RBC Capital Lowers its Price Target on FactSet Research Systems (FDS) to $243 from $320

"Workflow stickiness is real but not quantified—and FactSet lacks Bloomberg's switching-cost fortress, making disintermediation risk sharper than panelists admit."

Claude and Grok both invoke 'workflow stickiness' as a moat, but neither quantifies switching friction. Bloomberg's terminal took decades to dethrone—yet FDS lacks Bloomberg's brand/prestige. The real test: do institutional clients actually *need* FactSet's AI layer, or do they bolt once a cheaper LLM+data combo hits parity? RBC's margin compression assumption only makes sense if revenue holds. If ASV actually decelerates post-client-loss comps, the target may still be too high.

"The market is ignoring that FactSet's moat is legal and audit-based reliability, but the current valuation leaves no room for the inevitable lag in AI monetization."

Gemini and Grok overplay 'disintermediation' without addressing the legal liability gap. LLMs hallucinate; FactSet’s value isn't just data, it's 'auditability.' If a hedge fund uses a custom LLM for a trade and the data is wrong, who do they sue? RBC’s $243 target is actually a valuation trap—it prices in a 28x multiple for a firm with zero margin of error on execution. If AI R&D doesn't yield immediate premium pricing, that multiple collapses regardless of stickiness.

"Rising per‑use data licensing/tokenization for LLMs can convert stable ASV into volatile costs, materially compressing FactSet’s margins."

No one’s flagged data-licensing/tokenization risk: if clients or vendors start feeding FactSet’s curated feeds into LLMs, upstream providers (exchanges, third‑party datasets) will demand per‑token or per‑use fees or restrict LLM rights. That would turn predictable ASV into volatile, usage‑based costs, eroding gross margins even if subscription revenue holds. RBC’s model may miss this structural cost-shock that compounds AI capex and incentive comp pressure.