Oracle: The Second Battlefield Behind the Prediction Market War

Author: Chloe, ChainCatcher

Over the past two years, prediction markets have become the brightest narrative in the crypto industry. The entire sector’s total trading volume approached $10 billion by the end of last year, with monthly growth momentum significantly accelerating in the second half of 2025.

But on the other side of this celebration, there is a role that has always remained outside the spotlight, repeatedly criticized harshly by users: the oracle.

UMA’s Double-Edged Sword

In the past year, several major controversies surrounding Polymarket have emerged, such as the Ukraine President Zelensky “wearing a suit or not” (total trading volume of $237 million), Ukraine mineral rights agreements (involving $7 million, with major players manipulating votes using about 5 million UMA tokens), and whether the Trump administration will declassify UFO files in 2025 (a $16 million market, labeled by users as a “whale proof” scam). The root of these disputes all points to the same source: UMA’s Optimistic Oracle and its token governance structure.

UMA’s Optimistic Oracle design logic is: anyone proposes an outcome and posts collateral; if no disputes are raised during the challenge period (usually 2 hours), the result is considered true by default; if disputed, UMA token holders vote via the Data Verification Mechanism (DVM) to decide.

This mechanism has obvious advantages: low cost, capable of handling tail events, and addressing “subjectivity issues,” such as whether Zelensky’s clothing counts as a suit—problems traditional price oracles cannot handle.

However, several controversies involving Polymarket have exposed flaws in this design. For example, the Ukraine mineral rights event in March last year, which involved about $7 million in trading volume, concerned whether Trump would reach a rare earth mineral agreement with Ukraine before April.

Although no such agreement was reached, the market was settled as “Yes.” According to reports from The Defiant and Cryptopolitan, the main reason was a large UMA holder controlling about 5 million tokens across three accounts, accounting for roughly 25% of the voting power in that round, pushing the vote toward “Yes.” Subsequently, Polymarket publicly stated in Discord: “This is not a system malfunction but the result of governance mechanics, so refunds are refused.”

It can be said that Polymarket’s reliance on UMA is facing systemic risks. Originally designed as a “neutral truth arbitration layer,” the concentration of governance tokens now turns into a tool for a few to influence market outcomes.

Data from crypto asset data platform RootData shows that until September last year, when Polymarket began promoting crypto-related events, it urgently needed a more definitive data source, leading it to start delegating some settlement work to a completely different oracle system: Chainlink.

Chainlink: Another Dilemma for the Leader

CoinDesk reported that Polymarket began integrating Chainlink to improve its prediction result determination. Both parties announced that Polymarket would use Chainlink for automatic settlement of markets related to asset prices, reducing delays and tampering risks. Initially focusing on crypto asset price markets, they also explored applications in more subjective markets.

The significance of this cooperation lies in, from relying on UMA’s “group game-style subjective consensus,” Polymarket now has a path to directly read market prices and automate judgments via Chainlink.

From the market landscape, Chainlink is undisputedly the leader in the oracle space, holding over 87% of the market cap in the oracle sector, with a TVS share of 61.58% (about $62.9 billion), far ahead of second-place Chronicle (10.15%) and third-place RedStone (7.94%).

It can also be said that its penetration into DeFi is nearly saturated. Major protocols—from Aave, GMX, Synthetix’s liquidation and pricing, to Curve’s security references, Lido’s cross-chain standards—almost all rely on Chainlink’s various services.

Market share is reflected in its deployment. Chainlink provides around 2,000 price feeds (on-chain persistent price oracles) across approximately 27 chains, and has deployed Data Streams (low-latency, on-demand high-frequency price feeds) on 37 networks; its CCIP (Chainlink Cross-Chain Interoperability Protocol) mainnet covers 70 public chains and L2s, with about 200 registered cross-chain tokens supporting CCIP standards.

This scale effectively transforms Chainlink from a “single-chain price intermediary” into a “multi-chain information and asset exchange layer.”

But saturation also means DeFi is no longer its growth curve. According to Galaxy’s deep report, about 97% of Chainlink’s total revenue (around $399 million) comes from Price Feeds, while VRF (Verifiable Random Function, used for NFT minting and on-chain gaming), Automation, and CCIP account for only about 1.5%, 0.6%, and 0.5%, respectively.

In other words, Chainlink’s capital flow is highly concentrated in the most mature, commoditized price feed business, which has already been saturated, leaving very limited marginal growth potential.

In response, Chainlink is betting on three incremental growth paths.

The first is RWA and institutional finance.

From Chainlink’s partnership matrix, it has previously collaborated with Swift and several institutions to pilot cross-chain tokenized assets; last year, it worked with 24 major financial institutions to advance on-chain corporate actions data, with the DTCC Smart NAV pilot distributing mutual fund NAV data on-chain.

In the same year, Chainlink partnered with Mastercard to enable crypto purchases for over 3 billion cardholders; the U.S. Department of Commerce (BEA) also integrated macroeconomic data like GDP and PCE into Chainlink Data Feeds, initially covering 10 public chains.

The second is CCIP cross-chain communication.

CCIP has become one of the preferred standards for cross-chain interoperability. JPMorgan’s Kinexys, Chainlink, and Ondo completed a cross-chain DvP settlement test for tokenized U.S. Treasuries; Aave uses it to promote GHO cross-chain, Lido adopts it as the official cross-chain standard for wstETH; in the same year, CCIP launched on Aptos, extending into the Move ecosystem.

By October 2025, CCIP had processed nearly $2 billion in token transfers.

The third is prediction markets and “event settlement financialization.”

Polymarket’s integration marks the beginning of this path. It signifies Chainlink’s expansion from serving only “asset prices” to the broader domain of “event settlement.” As prediction markets see explosive demand for automated settlement of assets like U.S. stocks, commodities, ETFs, and macro indicators, Chainlink finds a natural extension of its original price business here.

Overall, while Chainlink remains the market leader, growth in traditional DeFi price oracles has peaked; it must rely on RWA, institutional finance, CCIP, and prediction market financialization to rebuild its next growth trajectory.

These paths hold significant potential. According to BCG estimates, the tokenized RWA market could reach $16 trillion by 2030, and SWIFT processes about $1.5 quadrillion in settlements annually, but with settlement cycles measured in “years,” while token holders’ patience is often measured in “days.”

This mismatch may be the core pressure Chainlink, as a leader, will face into 2026.

Multiple Oracle Projects Eroding the Prediction Market Pie

In early April, Polymarket announced a partnership with Pyth Network.

On this platform, prediction markets for gold, silver, WTI crude oil, natural gas, and more than ten U.S. stocks like NVDA, AAPL, TSLA, COIN, PLTR, as well as major indices and ETFs, will have settlement data provided in real-time via Pyth’s WebSocket, with sampling every second.

Pyth, as a first-party data provider (with market makers and institutions like Jump Trading, Jane Street, Blue Ocean, LMAX directly publishing data), uses a pull model, enabling low-latency data delivery to applications.

This division of labor is not exclusive to Polymarket. Regulated by the U.S. CFTC, Kalshi has also integrated Pyth as its new commodity center’s settlement data source, covering gold, silver, Brent crude, natural gas, copper, corn, soybeans, wheat, and other commodities; Pyth Pro provides direct market data access to Kalshi’s market makers, with plans to expand to indices, stocks, and forex.

When both Polymarket and Kalshi choose Pyth for traditional financial asset settlement, it reflects a broader industry trend toward “institution-grade high-frequency data settlement layers,” not just an isolated platform decision.

Pyth has carved out a segment of this market, but it is a subset of “traditional financial asset events,” separate from Chainlink’s crypto focus and UMA’s subjective data.

This three-layer division reveals the current reality of the oracle prediction market landscape.

First, no single oracle can fully serve a mature prediction market.

UMA’s community arbitration cannot handle high-frequency prices; Chainlink’s on-chain feeds are not optimal for millisecond-level event settlement; Pyth, while excellent for low-latency prices, cannot handle text-based data at all.

Second, each new oracle introduced by Polymarket expands its “tradable event” universe.

From UMA’s non-standard events, to Chainlink’s crypto assets, to Pyth’s traditional financial assets, each step incorporates more real-world uncertainty into on-chain betting. Following this logic, future markets could include macroeconomic indicators (GDP, CPI, interest rate decisions), central bank rate announcements, corporate earnings, and even AI model releases.

As long as verifiable data sources exist, corresponding markets can be built.

Conversely, for oracle projects, this also means prediction market expansion will not benefit any single oracle exclusively. Each new market will be allocated to the oracle best suited to handle that data type, leading to multi-party sharing and non-overlapping niches.

Conclusion

By 2026, the oracle sector has essentially evolved from an early “data pipeline” into a “verifiable facts layer” supporting the entire on-chain economy.

Its scope has expanded beyond DeFi liquidation and collateral valuation to include RWA on-chain compliance verification, cross-chain information transfer, and prediction market settlement of real-world uncertainties.

And prediction markets serve as a magnifying glass to observe this fiercely competitive red ocean.

Polymarket’s three-path division, along with Kalshi’s parallel focus on traditional financial assets, exposes a stark reality: no single oracle can fully serve a mature on-chain application. Every topic on the platform will be assigned to the most suitable oracle for that data structure.

Infrastructure differentiation is now a fact. But when no single project can monopolize the benefits, who can truly become indispensable?