Financial data companies have largely dispelled fears of being replaced by AI, but a new question is emerging for investors: who pays for the "tokens" when AI agents begin consuming vastly more data than human analysts?
The Tokenomics of Wall Street
Who Pays for the Tokens When AI Agents Devour the Data?
Financial data firms have survived the fear of replacement by leaning on their proprietary "data moat." Now a sharper question dominates: as AI agents consume orders of magnitude more data than human analysts, who absorbs the exploding token bill?
~$5M
Annualized token spend seen for some analysts ($15k–$20k / day)
+500%
RBC's AI token usage growth, quarter over quarter
26%
Of companies have a comprehensive grasp of their AI costs
Token consumption is exploding — not a forecast
Google's monthly token processing rose 50× year over year, reaching 480 trillion/month in 2025.
AT&T's multi-agent system, tokens per day
A single enterprise agent workload more than tripled its daily token processing.
Two paths to the token bill
Moody's — absorbs & prices in
Bakes token costs into contract pricing. Manages spend via volume deals and multiple model choices — a "clean, predictable relationship" where customers don't track token consumption.
FactSet — AI-native access
Rolling out FactSet AI for Banking, an MCP server for secure data access, and real-time bond pricing — a foundation for agents to reach AI-ready data directly.
The opportunity
Data firms reposition as amplifiers of AI, not casualties
Broad agent use cases: earnings retrieval, comps, filing summaries, portfolio monitoring, due diligence
The constraint
"Tokenmaxxing" and unpredictable costs blow up budgets
Agent adoption can push spend "orders of magnitude higher"
Data privacy, accuracy and reliability still limit reach
Continue reading The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in ✓ Signed in — this article isn’t included in your current plan.Unlocking the full article…