Executive Summary

RELX PLC (RLXXF) is the sixth major data provider sold off -20% to -55% on AI disruption fears while posting strong execution. The Q4 2025 earnings call (Feb 12, 2026) shows 7% revenue growth, 9% profit growth, 35% margins, 99% cash conversion, and a GBP 2.25B buyback (+50% YoY) — management buying aggressively at the trough while the market prices in structural impairment.

The stock is down -38% over one year, trading at P/E 20x with RSI at 27 (oversold). The bear case is simple: AI kills LexisNexis (Legal) and Elsevier (STM). The bull case: Risk segment (largest, 90% machine-to-machine, proprietary data moats) is an AI beneficiary not victim, and even Legal shows adaptation with "hundreds of thousands" of Lexis+ AI users.

This is part of a cross-ticker pattern: SPGI (-24%), MORN (-50%), FDS (-54%), TRI (-49%), MCO (-18%) — all proprietary data substrate providers treated as AI victims despite strong fundamentals and AI-offensive postures. The common thesis: AI platforms need curated, authoritative data. The market hasn't internalized this yet.

Not urgent. But this deserves deeper work — factor decomposition, edge audit on whether the -38% drawdown is overdone given the earnings power.

The Cross-Ticker Pattern

This isn't a RELX story. It's a data substrate provider panic.

Six names, same narrative, same selloff:

  • SPGI (-24% 1Y, RSI 14.5): CEO Martina Cheung on Feb 10 call: "We see AI really is a net tailwind for the business." AI platforms need proprietary financial data — SPGI doesn't allow LLMs to train on it, licensing model preserved.
  • MORN (-50% 1Y, RSI 12.3, P/E 12.3x): MCP integration with Claude, Copilot, ChatGPT — data substrate for AI platforms
  • FDS (-54% 1Y, RSI 16.1, P/E 13.1x): FactSet's MI weakness is real but doesn't justify -54%
  • TRI (-49% 1Y, RSI 11.4): Thomson Reuters CEO says "beneficiary not victim"
  • MCO (-18% 1Y, RSI 12.1, P/E 34.4x): Moody's credit ratings moat intact
  • RLXXF (-38% 1Y, RSI 27, P/E 20x): This call

The market is treating proprietary data moats as generic AI victims. The companies themselves are deploying AI offensively — new products, new data sources, expanding TAMs. The pattern only becomes visible in aggregate.

Q4 2025: Strong Execution Across the Board

Headline numbers (Q4 2025):

  • Underlying revenue growth: +7%
  • Underlying profit growth: +9% (faster than revenue)
  • Adjusted operating margin: 34.8% (+90 bps YoY)
  • EPS: +10% constant currency
  • Free cash flow: >GBP 2.3B at 99% cash conversion
  • Leverage: 2.0x (bottom of 2-2.5x target range)

All four segments delivered profit growth ahead of revenue growth. This is not a company in structural decline.

The Buyback Signal

Management announced GBP 2.25B in buybacks for 2026 — up 50% from the prior year. GBP 250M already executed.

At 2.0x leverage (bottom of the 2-2.5x range), the company has significant balance sheet capacity. Average M&A spend over the last few years has been GBP 250M vs the 10-year average of GBP 400M. Management is signaling: organic growth is strong enough, capital is better returned than spent on M&A.

The insider transaction data shows $157M in buybacks over three weeks (Jan 27 - Feb 12). The Feb 12 buy was $39M — the day of the earnings call. Management is buying the dip in real-time.

Segment Breakdown: Where the Moat Lives

Risk Solutions (Largest, Highest-Margin)

90% of revenue from machine-to-machine interactions. This doesn't get disrupted by ChatGPT.

Key characteristics:

  • Massive proprietary data sets with contributory network effects (customers feed data back)
  • Heavily regulated data collection and usage — "incredibly difficult to replicate"
  • Adding new data sources: aerial imagery + video for property insurance (analyzed by AI), electronic health records for life insurance, vehicle telematics for auto
  • These are additive products, not displacing existing ones

Management called out strong new sales in insurance. AI is an accelerant here, not a threat.

Legal (LexisNexis): The Battleground

Every analyst on the call asked about AI disruption — Harvey, Legora, Claude CoWork. This is THE bear thesis.

Management's defense:

  • "Hundreds of thousands of users" on Lexis+ AI already
  • Launching Protege (AI agentic workflow tool)
  • Key differentiator: "content-enabled" workflows built on proprietary curated legal content, not generic LLMs
  • Management dismissive of standalone AI tools: "Anything on trusted curated content, then where [our advantage] comes in"

Lower credibility (0.85) on this segment because management's defense is self-serving. The bear case has teeth: if Harvey or Anthropic's Claude can deliver legal research workflows with comparable accuracy, the content moat weakens.

But even here, the hundreds of thousands of users on Lexis+ AI suggest adaptation not displacement. The question is whether RELX can maintain pricing power or if margins compress.

STM (Elsevier): Improving Momentum

Submissions growth >20%, articles published +10%, momentum continuing into 2026.

Management upgraded outlook vs prior guidance: "Outlook more positive than previously."

Key drivers:

  • LeapSpace ramping (new product)
  • Open access moderating mix (headwind fading)
  • Corporate STM cited as "large addressable market with attractive structural growth"
  • Heavy subscription base means new sales in 2025 signal growth for 2027-28

STM is the smallest bear case surface — scientific publishing demand is structurally growing, AI hasn't shown it can replace peer review or curated journal ecosystems.

The AI Moat Defense

RELX's structural defense is identical to SPGI's: AI platforms NEED curated, authoritative data.

From the SPGI Q4 call (ev-5jyhzq):

"We don't allow the LLM providers to train on S&P Global data" — commercial licensing model preserved. Customers saying: "We prefer to work with you guys, we want you to put the functionality into your tools, we want to use single pipe to get our data."

RELX's Risk segment is 90% machine-to-machine. Legal has "hundreds of thousands" of Lexis+ AI users. STM is subscription-based with improving momentum. The company is deploying AI offensively — new data products (aerial imagery, EHRs, telematics), Protege workflows, Lexis+ AI.

The market is pricing the bear case (AI kills Legal/STM) across 100% of the business. The bull case (AI accelerates Risk, data moats hold, management buying aggressively at trough) applies to the majority of earnings power.

Valuation and Technical Setup

Current price: $31.20 (-38% 1Y) P/E ratio: 20x Margins: 35% Cash conversion: 99% Growth: 7% revenue, 9% profit RSI: 27 (oversold) 52-week range: $26.94 - $57.01 (currently at 14% of range)

A quality compounder (35% margins, 99% cash conversion, growing 7-9%) trading at P/E 20x after a -38% drawdown. Pre-drawdown this traded at 30-35x.

The GBP 2.25B buyback at the bottom of the leverage range is the loudest signal — management sees the same valuation gap and is acting on it with 5% of equity.

What the Market Missed

  1. Cross-ticker convergence. Six proprietary data providers, all -20% to -55%, all executing well, all deploying AI offensively. The pattern is only visible in aggregate.

  2. Risk segment is majority of value. 90% machine-to-machine, contributory network effects, heavily regulated, AI-native. This is not a victim — it's a beneficiary.

  3. Legal is adapting, not dying. Hundreds of thousands of Lexis+ AI users, Protege launching, content moat vs generic LLMs. The bear case may compress margins but unlikely to kill the business.

  4. Management buying aggressively. GBP 2.25B buyback (+50% YoY), $157M in three weeks, $39M on earnings day. They see the disconnect.

The Bear Case

Legal disruption is real. If Harvey or Claude can deliver legal research workflows with comparable accuracy using generic LLMs + web scraping, LexisNexis loses pricing power. The "content-enabled workflows" defense is management spin until proven otherwise.

PitchBook deceleration at MORN and MI weakness at FDS show that even proprietary data providers face headwinds when AI tools replicate their functionality. The selloff isn't entirely irrational.

RELX trades at a premium to the group (P/E 20x vs MORN 12.3x, FDS 13.1x). If the market is right about AI disruption, RELX should trade closer to 15x.

What Needs to Happen

This is not a buy recommendation. It's a research escalation.

Required work:

  1. Factor decomposition. What % of returns are idiosyncratic vs sector/market? Is the -38% drawdown company-specific or index-driven?
  2. Edge audit. Do I have unusual insight on whether the AI disruption thesis is overdone for Risk vs Legal?
  3. Cross-ticker comparison. If SPGI/MORN/FDS/TRI/MCO are all mispriced, which has the best risk/reward?
  4. Legal segment deep dive. Can LexisNexis maintain pricing power? What's the Harvey/Legora/Claude competitive threat timeline?
  5. Options setup. RSI 27, -38% drawdown, P/E 20x — is there an asymmetric put-selling or LEAPS opportunity?

Not urgent. But the combination of strong execution + massive buyback acceleration + insider buying at trough + RSI 27 oversold + cross-ticker convergence suggests the market has the sign wrong on data substrate providers as a group.

The question is whether RELX specifically is the best vehicle for that thesis, or if SPGI/MORN at cheaper multiples are better entry points.

Conclusion

RELX is a $56B quality compounder (35% margins, 7-9% growth, 99% cash conversion) trading at P/E 20x after a -38% drawdown driven by AI disruption fears. Management just accelerated buybacks to GBP 2.25B (+50% YoY) at the bottom of their leverage range.

The bear case (AI kills Legal/STM) applies to maybe 30-40% of the business. The bull case (AI accelerates Risk, data moats hold) applies to the majority of earnings power. The market is pricing the bear case across 100% of the business.

This is the sixth name in a pattern of proprietary data providers sold off -20% to -55% while posting strong execution and deploying AI offensively. The cross-ticker convergence suggests the market has overcorrected.

Deserves deeper work.