Mark Zuckerberg told a small team to build a standalone prediction market app. It uses points, not cash. It uses Llama to auto-generate questions and resolve markets. Arena uses Meta’s 3.56 billion daily users as its distribution moat. And it’s already DraftKings stock moving. It is not a side project.
Meta’s Arena prediction market app became public through reporting by The New York Times and internal documents obtained by NPR. Meta CEO Mark Zuckerberg has instructed a team to start building a standalone app called Arena where people can guess the outcome of real-world events. The app is expected to use artificial intelligence to drive the service. Meta’s AI will automatically generate questions from trending topics using Llama, the company’s large language model. Meta’s AI will make “personalized market recommendations” to users who download the app. That same AI will resolve markets in “near real-time,” the documents state.
Arena uses a points-based system rather than real-money wagering at launch. By contrast, real-money features have not been ruled out. Meta’s 3.56 billion daily active users could provide an unprecedented distribution advantage over smaller competitors. News of Meta’s reported plans caused declines in the market values of DraftKings and the parent of FanDuel, Flutter Entertainment. Prediction markets are no longer a niche product. They are the next engagement frontier — and Meta wants in.
What’s Happening & Why It Matters
A $130 Billion Market
Meta’s Arena prediction market app targets a sector that has grown faster than almost any other financial product category in 2026. Kalshi and Polymarket generated a combined $50 billion in trading volume during 2025. That figure has already surpassed $130 billion in 2026, highlighting the growing appetite for event-based trading. Kalshi currently leads the US prediction market, recording approximately $18.36 billion in contract volume in June 2026 alone, compared to Polymarket’s $6.77 billion in non-US volume and an additional $2 billion in US-based activity during the same period.
Analysts at Bernstein estimate prediction markets could reach as much as $1 trillion in annual trading volume by the end of the decade. Against that projection, Zuckerberg‘s decision to assign a dedicated team and make Arena “a top priority” is commercially legible.
Why Arena Is Different
Meta’s Arena prediction market app is not Meta‘s first attempt at this category. Meta ran a nearly identical product called Forecast from 2020 until 2022 before quietly shutting it down. The internal documents cited “the operational cost of manual question curation” as the reason Meta shut that effort down. The documents say the new app under development is a “rebuild” of that app.
The key structural difference between Forecast and Arena is AI. Llama generates the questions automatically and resolves the markets in near real-time. Llama personalises the market feed for each user. The manual curation cost that killed Forecast no longer exists. Additionally, Arena has been codenamed “Antwerp” and “FBForecast” in internal documents, suggesting multiple development tracks are running simultaneously.
Distribution — 3.56 Billion Users Against Kalshi’s 1 Million
The commercial case for Arena is not product quality. It is distribution. To understand the potential scale, consider that Meta reported 3.56 billion daily active users across its family of apps as of March. Kalshi and Polymarket are legitimate platforms with real traction — but their total user bases are a rounding error compared to Meta’s reach.
Meta plans to leverage its existing social media platforms to nudge people to its prediction market. The company could direct users from Facebook and Instagram toward the new app, potentially giving it a significant distribution advantage over smaller competitors. The points-based launch structure also expands the addressable audience. A casual user who will not deposit money to Polymarket will happily accumulate Arena points — especially if the app integrates with Facebook social graphs that make correct predictions visible to friends.
A Regulatory Calculation — Points Now, Cash Later
Meta’s Arena prediction market app uses a points system for a specific and deliberate reason. The design choice sidesteps immediate gambling licensing while leaving a clear path toward monetisation down the line. Real-money prediction markets in the US operate under CFTC oversight — a framework Kalshi has navigated through a landmark 2024 legal victory and Polymarket is still working toward following its 2022 CFTC settlement. Meta avoids that regulatory process entirely at launch — while testing the product at scale. Furthermore, Senator Richard Blumenthal specifically criticised Meta‘s reported plans. “Polymarket is flagrantly violating federal law and past legal orders. Any other administration, the DOJ, CFTC and FTC would be shutting down gambling operations like this.” As TF covered in its Polymarket fake videos article, the sector carries significant regulatory risk that a points-only launch explicitly sidesteps.
Who Arena Threatens
Meta’s Arena prediction market app threatens DraftKings, Flutter, Robinhood‘s event contracts, and the sports betting operators already facing competition from prediction markets. The market confirmed the assessment immediately. Following reports about Arena, shares of DraftKings, Flutter Entertainment, the parent company of FanDuel, and Robinhood experienced declines.
By contrast, the threat to Kalshi and Polymarket is more nuanced. Their crypto-native and regulated-exchange user bases are not the same demographic as Facebook users engaging with a points-based prediction game. The platforms that lose users to Arena are the casual engagement apps — not the serious trading venues. At least initially.
TF Summary: What’s Next
Meta has not confirmed Arena, commented publicly, or specified a launch date. Internal employee testing precedes public launch. No exact timeline is confirmed. Arena is one of several experimental projects Meta is currently developing — including Meta Photos, an AI-generated media standalone app. Meta declined to comment on the NPR report.
MY FORECAST: Meta’s Arena prediction market app will launch publicly before the end of 2026 — and it will reach more registered users in its first month than Kalshi has acquired in its entire operating history. Distribution always wins in consumer internet at Meta‘s scale. By contrast, the points-based launch is a temporary constraint, not a permanent business model. Once Arena demonstrates engagement metrics that justify the regulatory investment, Meta will move toward CFTC engagement for real-money features — probably in 2027. The Llama-powered auto-generation and AI resolution will be Arena’s true differentiator. Polymarket and Kalshi rely on manual or semi-manual question curation. Meta‘s AI can generate and resolve markets on any topic in near real-time at infinite scale. That capability — married to 3.56 billion daily users — is the product Forecast was always trying to be.
Related Stories
- Polymarket Paid Creators to Fake Winning Bets in Viral Videos
- Bernie Sanders Introduces $7 Trillion AI Public Ownership Bill
- Meta Suspended Its AI Employee Monitoring Programme After an Internal Data Leak