Building Market Data for a Machine-Readable Financial System

This article draws on a LinkedIn Live conversation with Mike Cahill (CEO of Douro Labs and Contributor to Pyth Network), hosted by Anthony J. Day. The conversation examined Pyth’s business model, the move from human-readable market data toward machine-readable infrastructure, and the role of AI, tokenization, and programmable markets in shaping the next generation of financial systems.

Details and full conversation here.

The next generation of financial data will be consumed by APIs, algorithms, and AI systems

For decades, market data was designed primarily for human users.

Professionals accessed information through terminals, dashboards, and specialist workflows. The data was presented for people to interpret, compare, and act on.

That model will remain important. However, it will increasingly sit alongside a much larger category of machine-readable financial consumption.

Trading systems, risk engines, autonomous applications, and AI models will need market data that is structured, normalized, low-latency, and available through clear programmatic interfaces.

This is a different market from simply building a cheaper version of an existing terminal.

The future buyer may be software

An AI system does not need a terminal in the traditional sense. It needs clean and consistent data, clear permissions, dependable delivery, and enough context to understand what it is consuming.

A trading system has similar requirements. It needs speed, reliability, defined usage rights, and data that can be integrated directly into its workflow.

That creates a growing need for market data that is designed for machines from the beginning.

Pyth was built with this direction in mind. Pyth’s focus is on making financial data available in a form that can move across applications, venues, and automated systems. That includes real-time pricing, continuous indices, publisher information, and data that can be accessed through APIs rather than only through a human-facing interface.

The opportunity is substantial because machine consumption is likely to expand well beyond traditional trading.

AI companies will need market data for analysis and decision-making. Financial applications will use it to power new products. Institutions will require structured data for internal systems. Autonomous finance will depend on reliable inputs that can be consumed without a human sitting between the data and the action.

A broader market data model

Serving this market requires more than one product.

The first layer is direct access to market data through an API. Customers pay for access to the coverage, performance, and rights that match their use case.

The second layer is index infrastructure. Exchanges and financial platforms increasingly want to create products that operate beyond traditional market hours. That requires continuous pricing, published methodologies, and commercial structures that align the data provider with the venue using the index.

The third layer is a broader marketplace for proprietary datasets.

Many valuable datasets do not originate from the traditional network of exchanges and market makers. They may include specialized financial information, reference data, economic indicators, or other datasets controlled by institutions.

Those institutions need a way to distribute their data while retaining control over attribution, access, and commercial terms. A modern data marketplace can provide that distribution layer and make institutional datasets available to a much wider range of machine-driven applications.

Business model before token economics

One principle is especially important: a token cannot substitute for a real business.

Pyth needs to generate value by providing useful market data and infrastructure. That means building products customers want to access, creating commercial relationships with data publishers, and developing sustainable revenue streams across data access, indices, and the marketplace.

The token economics should follow the network’s underlying activity.

When publishers contribute high-quality data and the network becomes more useful, the incentives should encourage deeper participation and long-term alignment. The purpose is to connect network growth with the people and institutions that make the network valuable.

That is a more durable model than starting with token economics and searching for a business case afterward.

Adoption will happen through trusted interfaces

The next billion users are unlikely to interact with financial infrastructure by managing every underlying technical detail themselves.

Many will access these markets through brokerages, exchanges, fintech applications, and trusted financial brands. Self-custody and direct onchain access will remain important options, but they will not be the only path.

This means the infrastructure must work behind the scenes as well as in the hands of expert users. It must be reliable enough for institutions, flexible enough for developers, and simple enough to power products where the end user never sees the underlying data architecture.

That is how new financial infrastructure becomes broadly adopted. The complexity moves into the system, while the user experience becomes more familiar.

Where Pyth is Going

The long-term opportunity is to make Pyth a market data layer for a financial system increasingly consumed by software.

That means broader asset coverage, more institutional datasets, stronger support for continuous markets, and better tools for AI and autonomous applications.

The market may continue to describe Pyth as an oracle because that is where the network began. The broader opportunity is larger.

Pyth is building a market data network for a world where financial products are more global, markets operate more continuously, and machines play a greater role in analysis, execution, and risk management.

The defining question is no longer whether financial data will become machine-readable. It is which infrastructure will make that data reliable, accessible, and useful at scale.

That is the system Pyth is building.