FactSet Finds Its Floor: MCP Turns From Threat to Tailwind
A terminal company the market spent two years pricing as an AI casualty just posted a record ASV quarter — and grew its stock 24% in 90 days while still sitting 44% below its peak.
FDS · Earnings Call · 2026-09-30
The bear case gets a stress test
FactSet reported fiscal 2026 fourth-quarter results on September 30 into an unusually loaded setup. The stock trades 44% below its November 2024 peak, yet it is up 24% over the last 90 trading days. That is the whole tension in one line: a company the market spent two years pricing as an AI casualty — a seat-based terminal business staring down disintermediation by large language models — just delivered its strongest quarter ever and a tape that says the narrative is bending.
The numbers underneath the headline support the tape. Total revenue reached $623M in the most recent filed quarter, up 6% year over year, and management reported 7% organic ASV growth for the quarter — the highest Q4 in the company's history and a "record" $86M of organic ASV added. The thing that makes this interesting is not the growth rate itself — 7% is modest — but the context: the market had de-rated this business to 3.4x price-to-revenue, versus a 2021 peak of 11.3x, and to 12.9x price-to-free-cash-flow against a 41.5x peak. A multiple that compressed that hard does not need perfection to re-rate — it needs a stress test, and this quarter was one.
The core evidence is MCP, or Model Context Protocol — the standard through which agents consume FactSet's data. Management disclosed “over 650 clients were actively engaged under contracts or trials to consume our datasets through MCP as of fiscal year end. API call volume continues to expand rapidly, with Q4 volumes at 7 times the level we experienced last quarter.” — Sanoke Viswanathan, Chief Executive Officer · 2026-09-30 Critically, the AI bear thesis was always that cheap model access would cannibalize seats. The company's answer is that MCP expands the wallet: with workstation clients who adopt MCP, management says the majority grew their total spend with FactSet by 20% to 50%. That is the crux of the MCP delivery story — data plus analytics plus agentic capability bundling into a larger contract, not a smaller one.
What is genuinely new: BCC, consumption pricing, one org
The freshest theme this quarter is the acquisition of BCC Group, a Frankfurt-based real-time data distribution platform. This is a strategic move, not a tuck-in. Sanoke Viswanathan framed it as a full-stack play to displace entrenched incumbents in real-time feeds: “We still have a small market share of a large TAM, and we see this acquisition as a real unlock in opening up a lot of new opportunities for us.” — Sanoke Viswanathan, Chief Executive Officer · 2026-09-30 That is the company pushing the workstation narrative toward data plumbing and entitlement management — the layer between clients and providers — which is a different growth vector than the terminal itself.
The second new theme is pricing architecture. Consumption based pricing appears for the first time as a top keyword, and it matters because it changes the revenue model's risk profile. Management described a structure where clients keep a large predictable subscription base and layer volume-based tiers on top. This is early — Viswanathan called it "very, very early days" — but it is the mechanism by which FactSet intends to capture upside if agents, not humans, become the primary consumers of its data.
Running underneath both is a quieter operational story: the collapse of separate business units into a single unified product organization. This is the connective tissue that explains why margins might expand even as headcount falls and ASV accelerates.
The floor, the caveats, and what management stopped saying
The most important forward number is the margin line. Fiscal 2026 operating margin compressed roughly 180 basis points as the company made foundational investments — multi-cloud infrastructure, cybersecurity, making data machine-ready. The latest quarterly operating margin was 26.7%, down 6.5 points year over year. On the call, CFO Josh Warren drew the line plainly: “We expect margin improvement in FY27 of between 25 and 75 basis points... we believe that FY26 operating margins represented a floor.” — Joshua Warren, Chief Financial Officer · 2026-09-30 A floor is a kept promise, and it matters here because the same "line of sight" on margin improvement was floated back in July — “we see a clear line of sight now on margin improvement coming up in the future quarters.” — Sanoke Viswanathan, Chief Executive Officer · 2026-07-01 This quarter converted that language into quantified guidance.
But the tension is real, and it shows in two places. First, guidance decelerates: FY27 organic ASV is guided to 5%-6.5%, below the 7% just delivered. Management insisted “we did not pull forward anything into FY26. It was just a question of great execution.” — Sanoke Viswanathan, Chief Executive Officer · 2026-09-30 The narrower, higher-than-history range is a bet that execution holds, not a bet that the environment improves.
Second — and this is the skeptic's anchor — the AI monetization claim is not new. The recurring AI monetization theme now spans multiple quarters, and the headline stat is essentially a repeat: last quarter management said “we saw over 10% of the ASV growth came directly from AI SKUs,” — Sanoke Viswanathan, Chief Executive Officer · 2026-07-01 and this quarter it reported the same double-digit share of new ASV. Consistency is fine, but it means the market has heard this before — the novelty this quarter is BCC, consumption pricing, and the margin floor, not the AI contribution percentage.
One telling omission: the partner emphasis has rotated. Prior quarters leaned hard on the Google/Gemini relationship and agentic infrastructure themes; this quarter the spotlight moved to competitive displacement — winning mandates away from incumbents in banking and wealth — and to the BCC acquisition. The Google narrative did not vanish, but it is no longer a top-of-mind keyword, which is worth noting given how prominently it was featured before.
Finally, a cross-market note. The global tape over the last 30 days is punishing the AI-infrastructure complex — AI data centers names, high-bandwidth memory, semis are among the biggest decliners. FactSet is not in that bucket. Its +24% 90-day run is idiosyncratic, driven by its own software-and-data re-rating rather than by the infrastructure wave. That decoupling is the cleanest way to read this report: a name that the market wrongly filed as an AI victim, now being re-filed as an AI beneficiary — one report at a time.