Meta Arena Rebuilds a Play-Money Market It Closed

Meta is building a standalone prediction app called Arena that would let people bet points, not cash, on sports, politics, and culture. People familiar with the work say Mark Zuckerberg made it a top priority and set a target of 100 million monthly active predictors, ages 18 to 34.

Internal company documents describe Arena as a rebuild of Forecast, the play-money prediction app Meta shut down on October 15, 2021. Executives have also been told to explore partnerships with Kalshi and Polymarket, the cash platforms Arena is supposed to resemble. Meta, Kalshi, and Polymarket declined to comment.

A Rebuild of the Forecast App Meta Already Closed

Forecast came out of Facebook’s New Product Experimentation group on June 23, 2020, first as an invite beta in the United States and Canada. Users spent virtual points to buy or sell outcomes on questions that ran from Covid-19 vaccination rates to the Great British Baking Show. There was a leaderboard for paper profit, and extra points for write-ups other people upvoted.

Rebecca Kossnick, the project lead, argued that a forecast frame could push people toward calmer talk. “We became interested in prediction markets because, when they work, they help push participants to be rational,” she wrote in October 2020. “In a period of increasing uncertainty and division, what we want most is to create a space in the world where people, especially those with divergent views, can find common ground.”

Reviewers still mixed up whether the point was accuracy or argument. The app stayed on iPhone only. Kossnick left in March 2021. On the Forecast site, the company said the experiment had been about crowdsourced wisdom arriving at truth, then set the shutdown for October 15, 2021, about 16 months after launch.

THE FORECAST-TO-ARENA CLOCK

1. June 23, 2020: Forecast launches as a points market for world events, including the early pandemic.
2. October 15, 2021: Meta sunsets Forecast after the team vets questions by hand.
3. September 2025: Combined monthly volume on Kalshi and Polymarket sits under $5 billion.
4. April 2026: That combined monthly volume reaches about $24 billion.
5. June 23, 2026: People inside Meta describe Arena as a new standalone prediction app ordered up by Zuckerberg.
6. June 24, 2026: Internal documents call Arena a rebuild of Forecast and assign Llama to write and settle questions.
7. June 26, 2026: Zuckerberg tells lieutenants to talk with Kalshi and Polymarket and to chase 100 million monthly predictors.

The new documents name the old cost problem in plain language: “the operational cost of manual question curation.” Arena is the attempt to run that same points market without paying a newsroom to write every question.

Zuckerberg Set a Target of 100 Million Predictors

Arena would sit outside Facebook, Instagram, WhatsApp, and Messenger, then draw people in from those apps. Meta’s second-quarter filing shows 3.60 billion Family daily active people on average in June 2026, up 3% from 3.48 billion a year earlier. That is the pipe. The product is still a prototype path: employees test first, then iPhone and Android. The documents do not name a public launch date, and people close to the work have said the app could stay on the shelf.

The audience Zuckerberg wants is 18- to 34-year-olds. The markets would cover sports, culture, entertainment, politics, and finance, with bets aimed at friends and family. Executives have called it an app “designed for everyone.” Arena is one of several stand-alone experiments; another, Meta Photos, would make new kinds of media with generative models.

THE SCALE META IS UNDERWRITING

• Family reach: 3.60 billion daily active people across Facebook, Instagram, Messenger, and WhatsApp in June 2026.
• Predictor goal: at least 100 million monthly active predictors on Arena.
• Age band: 18 to 34 as the first demographic Zuckerberg named.
• Cash at launch: none; a video-game-style points system, with real-money betting still not ruled out.

A hundred million monthly predictors would be a social product on Meta’s terms, not a trading floor on Kalshi’s. The company does not need Arena to clear billions of dollars in contracts for the app to show up in a weekly usage chart.

Why Arena Starts With Points, Not Dollars

On Kalshi and Polymarket, people stake cash on yes-or-no contracts. Arena’s documents instead promise “a daily virtual allotment” of “play money” to bet on “the outcome of future events.” Insiders have compared the points to a video game. The company has not closed the door on cash later.

That split is the whole product. Real-money markets got huge because a wrong call costs something. Points markets are easier to ship in countries that restrict cash betting, and they keep federal derivatives rules and state gambling statutes at arm’s length, so long as the points cannot be bought, sold, or turned into prizes. If Arena later lets people redeem a balance, the fight Kalshi is already in comes back through the same door.

WHAT THE INTERNAL DOCUMENTS SAY ARENA WILL DO

• Daily chips: each user gets a virtual allotment of play money to put on future events.
• Question mill: Llama, Meta’s large language model, writes markets from topics already trending.
• For-you rail: the same model serves personalized market recommendations after download.
• Instant judge: Llama also resolves each yes-or-no question in near real time, with no human review named in the documents.
• Old names: the work has also been called Antwerp and FBForecast, a label that still points at Forecast.
• Ship path: a prototype, then an employee test, then public iPhone and Android builds, with no date attached.

Play money only stays a legal small thing if it stays useless. Rank, streaks, and being right in front of friends can still pull people back the next morning. That loop is what Meta knows how to feed. It is also what Forecast already tried, without Llama and without a hundred-million user goal.

Llama Would Write the Questions and Call the Winners

Forecast died on the cost of people. Arena’s answer is to let Llama scrape whatever is moving online, spin it into a yes-or-no market, and then decide whether the thing happened. The documents say that call lands in “near real-time.” There is no challenge window in the pages that have been described, and no second human check.

Cash platforms live and die on that last step. A sports contract can lean on a box score. An election contract can lean on certified results. A fuzzy geopolitics question cannot. Handing that job to Meta’s own model is a cost cut, and it is also a trust cut. A wrong settlement on play money still publishes a public “winner” that friends will see. A feed that says an event is 80% likely can move people even when no dollar is at risk.

The same model would keep putting new markets in front of the same 18- to 34-year-olds. That is how a points app avoids the editorial payroll that closed Forecast. It is also how a company that already struggles to referee political speech ends up writing the questions and ringing the bell.

More Than 30 Lawsuits Sit Between Meta and Real Money

Daniel Wallach, a gaming lawyer who follows the sector, said a points launch could buy Meta time to seek a federal license and wait out the court map. “We’re clearly in legal limbo,” Wallach said, pointing to more than 30 pending lawsuits over whether these contracts are derivatives or bets. “And we might not have a clear answer for another year or two.”

The prediction markets of today are a novel creation that only arrived at the dawn of the second Trump administration. And it’s launched all these legal fights over whether Congress ever intended for this kind of activity to be regulated by the CFTC.

Daniel Wallach, gaming lawyer

Pew Research Center finds that 23 states prohibit election betting entirely, and that at least 16 states tried to regulate prediction markets in 2026. Federal officials argue the contracts are financial products and have sued states that treat them as gambling. Sportsbooks have opened their own prediction products. DraftKings and FanDuel have both shipped them. Trump Media’s Truth Social has announced plans. Event Horizon, a newsletter that tracks the industry, has counted more than 70 companies in the pile.

Sen. Richard Blumenthal of Connecticut, who has spent years on Meta and on youth addiction bills, treated Arena as the same business in a new jacket.

Meta copied slot machines to addict kids to Instagram. Now Zuckerberg is turning his company into a prediction market. Meta’s business model is profiting from addiction—kids, gamblers, & more. Stop it through KOSA & my prediction markets bills. https://t.co/e9tG1X8Fho

— Richard Blumenthal (@SenBlumenthal) June 23, 2026

Points let Meta stay out of that pile for now. They do not answer Blumenthal’s charge, and they do not settle whether a 100 million-person market on elections should be scored by Llama. They only delay the day someone at Meta has to file with the Commodity Futures Trading Commission.

Kalshi, Polymarket, and the Wager Meta Skipped

The cash side is not a side story. Pew’s tally of Kalshi and Polymarket, which leaves out Polymarket’s U.S. arm, shows combined monthly global volume rising from less than $5 billion in September 2025 to about $24 billion in April 2026. Sports, politics, and crypto made up 91% of Kalshi’s global volume and 90% of Polymarket’s since July 2024. Sports alone was 80% of Kalshi and 39% of Polymarket over that span. Kalshi closed a $1 billion Series F in May 2026 at a $22 billion valuation.

Those platforms take real questions with real stakes. Kalshi lists a Rotten Tomatoes score on a forthcoming movie. Polymarket lists contracts on whether Israel strikes Iran by a set date. Arena would use the same yes-or-no shape and none of the cash. Zuckerberg has still asked his team to talk to both firms. People familiar with those talks say it is not clear what a partnership would even be: a feed of their odds, a traffic share, or a quiet way to keep a door open if Meta later wants a regulated rail.

HOW ARENA COMPARES WITH THE CASH MARKETS AND WITH FORECAST

Copying the screen and skipping the stake is a familiar Meta move, and it is a weaker product on purpose. Kalshi and Polymarket sell prices that can be wrong with a bankroll attached. Arena would sell the argument around a live game or a vote, then send people back into chats they already open. If the partnership talks go anywhere, the cash firms could even gain a billboard inside that argument. If they go nowhere, Meta still has the distribution those firms do not.

The Social Conversation Is the Payoff

Ime Archibong, the Meta vice president of product leading Arena, put the thesis in a May 2026 internal post introducing the app. He did not talk about fees, order books, or a CFTC license. He talked about a new unit of content, and about showing off.

We believe that prediction markets are one of the more interesting new content types. With the right containers, the social conversation is the payoff as people aim to show off how good they are at predicting things to their friends.

Ime Archibong, Meta vice president of product, internal post

That sentence is why the missing cash is not a half-built Kalshi. It is the design. Forecast already proved that points and a leaderboard do not, on their own, become a daily habit. Llama is there to make the questions cheap. The Family apps are there to drop 18- to 34-year-olds onto a market the way Reels drops them onto a clip. The 100 million target is a conversation target. The open ask to Kalshi and Polymarket is an admission that the actual wager still lives somewhere else.

Employees will test a prototype before any iPhone or Android release. The documents still do not name a ship date. Until one appears, the cash markets keep the bet, and Meta keeps a points app it has already closed once.