In the past two years, the prediction market has become the brightest narrative in the cryptocurrency industry. The total trading volume in this sector approached $10 billion at the end of last year, with monthly growth momentum accelerating significantly in the second half of 2025.

However, on the other side of this celebration, there is a role that has always stood outside the spotlight and has repeatedly been scolded by users: the oracle.

UMA's Double-Edged Sword

The significant controversies surrounding Polymarket over the past year, including whether Ukrainian President Zelensky wore a suit (with a total trading volume of $237 million), the Ukrainian mineral agreement (involving $7 million, with a large account manipulating the vote with about 5 million UMA), and whether the Trump administration would declassify UFO files in 2025 ($16 million market, publicly called a "whale proof" scam by users), all point to the same source: UMA's Optimistic Oracle and its token governance structure.

The design logic of UMA's Optimistic Oracle is as follows: anyone proposes a result and stakes a deposit; during the challenge period (usually 2 hours), if no one raises a dispute, the result is deemed true by default; if there is a dispute, UMA token holders vote to decide through the Data Verification Mechanism (DVM).

The advantages of this mechanism are obvious: low cost, capable of handling long-tail events, and able to deal with "subjective issues," such as "does Zelensky's outfit count as a suit," which traditional price oracles cannot handle.

However, the controversies surrounding Polymarket expose flaws in this design. For example, in the Ukrainian mineral agreement event last March, the cumulative trading volume of this prediction event was about $7 million, focusing on whether Trump would reach a rare earth mineral agreement with Ukraine before April.

Despite the fact that no agreement was reached, the market was settled as "Yes." According to reports from The Defiant and Cryptopolitan, the main reason was that a large UMA holder controlled about 5 million UMA through three accounts, accounting for about 25% of the voting weight, and pushed the vote toward Yes. Subsequently, Polymarket explicitly stated in a Discord announcement: "This is not a system failure, but the result of the governance mechanism at work, thus refusing refunds."

It can be said that Polymarket's dependence on UMA is facing systemic risks. Originally designed as a "neutral truth judgment layer," the oracle's centralized distribution of governance tokens has instead become a tool for a few to manipulate market outcomes.

According to cryptocurrency asset data platform RootData, until September last year, when Polymarket began to focus on cryptocurrency events, it urgently needed to introduce a more certain data source, which is why it began to hand over some settlement work to another completely different oracle, Chainlink.

Chainlink: Another Dilemma for the Leader

CoinDesk reports that Polymarket has started to introduce Chainlink to improve the way it determines prediction outcomes. The two sides announced that Polymarket will use Chainlink to automatically settle markets related to asset prices, thereby reducing latency and risks of tampering. Initially focused on cryptocurrency price markets, they are simultaneously exploring application space in more subjective markets.

The significance of this collaboration lies in Polymarket's shift from relying on UMA's "crowd-gaming-style subjective consensus judgment" to directly reading market prices through Chainlink and moving toward automated determinations.

From a market perspective, Chainlink is the undisputed leader in the oracle space, with its market capitalization accounting for over 87% and TVS accounting for 61.58% (approximately $62.9 billion), creating a significant gap with the second-place Chronicle (10.15%) and third-place RedStone (7.94%).

It can also be said that its penetration in DeFi is nearly saturated. Mainstream protocols such as Aave, GMX, Synthetix's liquidation and pricing, Curve's security reference, and Lido's cross-chain standards almost all adopt different services provided by Chainlink.

Market share is reflected in its layout. Chainlink has provided 2,000 price feeds (Price Feeds, on-chain permanent price feeding services) across about 27 chains and deployed Data Streams (low-latency, on-demand verifiable high-frequency pricing services) on 37 networks; the CCIP (Chainlink Cross-Chain Communication Protocol) mainnet now covers 70 public chains and L2, with about 200 cross-chain tokens registered for CCIP standards available for use.

This scale means Chainlink has expanded itself from a "single-chain pricing intermediary" to a "layer for information and asset exchange between multiple chains."

But saturation also means that DeFi is no longer the growth curve for it. According to a deep report from Galaxy, about 97% of Chainlink's cumulative revenue (approximately $399 million) comes from Price Feeds, while VRF (Verifiable Random Function, used for NFT minting, on-chain games), Automation (automated execution), and CCIP together only account for about 1.5%, 0.6%, and 0.5%, respectively.

In other words, Chainlink's cash flow is highly concentrated in the most mature and commoditized pricing business, and the market for this part of the business is already filled, leaving very limited marginal growth space.

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

The first is RWA and institutional finance.

From Chainlink's partnership matrix, it can be seen that it has previously completed cross-chain tokenization asset experiments with multiple institutions in conjunction with Swift; last year, it collaborated with 24 major financial institutions to promote a plan to put corporate actions on-chain, and the DTCC Smart NAV pilot distributed mutual fund NAV data on-chain.

In the same year, Chainlink partnered with Mastercard to provide on-chain cryptocurrency purchase processes for over 3 billion cardholders; the U.S. Department of Commerce (BEA) has also put core macro data such as GDP and PCE on-chain through Chainlink Data Feeds, initially covering 10 public chains.

The second is CCIP cross-chain communication.

CCIP has become one of the standards for cross-chain communication. JPMorgan's Kinexys, in collaboration with Chainlink and Ondo, completed cross-chain DvP settlement experiments for tokenized U.S. Treasury bonds; Aave is using it to promote GHO cross-chain, and Lido has adopted it as the official cross-chain standard for wstETH; CCIP was also launched on Aptos that year, extending its reach to the Move ecosystem.

As of October 2025, CCIP has accumulated nearly $2 billion in token transfer volume.

The third is prediction markets and "event settlement financialization."

The integration with Polymarket is the beginning of this curve. It represents Chainlink's expansion from a sector that originally only served "asset prices" to a broader area of "event settlement." As demand explodes for automated settlement asset categories in prediction markets—such as U.S. stocks, commodities, ETFs, and macro indicators—Chainlink finds a natural extension of its original pricing business.

Overall, while Chainlink holds a leading position in the market, the growth of traditional DeFi price oracles has already peaked; it must rely on RWA, institutional finance, CCIP, and the financialization of prediction markets to rebuild its next growth curve.

The potential in these curves is considerable. According to BCG estimates, the scale of RWA tokenization could reach $16 trillion by 2030; the SWIFT track processes $150 trillion in settlements annually, but the cash-out cycle is measured in "years," whereas token holders' patience is usually measured in "days."

This mismatch may represent the core pressure that Chainlink, as the leader, must face in 2026.

Multiple Oracles Erode the Big Pie of Prediction Markets

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

On this platform, a short-term rise and fall prediction market for commodities such as gold, silver, WTI crude oil, and natural gas, as well as over a dozen U.S. stocks including NVDA, AAPL, TSLA, COIN, and PLTR, and major stock indices and ETFs, will have settlement data provided in real-time by Pyth via WebSocket, with Polymarket sampling once per second.

As a first-party data provider (with market makers and institutions like Jump Trading, Jane Street, Blue Ocean, and LMAX directly publishing), Pyth uses a pull model, allowing data to be delivered with low latency to the application layer.

Moreover, this division of labor structure is not just a choice made by Polymarket alone. Kalshi, which is regulated by the U.S. CFTC, has also integrated Pyth as a data source for its newly launched commodity center, covering commodities like gold, silver, Brent crude oil, natural gas, copper, corn, soybeans, and wheat; Pyth Pro is also providing direct market data access to Kalshi's market makers, with future expansions planned for indices, stocks, and foreign exchange.

When both Polymarket and Kalshi choose Pyth as their settlement layer for traditional financial assets, it reflects an overall convergence demand in the prediction market sector for an "institutional-grade high-frequency data settlement layer," rather than being just an individual platform's engineering decision.

Pyth thus claims a portion of this market, but this position is a subset of the "events related to traditional financial assets," with Chainlink holding the cryptocurrency niche and UMA the subjective category each having their own stake.

We can observe the reality of the oracle space revealed by prediction markets from these three layers of division of labor.

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

UMA's community decision mechanism cannot handle high-frequency prices; Chainlink's on-chain feed model is not the optimal solution for millisecond event settlements; although Pyth has obvious advantages in low-latency pricing, it cannot fully address text-based issues.

Second, every time Polymarket introduces a new oracle, it expands the map of "tradable events."

From UMA's non-standard events to Chainlink's cryptocurrency assets, and now to Pyth's traditional financial assets, each step incorporates more uncertainties from the real world into the on-chain betting spectrum. Following this logic, future macroeconomic indicators (GDP, CPI, interest rate decisions), central bank interest rate decisions, listed company profits, and even AI model releases could all potentially become market categories for Polymarket.

As long as there are verifiable data sources, corresponding markets can be constructed.

Conversely, for oracle projects, this also implies that the wild expansion of prediction markets will not allow any single oracle to enjoy the benefits alone. Each new market will be assigned to the oracle "best suited to handle that type of data structure," with many sharing the pie without overlap.

Conclusion

As the oracle space evolves by 2026, it has essentially transitioned from the early "data pipeline" to a "verifiable fact layer" supporting the entire on-chain economy.

Its targets no longer include only DeFi liquidation and collateral valuation, but also compliance verification for RWA on-chain, trustworthy transmission of cross-chain information, and settlement of prediction markets against real-world uncertainties.

Prediction markets serve as a magnifying glass to observe the competition in this red ocean.

The three-track division of labor of Polymarket, along with Kalshi's synchronous choice in traditional financial assets, reveals a reality: no single oracle can fully serve a mature on-chain application. Every topic on the platform will be assigned to the oracle best suited to handle that type of data structure.

Infrastructure differentiation is already a fact. But when no single project can exclusively enjoy the profits, who can truly become irreplaceable?

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。