Kalshi and Polymarket Cross $50B in June as World Cup Drives Bets

The number that actually matters in the New York Times' reporting on this World Cup is not the odds on Sunday's final, it is that two prediction markets most people had never heard of a year ago have quietly become a piece of financial infrastructure. Kalshi posted about $31 billion in total notional trading volume in June, and Polymarket's international exchange set a monthly record of $10.8 billion, with its regulated U.S. arm adding another $3.5 billion. Put together with the wider category, prediction markets cleared roughly $50 billion in June, which is a scale traditional U.S. sportsbooks are not going to touch on this tournament.

The World Cup is doing most of the pulling. Kalshi's World Cup-specific markets alone generated $7.4 billion in June, and the single contract on the Argentina-Spain final has already surpassed $1.27 billion in trading volume, with Spain priced at 61% to beat Lionel Messi's side. Polymarket's World Cup Winner market has generated $4 billion in cumulative volume since it launched in July 2025. These are not toy numbers on toy exchanges anymore.

The reason this belongs in an AI newsletter is the OpenAI piece. ChatGPT Search has quietly started displaying Kalshi's World Cup odds inside its answer cards, with a 'Source: Kalshi' label but, according to The Next Web's reporting, no outbound links or Kalshi branding, and users cannot actually place bets through ChatGPT. It is OpenAI's first prediction-market integration, and the interesting part is what it implies about the roadmap. If you are going to answer probabilistic questions inside a chat interface, a real-money market is a defensible source of a number, in a way that a model's own guess is not.

The honest caveat is that the reporting does not say whether money changed hands in the OpenAI-Kalshi arrangement, and it does not break out how much of Kalshi's June volume is genuine retail participation versus market-maker and arbitrage flow between platforms. A tournament-driven spike is also not the same thing as durable demand, and the post-final comedown will be the real test of whether $31 billion months become a floor or a peak.

The thing to watch is not Sunday's result. It is whether the odds card that showed up in ChatGPT this month expands to elections, macro contracts, and sports outside the World Cup, because that is the moment prediction-market probabilities become the default answer to 'what are the chances' inside consumer AI, and every product built on top of that answer surface has a new dependency.