low angle view on modern office building with blue glass windows
getty
Global finance is shifting faster than its infrastructure can keep up. Tokenized assets, prediction markets and AI agents are all scaling at once, but each depends on something the current system was never built to deliver: market data that is verifiable, licensed, and machine readable by default. The real story of 2026 isn’t the products on the surface — it’s the race to rebuild the data layer beneath them.
Each of these shifts depend on market data that is reliable, licensed and machine-readable in ways legacy infrastructure was not built to deliver. The plumbing that has served a world of nine-to-five equity desks and proprietary terminal feeds is now asked to support continuous on-chain settlement, real-time political and economic probability pricing as well as autonomous software counterparties.
Tokenization Has Outrun Its Reference Data
The institutional pivot toward tokenized assets is no longer speculative. Tokenized real-world assets reached $19.3 billion by the first quarter of 2026, with tokenized U.S. Treasuries alone crossing $10 billion in February of this year. Clearer regulatory frameworks for digital commodities crafted under SEC Chairman Paul Atkins’s leadership have removed the enforcement hangover, accelerating issuance from firms that previously stayed on the sidelines.
Yet, moving assets on-chain does not replace the need for underlying reference data. For example, a tokenized Treasury bill still requires a price, a yield curve and a valuation methodology, all of which must be sourced off-chain and then delivered with cryptographic assurances about provenance and licensing. The first generation of oracles solved the bridging problem, but the question of integrity has remained unanswered. For institutional balance sheets, an unlicensed or nonauditable data feed can be a nonstarter, hindering adoption. Providers that can deliver this combination are positioned to become the pricing utility for tokenized markets in the same way index providers and rating agencies are instrumental in traditional finance.
Prediction Markets Are Liquid, But Their Information Is Still Gated
Prediction markets like Polymarket and Kalshi have emerged as some of the most watched real-time signals in macro and political analysis. Monthly trading volume climbed from roughly $1.2 billion in 2025 to more than $25.7 billion in March 2026, with a breadth that increasingly attracts professional market makers alongside retail participants.
The rise of prediction markets is reshaping demand for cross-asset data. Making markets across thousands of contracts spanning elections, commodities, corporate events and macro indicators requires high-fidelity inputs across asset classes that were not designed to be priced together. The participants who solve the data-assembly problem first, whether it’s incumbents repositioning their feeds or new entrants packaging data natively for event markets, will capture the margin that today accrues only to the largest liquidity providers. Broadening access to that data stack is what allows prediction markets to deliver on their promise as price-discovery venues rather than a venue dominated by a handful of well-resourced trading desks.
AI Agents Need Data Built For Machines, Not Terminals
Agentic AI systems are beginning to act as market participants, executing workflows that previously required a human at a terminal. Estimates put the AI agent market above $12 billion in 2026. At the same time, research notes growing constraints and commercialization around data and AI as real-time market data becomes increasingly copyrighted and restricted from AI use.
Pricing and licensing data for machine consumption remain unsettled. AI developers are struggling to figure out how to price data, whether by the token, by usage or through a fixed subscription, as licensing becomes the next frontier. The future requires data that is natively formatted and licensed for machine consumption.
The Race That Actually Matters
It is tempting to frame tokenized markets, prediction markets and autonomous agents as separate stories about distinct venues, new participants and emerging technologies. However, a more holistic view is that they are better understood as a single demand on the same underlying layer: market data that is verifiable, permissible and machine readable by default. These three trends share the same dependency on data as they shape the next generation of finance, and that convergence is where the next generation of market-data businesses may be built.
