Financial data incumbents turn MCP into distribution, and buy credit data to feed it
Licensed financial data served straight to agents with permissions intact. Funds pay, and terminal UIs lose value.
Competitive · S&P iLEVEL MCP + LSEG buys Octus
Written for: Investor or analyst
Next step
Thesis: agents become the main reader of financial data, so proprietary feeds gain value while seat-priced terminals lose it. Confirming signal: an incumbent reporting agent or MCP usage as a revenue line, or another data acquisition at a similar multiple. Already building: S&P Global, LSEG, V7, Reducto, Moshi.
Why now
LSEG is paying $3.535bn in cash for Octus, a credit data provider. That is a clear price for proprietary data that agents can use. With agents as the reader, the terminal's interface stops being the moat. Permissioned feeds and unique datasets become the moat, and document parsing (Reducto) fills the gaps. European banks building on open models make the same bet: own the data, rent the model.
What became possible
S&P Global's iLEVEL now exposes private-markets data through an MCP server to Copilot and Claude, keeping existing access controls. V7 Go reports 89% accuracy on its hardest graph queries for deal screening.
Two ways in
- Entitlement and metering for data over MCP
Usage accounting and licence enforcement when an agent, not a seat-holder, queries licensed data.
- Customer
- Mid-size data vendors such as rating, credit and alt-data shops
- Model
- Share of metered revenue
- Niche dataset roll-up
Buy and package specialist credit, covenant and private-markets datasets as agent-ready feeds, priced for incumbent acquirers.
- Customer
- Strategic acquirers like LSEG, S&P and Moody's
- Model
- Licensing, then exit
Competition
The major incumbents (S&P via Kensho, LSEG, and reportedly Bloomberg) already serve licensed data to agents through their own MCP servers with entitlements. viaNexus already offers per-agent identity, permissions and usage-based paywalls to mid-size providers, so basic entitlement plus metering over MCP is taken.
Full competition notes (10 competitors)
The major incumbents (S&P via Kensho, LSEG, and reportedly Bloomberg) already serve licensed data to agents through their own MCP servers with entitlements. viaNexus already offers per-agent identity, permissions and usage-based paywalls to mid-size providers, so basic entitlement plus metering over MCP is taken. The open problem is licence enforcement for vendors that want to stay independent rather than join an aggregator. No one I found turns contract terms into rules enforced at query time: telling display use from non-display use, judging when an agent's output reconstructs the original data, enforcing redistribution and territory limits, and producing usage records that audits will accept. Industry commentary says outright that MCP does not solve licensing. A second opening is the fund side: an agent-era version of TRG Screen that shows which agent pulled what data under which licence. On the roll-up idea, LSEG's $3.535bn purchase of Octus confirms the thesis but takes the largest independent target. A new entrant could still assemble smaller covenant, loan-document and niche private-markets datasets below Octus's scale. That is an M&A play that needs capital, and it would compete with the acquirers' own deal teams.
- S&P Global / Kensho LLM-ready API MCP server
Remote MCP server for Capital IQ data, live in Claude, ChatGPT, Amazon Quick Suite and Databricks. Uses OAuth, and access is provisioned through the customer's account team.
- LSEG MCP server (Databricks Marketplace and Microsoft Copilot Studio)
Managed MCP server that serves data customers already license from LSEG to agents. Launched on Databricks in July 2025 and in Copilot Studio from October 2025, rolling out in phases.
- Bloomberg Enterprise MCP (reported)
A third-party blog reports a 29 September 2026 launch built on Data License Plus, with entitlement checks before data is returned. Only secondary coverage was found, so this is unverified.
- LSEG acquisition of Octus
$3.535bn deal for a private and distressed credit data firm with over $300m in revenue, announced 22 October 2026. It confirms the credit data roll-up thesis and takes the biggest independent target off the market.
- viaNexus vAST (BlueSky API)
The closest direct competitor for the entitlement product. Agent-ready MCP layer with a scoped identity and permissions per agent, server-side row-level permission checks, and a paywall for usage-based pricing. Brings together over 30 mid-size providers, including MT Newswires, Aiera, BMLL and ExtractAlpha.
- Moesif MCP monetization
Generic per-tool-call metering and billing for MCP servers, connected to Stripe, Chargebee and Zuora. It has no awareness of financial data licensing.
- Zuplo MCP monetization
API gateway with MCP metering and monetization. A general-purpose tool not built for licensed data.
- TRG Screen
Incumbent in tracking market data usage, spend and compliance on the buy side. It is adding AI features but has no agent-specific metering product that I could find, so it is a likely fast follower.
- Intrinio MCP server
Mid-size data vendor that ships its own entitlement-aware, read-only MCP server, which shows vendors may build this themselves.
- 9fin
Independent credit intelligence platform and a likely next roll-up target. Shows that specialist credit data already has well-funded owners.
Checked 2026-10-23
Risks
- Labs may license data directly and cut out middle layers
- Seat-based pricing may resist metered agent pricing
- MCP servers add a new leak and injection surface
Strong team fit; open space is the weak point (5).
tech 8 · pull 8 · space 5 · fit 9 · build 7
Exposed
Companies that could benefit or be threatened. News and context, not investment advice.
Momentum of its main topic
Sources
- S&P Global expands AI-powered capabilities across private marketsFinextra
- LSEG to acquire data provider Octus for $3.5bnFinextra
- V7 cuts costs 78% while boosting accuracy with GPT‑5.6 LunaOpenAI
- The Missing Infrastructure for AI Agents: Documents | Reducto’s Adit AbrahamThe a16z Show
- The Recommendation Engine for Commodity Traders — Anekha Sokhal, MoshiLatent Space
- Leading European Financial Services Organizations Embrace Open Models to Build Proprietary IntelligenceNVIDIA Blog
Status history
- 2026-10-23 Flagged as buildable