Built for the next 50 billion market data users

Market data wasn’t built for machines but 50 billion AI agents could become its hungriest users.

Finance is acquiring a second population. It is made of software: AI agents that monitor markets, compare scenarios, retrieve prices, prepare analysis and, where authorized, execute instructions. These agents do not show up in headcount, yet they consume data, make decisions, and increasingly move capital.

“The next 50 billion users” is a provocation that reflects what happens when every human professional is paired with hundreds of machine counterparts.

In that world, the number of market data consumers can dwarf the number of human market participants. Infrastructure needs to scale for both humans and machines.

The second population is already reshaping market infrastructure

AI agents operate continuously, query across venues in seconds, and repeat workflows at massive scale. One analyst checks a handful of prices. An agent can monitor equities, FX, commodities, futures, and digital assets simultaneously, then recheck those markets every few seconds.

This changes what “good market data” means. Agents need machine readable outputs, API delivery, cross asset coverage, real time freshness, transparent metadata for auditing, and licensing that supports programmatic use and redistribution.

Macro signals point in the same direction. Tokenized real world assets are projected to approach **$19T by 2033.** Agentic AI is projected to grow from $5.2B in 2024 to nearly $197B by 2034. Market infrastructure has to serve both trends at once: programmable assets and autonomous decision systems.

Data quality becomes risk management for autonomous finance

For decades, market data products were designed for humans sitting at terminals. AI agents call APIs. They require structured responses. They query repeatedly. They merge data across markets. Their outputs can influence trades, risk limits, portfolio construction, and product pricing.

When data becomes an upstream input to thousands of automated decisions, quality issues compound. A stale price propagates. An unclear symbol becomes a confident answer about the wrong instrument. A restrictive license can block production deployment. In an agentic environment, market data is not only information. It is a control surface for risk.

Pyth Terminal is the evaluation layer for humans and teams

Pyth Terminal is the human front door to market data before integration. Users can browse more than 3,500 feeds across crypto, equities, FX, metals, and commodities, then validate behavior in context: updates, coverage and confidence.

That matters because institutions often purchase data before they can properly test it. Terminal flips that sequence. Teams see the data first, understand coverage, and decide where it belongs across trading, risk, and product systems.

Pyth Pro delivers machine native access through MCP

Pyth Pro connects Pyth market data to AI agents through the Model Context Protocol.

Pyth Pro AI supports agent workflows that need to discover feeds, retrieve current prices, access historical prices, and request candlesticks for analysis and backtesting. Four requirements matter for autonomous finance.

First party sourcing from institutions active in price discovery.

Cross asset coverage across thousands of feeds through a single connection.

Programmatic distribution rights designed for downstream systems to consume, process, and display data to end users within their applications and workflows.

Open delivery through MCP compatible environments so data reaches the tools where agents operate.

Explore Pyth Terminal and join the next generation of finance APIs.