Who Gets Paid When AI Agents Subscribe to Data?
The market sold off data providers along with SaaS companies. Our own workflow suggests the former may actually benefit from AI.
This piece is based on my own experience. As I have experimented with AI in my workflow, both as a private investor and CEO of MOI Global, I have accumulated subscriptions to a growing number of data APIs.
When I share an API key with an LLM such as Claude, the model stops guessing. It pulls from a credible source with a schema, a version, and a contractual provenance. Research gets materially better because the retrieval surface is no longer the open web but an authoritative feed.
That reframed the question for me. If I, as a niche participant, am paying for API access because it makes AI more trustworthy, then every enterprise AI deployment (e.g., credit, travel, compliance, or research agents) will eventually do the same. The companies that supply the data on the other end of those API calls could see a step-change in demand. Before LLMs, MOI had not subscribed to any APIs; now we are up to a half-dozen and growing.
I asked Claude to run a comprehensive global screen. The report came back with more than sixty names across financial, geospatial, legal, health, industrial, and consumer data categories. What follows is a curated summary of the Top 10 ideas I found most intriguing.
What makes data valuable to an agent
Five properties matter:
Authority. Regulators, auditors, or litigants are required or expected to cite the source.
Sensor production. The data is generated by owned infrastructure (satellites, probes, labs, payment rails, panels) and cannot be scraped.
Network exclusivity. Two-sided platforms where buyers and sellers both stand on one surface.
Temporal depth. Multi-decade archives that cannot be recreated by a new entrant.
Agent-callable surface. The data is served through versioned APIs with machine-readable schemas and transparent pricing.
The durable opportunities score on at least three of the five. “We own data” on its own is not a moat. And, the revenue model has to capture per-query agent economics, not per-seat SaaS that agents will compress.
The paradox the market is handing us
In 2025–26 the market repriced data businesses downward on the fear that frontier LLMs would disintermediate them rather than elevate them. Wolters Kluwer is off ~55% YoY; Thomson Reuters ~47%; FICO ~47%; Equifax ~28%. Rightmove, Hemnet, and Scout24 have been halved. RELX sold off ~20% on the February 2026 Anthropic news. Clarivate trades at roughly 0.7× sales on a high-margin recurring-revenue base.
If the “agents need authoritative data” thesis is directionally right, then this cohort represents the cleanest contrarian setup of the cycle. Here are the Top 10 names I am going deeper on in my own research.