CME's Kalshi lawsuit is also a fight over who owns prediction-market AI training data

On June 9, Kalshi's new crypto perpetual-futures product crossed $1 billion in trading volume inside its first week, according to CNBC. Nine days later, CME Group sued the U.S. Commodity Futures Trading Commission in federal court in Washington to block that product from existing at all (Reuters). The wire has framed the suit as a derivatives turf war. The underlying dispute, however, runs through a different asset entirely: the outcome-tagged dataset Kalshi produces every time one of its event contracts resolves, and the AI laboratories that founder Tarek Mansour says are already ingesting it.

CME's complaint, filed June 18 in the U.S. District Court for the District of Columbia (docket recap here), asks a court to set aside the CFTC order 26-601 that lets Kalshi and Coinbase offer perpetual futures to U.S. retail customers. CME argues the agency exceeded its authority. Dechert's OnPoint addendum and DeSilva's analysis treat the case as a question about whether event-contract platforms can sit adjacent to, or inside, the traditional derivatives perimeter.

Kalshi's growth story this year, in Mansour's telling on Podcast Alpha, is roughly 25 times larger than 2025, driven mostly by the crypto perpetuals launch. The same conversation introduces a separate, larger claim: that Kalshi's prediction-market order book is the only large, outcome-verified, forward-looking dataset of its kind, and that Kalshi is currently providing it to AI laboratories at no charge. Mansour calls it "data nobody has" on the show, and a Coindoo feature frames the perpetuals launch as a challenge to Wall Street's crypto derivatives franchise.

Mansour's claim is founder-attributed, not an established market fact. Whether Kalshi's outcome data is genuinely unique, whether AI labs would pay a premium for it once alternatives exist, and whether CME's suit would change any of that, are all open questions.

The data angle is the second-order effect of the suit. A Kalshi win in D.C. would establish two things at once: that event-contract platforms can offer perpetuals under CFTC oversight, and that the dataset generated inside those platforms is a paid AI-training input rather than a free one. A CME win pulls that data asset back inside the traditional derivatives stack, where CME already controls the order flow, the licensing, and the API surface that AI labs currently consume for futures-market signals.

Mansour adds two adjacent claims on the same podcast that the lawsuit, if anything, makes more legible. First, more than 80% of Kalshi's active users, he says, consume prediction probabilities as a read-only news feed rather than as traders. Second, the single most accurate inflation forecaster on Kalshi's platform is an unbranded individual based in Kansas. Both are founder-attributed and not independently verified. They matter because, taken together, they describe a market whose economic value lives in the signal layer, not the trading fees, and whose signal is being generated by users who may never place a trade.

How that value gets priced now depends on the CFTC's next move. The agency has not signaled whether it will defend the order or settle. CME has incentive to settle only on terms that either preserve its existing crypto derivatives franchise or force Kalshi into a registered-futures structure where CME retains connectivity. Kalshi's incentive is the opposite: keep the event-contract wrapper, keep the dataset inside it, and turn the AI-training-license question into a separate revenue line.

The next concrete trigger is the court's scheduling order on the government's response to CME's complaint. A preliminary injunction hearing, if granted, would compress the timeline from months to weeks and force Kalshi to defend the perpetuals product while the merits are still being briefed. The longer the order stands, the more AI labs are likely to treat Kalshi's data feed as a default input rather than a negotiable one.