Meta Weighed Kalshi Buyout Before Building Play-Money 'Arena'

The specifics tell a familiar Silicon Valley story with a twist. Zuckerberg reportedly sat down with Kalshi cofounder and CEO Tarek Mansour last year to talk about buying the prediction-market startup outright, but the conversation stayed at an informal stage and never became a deal, NPR reports. The two accounts of why it fell apart don't quite match: some sources told NPR that Mansour wasn't a willing seller, while others said Zuckerberg looked at the legal and regulatory battles Kalshi is contending with and decided the mess wasn't worth it.

Instead, Meta built its own thing. Internal documents reviewed by NPR describe a Meta app called Arena where users guess at future news events and trending topics but wager only play money, not the real cash Kalshi and its main competitor Polymarket take. That distinction is the whole point. It sidesteps the legal fights that have shadowed the real-money side of the market while still borrowing the engagement mechanic that has fueled the category's growth.

And that growth is why any of this matters. NPR reports that in June 2025 the two big real-money venues were handling about $28 billion in monthly volume between them; a year later that figure is nearly $220 billion a month, driven mostly by sports-related betting. Kalshi's private valuation has moved in sympathy, from roughly $2 billion in mid-2025 to about $22 billion by March 2026, according to a casino.org summary of the funding rounds. Meta and Kalshi also struck a partnership in March that put Kalshi markets inside Threads, so the two are not straightforward rivals.

The honest caveat is that the reporting on why the talks died is single-sourced on either side, and neither piece gives an Arena launch date; treat the ethics-concerns version as one account, not a settled fact. What the reporting doesn't cover is how Meta plans to monetize play-money predictions, what happens to the existing Threads integration once Arena ships, or whether regulators will treat a no-cash prediction app at Meta's scale as a gambling on-ramp anyway. The forward-looking thing worth watching is whether play money at Threads distribution ends up training the audience for real-money prediction markets, and who owns the funnel when it does.