When the Industry Starts Saying It: Pyth Is Becoming a New Model for Market Data

Every company has a vision for what it’s building. But it hits differently when the market starts saying it for you.

Benchmark, the equity research arm of StoneX, just published a research note on Pyth Network. The framing it lands on is the most important part: Pyth is no longer being described as crypto middleware. It is being described as a direct challenge to the institutional data vendors that dominate traditional finance.

“Pyth is beginning to operate less like a crypto oracle and more like a next‑generation market data platform, one that challenges legacy providers and redefines what real‑time financial data can look like.”

That framing did not come from within. It comes from analysts who cover the financial data industry for a living, after looking at the architecture, the adoption curve, and the commercial trajectory.

From oracle protocol to market data network

The report's central argument is structural. Pyth sources prices first-hand from the institutions that actually set them — exchanges, market makers, and trading firms — and distributes those prices with sub-second latency across the systems that need them, onchain and off. The result, in Benchmark's words, "functions less like a price oracle and more like a price layer that sits between data producers and data consumers.”

That sentence is doing a lot of work. It captures a transition that has been underway for some time but rarely articulated this cleanly from the outside: Pyth is operating as market data infrastructure, with an oracle as one delivery surface among several.

The report frames Pyth's first phase as establishing technical differentiation through first-party sourcing and an architecture suited to active trading. The second phase, now underway, is building the commercial layer on top: subscription products, proprietary datasets, and direct distribution to institutions.

The numbers behind the framing

Benchmark's note pulls together a set of operational data points that, taken together, support the market data thesis:

• 710+ businesses now use Pyth data

• $2.8 trillion in cumulative transaction volume secured

• 60% of the on-chain perpetuals market runs on Pyth

• Pyth Pro delivers feeds across more than 2,200 instruments

• 138+ first-party publishers contributing data

• 114+ blockchains receiving Pyth feeds

The commercial trajectory is the part that matters most for the framing shift. Pyth Pro launched in September 2025. By the end of Q1 2026, it had reached a $3M ARR run-rate. Cahill, Douro Labs CEO, told Benchmark the platform is on track for $10M ARR by year-end, with only around 7% of network users having upgraded to paid tiers so far. He described the recent ramp as "a five-year overnight success."

Those numbers are early-stage by any traditional benchmark. They are also the first externally cited revenue trajectory for an on-chain market data business, which is the broader point.

Three sources of demand

The report breaks Pyth's demand into three categories, and the framework is worth borrowing because it captures where the next leg of growth is coming from.

Crypto and derivatives remain the core. The oracle layer created the wedge — first-party institutional data, low latency, an architecture designed for active trading. That wedge is now being pulled into adjacent categories.

Prediction markets are the clearest example of that pull. Polymarket uses Pyth Pro to power real-time pricing across gold, silver, equities, and indices. Kalshi, the CFTC-regulated venue, also relies on Pyth data. These are environments that demand timely, accurate, cross-asset reference data — and they have chosen Pyth.

AI agents may be the most consequential category over time. Pyth Pro for AI Agents delivers more than 3,000 institutional price feeds through a single integration. The data is already accessible inside Claude and Cursor, with thousands of agents calling Pyth feeds every day.

The AI angle is where the licensing argument gets interesting. As Cahill noted in the report, building extensive machine-readable datasets is hard without clarity about what rights are being sold and reused. The platforms that win, in his framing, will be the ones that provide not just data, but "licensed data with clear distribution rights and programmatic access."

That is a different kind of competition than the one the oracle category was built around.

Continuous markets, continuous data

Benchmark also flagged a product that crystallizes the broader thesis: the Pyth 24/7 Oil Index, a continuously updating crude composite built from both institutional and on-chain sources. Cahill's observation in the report is that no single exchange or legacy data vendor has a comparable view across nights, weekends, and holidays.

This is the structural argument applied to a single asset class. As trading hours extend and asset classes go continuous, data infrastructure built on closed-hours conventions starts to break down. The opening that creates is exactly the kind of category — commodities, FX, tokenized assets — where Benchmark suggests Pyth has the potential to challenge legacy incumbents directly.

The Spotify of market data

Cahill's framing for where this is heading, captured in the report, is "the Spotify of market data": a data layer that acquires financial information directly from the best sources in the world and distributes it across modern infrastructure, with premium subscriptions and AI integrations off-chain and cryptographic verification on-chain.

The contrast he draws is with the current market data model, which the report describes as "fragmented, expensive, and mediated by data vendors." That is the legacy structure Pyth is positioned against, and it is the structure Benchmark's note is implicitly arguing is now contestable.

What the outside view confirms

The most useful thing about an external research note is that it forces a thesis to be stated cleanly. Benchmark's contribution here is a clear articulation of something that has been visible in the data for a while: the next market data platform will not look like a terminal. It will look like a programmable price layer built from first-party sources, continuously updating for continuous markets, spanning venues, assets, and applications, and delivered in a machine-native format that powers traders, algorithms, and agents alike.

Pyth has been building toward that for a while. What changed this month is that the market started describing it the same way.