The panic, and why the numbers don't quite match it
Three high-quality dividend growers have been sold hard on one story: artificial intelligence is coming for business-services and information-services companies. Paychex runs payroll. FactSet sells financial data. Thomson Reuters sells legal, tax, and news information. If AI commoditizes all three, the thinking goes, the moats melt.
Take that risk seriously — these models are exposed, and we'll quantify the bear case for each. But notice the mismatch first. All three are asset-light, high-margin, recurring-revenue businesses. And every one of them grew revenue in each of the last four fiscal years, right through the drawdown.
- PAYX revenue: $5.01B → $5.28B → $5.57B → $6.51B
- FDS revenue: $1.84B → $2.09B → $2.20B → $2.32B
- TRI revenue: $6.63B → $6.79B → $7.26B → $7.48B
That's an unusual financial signature for companies supposedly being disrupted in real time. The share prices fell; the top lines didn't. That gap is the whole reason this list exists.
The rule for the rest of this piece: "it's down, so it's cheap" is not a thesis. A markdown only matters if the dividend behind it is funded by real cash. So for each name we walk the same spine — the drawdown, the yield, and the crux, which is how comfortably free cash flow covers the payout.
Performance and yield figures are historical and may change. Total return includes price movement and distributions where available. Past performance does not guarantee future results. Yield is not the same as total return.
A note on the numbers we're using
Every figure below is drawn from the latest annual financials. We lead the dividend-safety verdict with free-cash-flow coverage — cash generated versus cash paid out — because it's a harder test than the earnings-based payout ratio. Exact dividend-growth streaks and multi-year dividend CAGRs vary by source and record date; for those, check each company's own dividend history rather than treating any single number as gospel. (The performance chart above plots 1-year total return with dividends included, which runs a few points shallower than the peak-to-trough drawdowns quoted in the text.)
1. Thomson Reuters (TRI) — the deepest markdown, the lumpiest earnings
TRI trades at $104.36, which is 43% below its 52-week high of $182.82 — the steepest drawdown of the three. The trailing yield is ~2.4%, and the stock carries the group's richest trailing multiple at 27.5x earnings.
The earnings-based payout ratio reads 80%, which looks tight. But GAAP net income here is lumpy — $2.69B, $2.21B, then $1.50B across the last three years, distorted by items unrelated to the core subscription business. Free cash flow tells a steadier story: $1.80B → $1.85B → $2.02B.
Against that cash, the dividend consumed $1.06B. Free-cash-flow coverage is 1.91x — the payout takes 53% of free cash flow. That's the number to weigh, not the 80% earnings figure.
The balance sheet is the lightest in the group: net debt of $1.61B, or about 0.5x EBITDA. Interest expense runs ~$183M a year, and free cash flow covers interest roughly 11x. FCF yield is 4.5% (4.3% on enterprise value).
The AI risk, quantified: Interest plus the dividend already claims ~$1.24B, or 46% of operating cash flow, before any capex — and TRI's capex is the heaviest of the three at $645M, because it's actively rebuilding its products around AI. That spending is the bet: it can widen margins, or it can be a treadmill if AI compresses the value of legal and tax content faster than TRI can re-price it. The coverage cushion exists; the reinvestment burden is real.
