A $3.535 billion agreement places the question of information ownership in focus

On October 22, 2026, London Stock Exchange Group agreed to acquire Octus, a credit intelligence, data and workflow specialist, for $3.535 billion in cash. Completion is expected in the first half of 2027, subject to regulatory approval and customary conditions. This is an agreed transaction, not a completed acquisition.

Octus is not a consumer credit-scoring business. It serves investment managers, banks, law firms and advisers across private, illiquid and distressed credit. LSEG is acquiring capabilities to obtain and structure hard-to-access documents and connect them to the work of making and managing financial decisions.

Summarizing a contract and comparing its terms are different tasks

An AI system given one loan agreement can be asked to locate and summarize its clauses. But a question such as 'Are these covenants unusually restrictive relative to comparable deals?' cannot be answered reliably from that agreement alone. It requires a body of comparable transactions and a method for normalizing terms that are expressed differently.

In October 2025 Octus launched Deal Term Analytics and CovenantAI, combining legal analysis, structured deal-term comparisons and AI-assisted document review. Its MCP product description references more than 170 covenant data fields per instrument. Those are company-described product capabilities, not an independent guarantee that every AI output is accurate.

The value lies partly in producing comparable data again and again

LSEG's announcement describes Octus teams sourcing difficult-to-access covenant and deal-term documents, then transforming them into structured, searchable information. Access to source documents and the work required to classify them consistently are not identical to running a model over publicly indexed text.

Banseog's interpretation is that the production process matters as much as the inventory: source material must be acquired lawfully, interpreted on a consistent basis and updated as new transactions occur. AI may reduce processing time without eliminating acquisition rights, quality control or maintenance costs. None of this establishes that a provider has a permanent monopoly.

Octus expanded from reading credit news to managing portfolios

Octus acquired portfolio-management specialist Sky Road in June 2025. It later introduced Sky Road Research Management in December, bringing credit intelligence, company financials and document management into an environment for order management, compliance monitoring and portfolio optimization.

The strategic extension is from answering 'What happened in this credit?' to helping users connect information to the assets and decisions already being managed. That can reduce the distance between research and execution, though vendor product releases alone do not establish measured productivity gains or investment performance.

The interface and the supplier of information need not be the same company

Octus also offers an MCP Connector that lets entitled customers and their AI agents query Octus data through systems such as ChatGPT and Claude. The product documentation describes permission synchronization, links to source records and an additional subscription cost. This is licensed distribution, not free publication of proprietary materials.

Octus therefore supports both its own integrated products and external AI interfaces. Banseog reads this as a separation of the place where a customer asks a question from the supplier maintaining the evidence behind the answer. But opening a new channel also raises pricing, attribution and customer-relationship questions that are not resolved merely by offering a connector.

LSEG's quarterly disclosure provides an earlier signal

LSEG's separate October 22 Q3 2026 update says more than two-thirds of its top 50 Data & Analytics customers were connected to its MCP servers. It also says 80% of its AI agreements had involved the purchase of additional datasets. These are LSEG-wide disclosures and must not be presented as Octus customer adoption or revenue.

At minimum the company is describing AI as a new way to distribute and sell information, not simply a substitute that makes information unnecessary. Acquiring Octus can be read as an attempt to extend that model into harder-to-access private and illiquid credit data. Whether cross-selling and integration deliver the intended benefits remains to be seen.

A valuable information asset does not automatically justify its purchase price

LSEG forecasts that Octus will generate more than $300 million in 2026 revenue and an adjusted EBITDA margin of about 40%. The company plans to fund the purchase through external debt and expects neutral adjusted earnings per share in the first full year after closing, followed by accretion. These are management expectations, not results.

An investment judgment must still weigh the cost of sourcing and verifying information, recurring client use, integration expense, financing and whether outside AI distribution preserves pricing power. The difference between 'hard to replicate' and 'worth $3.535 billion' is substantial, and the transaction itself cannot settle it.

What disappears for the customer if the data provider disappears?

If a research service provides only a convenient summary of broadly available materials, speed and presentation will be central competitive pressures. If losing the service also means replacing expensive permissions, reconstructing comparable transaction histories and rebuilding workflow connections, a different source of value may be present.

That is not a prescription for every firm to build a costly proprietary database. The relevant questions are whether the information is actually needed, what buyers will pay for it, and whether the expense of producing it can be justified. The Octus case turns 'Can AI answer?' into 'Who continuously supplies the information required for a reliable answer?'

BANSEOG VIEW

BANSEOG VIEW | The answer engine and the evidence behind it are different businesses

Octus' public sequence runs from sourcing credit documents to standardizing covenant data, integrating portfolio workflows and distributing permissioned information through external AI. A large archive alone does not explain the strategy.

A provider may remain relevant even when customers use someone else's AI interface, provided its data are valuable, permissioned, current and genuinely useful in a decision workflow. Yet the existence of that capability does not by itself prove retention, pricing power or profitability.

The decision question for information businesses is: if our service vanished, would clients lose a better-written explanation, or a body of evidence and workflow connections they would struggle to reconstruct? The success of LSEG's deal requires future evidence.

SOURCES

Primary sources and references

Prepared Oct 23, 2026. LSEG Oct 22 official RNS confirms proposed $3.535bn cash acquisition and forecast Octus revenue >$300m, ~40% adjusted EBITDA margin, H1 2027 expected closing and debt financing. LSEG same-day Q3 update provides its own top-50-customer MCP connections and 80%-of-AI-agreements additional-data claim: not Octus KPIs. Octus supplier releases and product descriptions support Deal Term Analytics/CovenantAI, Sky Road workflows, permissioned paid MCP. No claims of finalized acquisition, guaranteed AI protection, independently verified model accuracy, or realized synergies.