by modest acquisitions such as Clairvoyant (December 2021) and ITI Data. HOW WE MEASURED EARNING POWER We start from operating cash flow, remove swings in receivables and other working-capital items, subtract stock-based pay as a real cost (about $79M in 2025), and subtract the capital spending needed to stand still. EXL spends only about 3.6% of revenue on equipment and facilities; the slice tied to revenue growth is treated as optional investment. Office rent is kept as a cost. On that basis owner earnings per share rose from $0.35 in 2016 to $0.80 in 2022, $1.07 in 2023, $1.15 in 2024 and $1.47 in 2025 (about $240M in total), a 17.5% annual pace per share, helped by steady buybacks that cut the diluted share count from about 171M in 2021 to about 157M in early 2026. Acquired-intangible amortization is small (about $13M a year), so little of the earnings rests on wasting acquired assets. The starting level we value from is $1.15 per share, the median of the last three years. That is deliberately below the 2025 figure of $1.47: the business is steady rather than cyclical, but we will not build on the single best year. The 2025 jump looks genuine (operating income rose 19%, and in the first quarter of 2026 revenue rose 14% and operating income 17%; net income was flat in that quarter only because of a higher tax charge), so the cases grow from the lower level rather than haircutting it further. THE CASES (6% discount rate) Base ($30.44 per share): owner earnings per share grow 12% a year for five years and 8% a year for the next five, reaching about $2.98 in year ten. That is well below the 28% achieved in 2025 and below its ten-year history, reflecting the company's own disclosure that some services and tasks its employees now perform may be replaced by automation and AI, and that clients may develop their own AI tools. The drivers are evidenced: existing clients expanded their spending 10.6% in 2025, capital needs are light, and buybacks keep shrinking the share count. After year ten, earnings hold flat for ten years on the strength of long, embedded client relationships (EXL has managed insurance and healthcare operations for more than two decades), then fade to zero over the following ten. That fade is not a forecast that something ends the business; it is simply where our evidence runs out, and we count nothing beyond it. Because this case describes a continuing business that can keep refinancing its borrowing, we neither subtract its debt nor add its cash; the interest on that debt is already paid out of the owner earnings we count. Bear ($11.66 per share): the AI risk the company names plays out against its headcount-priced work. Growth slows to 5% a year for three years and stalls for four; then, as clients automate work EXL used to do and push prices down, earnings shrink 5% a year for eight years. From year 16 the remaining earnings fade in equal steps to nothing by year 20, when the business ends. Client concentration makes this worse: the top ten clients were 34% of 2025 revenue, and the company says losing any one of them could have a material adverse effect. Because this case ends the business, its debt cannot be rolled over forever and must be repaid out of the business before the earnings stop. EXL owed about $417M at March 31, 2026 under a bank credit agreement that matures on April 18, 2027; the balance had risen from about $299M at year-end, mainly to fund $166M of share repurchases in the first quarter. It held about $254M of cash and short-term investments. We keep $100M of that as working cash (routine bills are covered by incoming receivables in the normal course) and apply the other $154M to the debt right away, which the agreement allows without penalty. The remaining roughly $264M is refinanced at the 2027 maturity, which this case can support because it still has close to twenty years of earnings left, and is then repaid in six equal payments of about $44M over roughly 2028 to 2033, before earnings start to shrink. We credit back the after-tax interest no longer paid once each payment is made (the credit agreement cost 5.7% in 2025). Net of that credit, repaying the debt costs about $0.71 per share. Even in this case EXL is worth more than it owes. Bull ($52.79 per share): the business keeps doing what it has done for a decade. Owner earnings per share grow 17% a year for five years (in line with the 17.5% pace from 2016 to 2025), 11% for the next five and 6% for five more, as its data, analytics and AI offerings (EXLData.AI and its insurance-specific large language model) win more work than automation takes away, and buybacks continue. Earnings then hold flat for ten years and fade out over another ten, again because the evidence runs out rather than because a named event ends the business. As in the base case, the debt is treated as rolled over, and neither the debt nor the cash is counted separately. Our earlier bear case did not say how the debt was handled. Because that story ends the business, the borrowing now has to be repaid within it, which lowers the bear case from $12.37 to $11.66 per share. The base and bull cases are unchanged. WHAT THE SHARES APPEAR TO BE WORTH Our central estimate is about $30 per share, with a range of roughly $12 to $53. The width of that range comes down mostly to one question: whether AI turns out to be a tool EXL sells or a tool that replaces EXL. The company is repositioning toward the former, and so far the numbers support it, but that has not yet been proven over a full technology cycle. MARGIN-OF-SAFETY PRICE The margin-of-safety price is $18, about 40% under the base case. This is a well-run, cash-generative, lightly capitalized business, which would normally warrant a smaller discount, but the disclosed risk of AI eroding demand for outsourced labor is real and hard to size, so the margin-of-safety price allows a margin closer to that of a business facing technology obsolescence. At $18 the price sits about 54% above the bear case, a gap the estimate allows for a company that has grown owner earnings per share in every year but one since 2016. WHAT WOULD MAKE THIS WRONG The estimate falls if revenue growth from existing clients drops toward zero for several quarters, if gross margins compress as pricing shifts from people to outcomes, or if a top-ten client leaves. A further debt-funded buyback program would also enlarge what the bear case must repay. The estimate rises if growth stays in the teens while headcount grows more slowly than revenue, which would show AI is expanding EXL's margins rather than shrinking its market.