Business

Business

TaskUs sells human judgment at offshore wages, wrapped in software. In FY2025 it turned $1.18 billion of service revenue into $140.6 million of operating income and $102.3 million of net income, on a workforce of roughly 65,500 people billed to about 200 clients [1][2]. The economics are a cash-wage labor spread, not an equity-funded tech model — a distinction that decides how real the margins are, and it is where this chapter starts.

Sources: FY2025 10-K, Consolidated Statements of Income [3]; Adjusted EBITDA reconciliation [4]; Free Cash Flow reconciliation [5]; net debt derived from the FY2025 balance sheet [6].

Who pays, for what

TaskUs is paid by large digital companies to run operational work they would otherwise staff in-house. That work is organized into three service lines — Digital Customer Experience (customer care, 56% of FY2025 revenue), Trust & Safety (content moderation and risk, 26%), and AI Services (data labeling and annotation for machine-learning models, 18%) [7][8]. The industry tab defines what each line does operationally; the point for the business itself is that all three sell the same underlying thing — trained people running a client's process on TaskUs's cloud infrastructure — priced in dollars regardless of where the work is performed [9].

Billing follows the work rather than a single meter. Contracts run on "time-and-materials, cost-plus, unit-priced, fixed-price, or outcome oriented" terms [10]. Historically most revenue was effectively per-seat or per-hour; management is now steering newer work toward outcome-based pricing, and some contracts already commit TaskUs to deliver productivity gains — reduced handle or response time — for the client [11][12]. That shift matters because it moves the revenue meter away from headcount, which is the model's own cost base.

Source: FY2025 10-K, Service revenue by service offering [13].

The mix is rotating. AI Services grew 58.6% and Trust & Safety 23.9% in FY2025, while Digital CX — still the largest line — grew only 8.2% [14]. A year earlier Digital CX was 61% of revenue; it is now 56% [15]. The company is becoming less a call-centre operator and more a supplier of moderation and AI-training labor, though the filing does not disclose the relative margin of the three lines — so whether that rotation helps or hurts profitability cannot be settled from the primary record.

Concentration: a few clients, one very large

The customer base is narrow at the top. The largest client was 26% of revenue in FY2025 (up from 22% in FY2024); the top five were 45%, the top ten 58%, and the top twenty 71% [16][17]. That anchor client — identified elsewhere in the record as Meta (competition) — is both the biggest source of revenue and, because it is automating its own moderation, the clearest single risk to it.

The contracts holding these relationships are thin. They typically run one to three years with automatic renewal, but are generally non-exclusive, carry no minimum-volume commitment, and can be terminated for convenience with notice — some are month-to-month [18]. Against that portability, TaskUs points to cross-sell depth: 73 clients now use more than one service line, up 14% year over year, and it added 34 new clients in FY2025 at a 36% new-client win rate [19]. Entanglement, not contract length, is the retention mechanism.

The unit economics: a cash-wage spread

The engine is labor arbitrage. Cost of services — overwhelmingly the wages of frontline staff in the Philippines, India, and elsewhere — was $736.4 million in FY2025, or 62.2% of revenue, and its increase came almost entirely from $108.2 million of additional personnel cost as headcount grew [20][21]. Roughly 54% of revenue is delivered from the Philippines and 13% from India — the low-cost delivery base against which US-dollar billing rates set the spread [22].

Source: FY2025 10-K, Consolidated Statements of Income and Results of Operations [23].

What makes the margins credible is that they are cash margins. Stock-based compensation was $29.7 million in FY2025 — about 2.5% of revenue — and falling, from $52.8 million in FY2023 [24]. More telling, almost none of it sits in the delivery cost base: only $1.1 million of that $29.7 million ran through cost of services, the rest through corporate SG&A [25]. The frontline that generates the revenue is paid in cash wages, so the gap between billing rate and wage is a real spread, not an accounting one — unlike the equity-heavy compensation that flatters many tech-adjacent operating margins.

Operating leverage lives above that spread, in SG&A. Cost of services scales roughly with revenue — more clients need more people — but SG&A rose only 2.2% while revenue rose 19.0%, so SG&A fell from 24.1% to 20.7% of revenue and carried most of the 260-basis-point improvement in operating margin, to 11.9% [26][27]. Two of those points, though, were non-recurring: SG&A carried $11.9 million of transaction costs from the failed 2025 take-private (history) and shed $15.4 million of prior-year litigation cost [28]. The underlying leverage is real but smaller than the headline margin jump implies.

The shape over time: fall and recovery

Revenue compounded from $359.7 million in FY2019 to $1.18 billion in FY2025, but not in a straight line. FY2023 was a genuine decline — revenue fell 3.8% to $924.4 million as the digital clients TaskUs serves cut their own outsourcing budgets — before the business reaccelerated to 19.0% growth in FY2025 [29]. The FY2021 operating loss was an IPO artifact — a one-time surge in stock-based compensation — not an operating failure; the business was cash-profitable before and after.

Source: FY2025 10-K and prior filings; revenue as reported [30].

Source: derived from reported operating income and revenue, FY2019–FY2025 filings [31].

Where the cash went in FY2025

Reported profit and owner cash diverged in FY2025, and the divergence is the most important thing on the P&L this year. Net income rose 123% to $102.3 million and adjusted EBITDA rose to $249.1 million, yet free cash flow fell to $73.7 million — a 6.2% FCF margin, down from 10.0% [32][33][34]. Operating cash flow held roughly flat at $137.2 million; the drop in FCF came from capital spending nearly doubling to $63.5 million as TaskUs expanded sites, and from a receivables build — accounts receivable rose 27.7% against 19.0% revenue growth [35][36]. Management's own conversion metric — adjusted EBITDA to free cash flow — fell from 47.5% to 29.6% [37].

Whether that is a structural reset or a growth-investment trough is unresolved from the filing, which does not split maintenance from growth capex. It is the first question the cash-quality reader will want answered, and the record does not yet answer it — the Financials and Estimates and Cash Conversion chapters take up the operating-cash-flow, capex, and free-cash-flow trend in full.

The balance sheet: near net cash, LBO goodwill

For an investor whose first test is that the chance of insolvency is near zero, the balance sheet is reassuring. Against $241.4 million of total debt — a term loan maturing in September 2027, with a $190 million revolver undrawn — TaskUs held $211.7 million of cash, leaving net debt of just $29.7 million, well under one-eighth of adjusted EBITDA [38][39]. The operating business is light: property and equipment is $95.4 million on $1.05 billion of assets, consistent with a people-and-software model that leases its sites [40].

The one heavy item is history, not operations. Goodwill of $219.5 million and intangibles of $153.5 million — together 35% of total assets — are carried from the 2018 Blackstone leveraged buyout (history), not from operating acquisitions; TaskUs has made no cash acquisitions in any year of the record [41]. Neither has been impaired. A stripped-down, near-net-cash balance sheet keeps the bankruptcy risk that would end the investment case remote, and that is a load-bearing fact for this business.

The question this report is built to answer

At roughly $565 million of market value in mid-2026, TaskUs trades about 68% below its three-year high and near five times forward free cash flow — a former high-growth IPO the market now treats as a structural loser [42]. Yet the business underneath is growing 19%, cash-profitable, near net cash, and controlled by a founder and Blackstone who together own most of the economics (people). The tension between those two facts is the report's spine:

Is TaskUs a durable, cash-generative outsourcing franchise that the market has mispriced as an AI casualty — protected by a near-net-cash balance sheet and owner-operator control — or is the labor-arbitrage engine at its core being automated away, at its own largest clients, faster than its AI-Services line can replace it?

Everything that follows tests one side of that question or the other: how reliably the reported profit becomes owner cash, whether the client relationships are a moat or a liability, what the controllers' actions say about value, and what price the evidence can justify. The forward consensus already assumes the answer is caution — revenue growth of about 4% in FY2026 against 19% delivered — which is precisely the gap a value reader is paid to test.