FactSet at 12x: Where the Displacement Evidence Isn't

FactSet Research Systems (FDS) filed its Q2 FY2026 10-Q on April 2, covering the quarter through February 28 — a period that includes the Anthropic-triggered AI panic selloff of early February. FDS trades at 11.6x forward earnings, a 50-60% discount to peers (SPGI 29x, MCO 32x, TRI 27x, MORN 19x). This is the first hard operating data from the quarter the market repriced the entire financial data sector.

What the Filing Says

The subscription business is intact. Organic ASV grew 6.7% to $2.45 billion, with dollar retention above 95% — unchanged from prior year. Clients increased 5.3% to 9,101. Users increased 10.1% to 241,352. These are the metrics that would crack first if AI were displacing FactSet's products. They haven't cracked.

Adjusted EPS grew 4.2% to $4.46. The GAAP print looks worse (EPS -4.5%, net income -8.1%) but three non-recurring items distort it: $10.6M in make-whole awards for the new CEO, $2.9M from India's labor code reform, and a $10.2M equity investment write-down. Strip those and the underlying business grew mid-single digits.

Free cash flow surged 31% to $276 million in the first half. Buybacks accelerated 2.68x to $303 million, with the heaviest buying in February — 298,000 shares at an average $207 — the month the panic hit hardest. The board authorized $600 million in December 2025. $697 million remains, roughly 8-10% of the current float.

The filing positions FactSet explicitly as AI infrastructure. The strategic framing — "sustainable success in enterprise AI depends on trusted, high-quality data" — is offensive, not defensive. The CGS/CUSIP business was rebranded "Market Infrastructure."

Two problems are real. Operating margins compressed 220 basis points (GAAP) and 230 basis points (adjusted), with costs growing 10.6% versus revenue at 7.1%. Cloud and AI infrastructure spend rose 40-60 basis points as a percentage of revenue. EMEA operating income fell 6.1% despite 4% revenue growth. And a material weakness in IT general controls — affecting revenue recognition — continues for a third consecutive year, with disclosure controls deemed "not effective."

What the Market Thinks

At $225 and 11.6x forward earnings, the market prices FDS at a deep discount to every peer. Reverse-engineering probability weights from the peer-relative valuation gap, the implied displacement probability falls in the range of 50-65%, depending on what you assume for the coexistence scenario. The sell side confirms: Goldman has a Sell at $217, Wells Fargo Underweight at $200. Only 3 of 18 analysts are buyers. Short interest is 10.4%.

The options market offers no additional signal — 6 total contracts of open interest on the September expiration. No derivatives-based price discovery exists for this name.

Why the Gap Exists

The selloff was sector-wide and indiscriminate. The February panic hit every financial data name — SPGI, MCO, MORN, TRI, RELX, CLVT — regardless of business model or moat structure. FDS, the smallest and least covered, took the worst hit (-55% peak to trough). A factor regression confirms this: 45% of FDS's trailing variance is explained by a shared thematic factor (the peer basket orthogonalized to SPY), with another 45% truly idiosyncratic and only 10% market beta. The selloff was one trade across nine tickers.

AlphaSense gave the threat a name. At $500 million+ ARR, 73% growth, and 88% of the S&P 100 as customers, AlphaSense is building AI-native financial data products that compete directly with FactSet and Capital IQ. Bloomberg reported in March 2026 that it's seeking fresh funding at "well over" $4 billion, with an IPO likely in H2 2026 or H1 2027. A visible, fast-growing competitor makes the disruption thesis more salient than the abstract question of whether incumbents are actually losing customers.

The cross-ticker corroboration hasn't been synthesized. The strongest evidence against displacement comes from reading across all seven peers simultaneously. Every name shows stable or improving retention. Every management team is accelerating buybacks — $10 billion combined in 2025, all accelerating year-over-year. Every strategic narrative frames AI as additive. Morningstar's 10-K explicitly states the selloff was driven by "sentiment regarding the impact of AI on software and data company growth prospects... despite strong operating performance." Thomson Reuters' CEO said "beneficiary, not victim" on consecutive earnings calls. These are statements made under securities law obligations, by management teams simultaneously committing capital at depressed prices.

FDS-specific factors explain the extra discount. FDS trails the peer group by approximately 16 percentage points even after adjusting for the shared thematic factor. The gap has identifiable causes: margin compression while all six peers expand 100-300 basis points, three years of material weakness in IT controls, and no regulatory moat equivalent to SPGI's ratings franchise. Peer comparison suggests the margin issue is fixable — FDS's compression traces to one-time items (CEO comp, India labor reform) plus AI infrastructure investment — but FDS is behind the curve on converting AI spending into productivity gains. SPGI invested over $1 billion across several years before AI became margin-accretive.

What We Don't Know

We cannot determine from this filing whether aggregate retention masks product-level churn. If data solutions are growing while workstation seats decline, the 95% retention figure hides competitive erosion. User counts (+10.1%) argue against this, but they're not broken out by product line.

We don't know whether EMEA's margin deterioration (-6.1% operating income on +4% revenue) reflects competitive pressure from AlphaSense's European expansion, cost structure problems, or something else. One quarter is not a pattern.

We can't confirm that retention will hold beyond current contract cycles. Contracts run 1-3 years. If new analyst cohorts are adopting AI alternatives, the retention data won't show it until 2027-2028 renewals.

Risks

1. Retention is a lagging indicator. Current data only proves displacement hasn't happened yet, not that it won't. The real test comes when 2025-2026 vintage contracts renew in 2027-2028.

2. AlphaSense IPO above $10 billion would renew the displacement narrative regardless of incumbent operating data. Markets may respond to a growth story's valuation signal more than to seven incumbents' retention metrics.

3. Margin compression may be structural. Cloud and AI infrastructure costs rose 40-60 basis points and may not reverse. SPGI's experience — years and over $1 billion before AI spending became margin-positive — suggests FDS could face a multi-year investment cycle with uncertain returns.

4. Material weakness in IT controls affecting revenue recognition, now in its third year. Some institutional funds exclude names with material weaknesses. Restatement tail risk is low but nonzero.

5. FDS is the most exposed peer. Smallest ($8.4B market cap), no regulatory moat, most concentrated in the workstation layer that faces the most direct AI competition.

Catalysts

June 25, 2026 — Q3 FY2026 earnings. The nearest decision point. Margin stabilization (adjusted margin at or above 35.5% with ASV at 6%+) would support the execution recovery thesis. Further compression keeps the stock in the penalty box.

April-May 2026 — Peer earnings. SPGI, MCO, and MORN report Q1 2026. Continued retention stability weakens the displacement narrative ahead of FDS's own report.

H2 2026 — AlphaSense IPO. Below $8 billion deflates the narrative. Above $10 billion reinforces it.

October 2026 — FY2026 10-K. Material weakness resolution, full-year retention data, and updated risk factors either clear or compound the governance overhang.

What Would Change Our Mind

Product-level churn becoming visible: flat or declining user counts in future filings, or earnings call commentary about workstation-to-data-solution mix shifts, would indicate that aggregate retention is masking competitive erosion. This is the single most important thing to watch.

A named institutional client publicly switching from FactSet to AlphaSense or Bloomberg AI would be a concrete displacement signal that retention data cannot capture.

EMEA operating income declining for two more consecutive quarters alongside revenue growth would suggest geographic competitive pressure, not just cost structure.

FDS adjusted operating margins failing to reach 34% by Q4 FY2026, after one-time items have fully rolled off, would indicate the cost structure has permanently shifted.

Evidence