The 91% Problem
Kalshi's revenue breakdown tells you exactly what prediction markets are. The industry's survival depends on whether anyone is allowed to say it out loud.
Kalshi cleared $43.1 billion in trading volume in 2025. Of that, 91.1 percent came from sports contracts. Strip the sports out and the remaining volume — the entire commercial footprint of “information finance,” the thesis used to justify an $11 billion valuation and win a federal court battle against the CFTC — would not rank Kalshi among the fifty largest futures exchanges in the world.
The fee revenue tells the same story at a finer resolution. Kalshi earned $263.5 million in fees in 2025. Sports generated $234.6 million — 89 percent. For every month from September through December, a stretch covering the NFL season, college football, and the NBA, sports exceeded 90 percent of fee revenue. The Super Bowl winner contract alone surpassed $150 million in volume, a 6,400 percent increase from $2.3 million one year prior. On Conference Championship Sunday, a single day of NFL parlay fees exceeded the platform's highest-ever non-sports fee day — which was Election Day 2025.
These are not the metrics of an information aggregation platform. They are the metrics of a sportsbook that operates in all 50 states while licensed sportsbooks are limited to 39.
The classification question
The prediction market industry now faces a binary question that will determine whether it survives in its current form: are these platforms financial instruments that happen to cover sports, or sportsbooks that have obtained a financial classification?
The answer carries structural consequences. If financial instruments, the CFTC's jurisdiction holds. Federal preemption overrides state gambling laws. Kalshi, Robinhood, Coinbase, and every other CFTC-registered platform continue operating nationwide, including in California, Texas, and Florida, where sports betting remains illegal. The 18+ age threshold stays, expanding the addressable market beyond the 21+ requirement for traditional sportsbooks. The growth trajectory — potentially $100 billion in 2026 volume — remains intact.
If sportsbooks with a regulatory loophole, then state gaming regulators have standing. The American Gaming Association's argument — that the CFTC is “an agency with no gambling expertise that oversees cattle futures and derivatives, not sportsbook operations and responsible gaming protocols” — becomes difficult to dismiss. The 60 tribes, 34 state regulators, and 9 tribal groups that filed amicus briefs supporting New Jersey's case against Kalshi would have a path forward. The 19 federal lawsuits Kalshi currently faces would carry real weight.
The honest assessment is that both descriptions are simultaneously true. Prediction markets are, structurally, derivatives contracts settled against real-world outcomes. They are also, functionally, a way to bet on football games. The regulatory question is not which description is correct. It is which description wins.
The sportsbook industry has already answered for itself. DraftKings launched DraftKings Predictions in December 2025, available in 38 states, routing through CME Group. FanDuel launched FanDuel Predicts the same month through a CME joint venture, reaching all 50 states by January 15, 2026. Fanatics launched Fanatics Markets through Crypto.com's exchange. These are the three largest sportsbook operators in America, now simultaneously running licensed sportsbooks under state gaming commissions and prediction market platforms under CFTC oversight. DraftKings and FanDuel deliberately avoid offering sports prediction contracts in states where they hold sportsbook licenses — an implicit acknowledgment that the products are substitutes.
Both DraftKings Predictions and FanDuel Predicts route through CME Group, meaning they share an underlying order book. The distinction between a sports bet and an event contract is not architectural. It is regulatory. The sportsbooks know this. They left the American Gaming Association over it. DraftKings, FanDuel, and Fanatics all withdrew from the AGA in late 2025 after disagreements about prediction markets. The largest operators in American sports betting looked at prediction markets and chose to join rather than fight.
The Robinhood dependency
Layered on top of the classification risk is a distribution dependency.
Robinhood drives more than half of Kalshi's total volume. It has 24 million users and a market capitalization exceeding $100 billion — more than DraftKings, FanDuel, Kalshi, and Polymarket combined. It called prediction markets the fastest-growing product in its history, contributing $100 million in annualized revenue.
In January 2026, Robinhood closed its acquisition of MIAXdx, a CFTC-licensed exchange, through a joint venture with Susquehanna International Group. The exchange has been rebranded as Rothera and is expected to go live in Q2 2026. When it does, Robinhood will have the infrastructure to migrate volume off Kalshi and onto its own exchange. The company has said “there will be no changes in the short term.” That sentence has a shelf life.
The arithmetic is uncomfortable. If Robinhood represents more than 50 percent of Kalshi's volume and migrates to Rothera, Kalshi loses its majority volume source. The remaining distribution partners — Coinbase, Webull, and direct users — may not sustain current fee economics. Kalshi's 91 percent sports revenue is flowing predominantly through a partner that is building a competing exchange. The information finance markets are not large enough to sustain the business independently.
The reflexive trap
The 91 percent figure is not just a revenue composition metric. It is the mechanism of a feedback loop the industry has built for itself.
Sports volume funds the infrastructure that makes non-sports markets possible. Kalshi's 3,500 active markets — spanning economics, geopolitics, weather, and culture — exist because the NFL and NBA generate enough fee revenue to fund the exchange technology, the compliance apparatus, the market-making relationships, and the legal defense. The non-sports markets are a public good subsidized by sports betting.
This creates the trap. The more successfully prediction markets function as sportsbooks, the stronger the case that state gambling regulators should have jurisdiction. But if state regulators gain jurisdiction, they impose geographic restrictions, age requirements, and licensing regimes identical to those governing DraftKings and FanDuel. That eliminates the regulatory advantage driving sports volume. Which eliminates the revenue funding non-sports infrastructure. Which kills the information finance use case that justifies the whole enterprise.
The loop runs in both directions. Every additional dollar of NFL volume simultaneously grows the business and strengthens the argument for why the business should be restricted. The industry needs sports revenue to fund information finance infrastructure. It needs the information finance classification to protect sports revenue from state regulators. The entire structure depends on both halves holding — even though the commercial engine is the primary evidence against the regulatory framework.
The CFTC withdrew its proposed ban on sports and political event contracts on January 29, 2026. Chairman Michael Selig directed staff to draft new rulemaking. The Coalition for Prediction Markets — Kalshi, Robinhood, Coinbase, Crypto.com, and Underdog — called it “a key step to foster market clarity.” But the CFTC has one of five commissioner slots filled. The agency has been gutted by broader government cuts and has approximately one-eighth the staff of the SEC, despite Kalshi processing more than $2 billion in trades in a single week. The regulatory body tasked with deciding what these markets are does not have the resources to decide.
Meanwhile, the appeals case between Kalshi and Nevada gambling regulators is set for the 9th Circuit in April 2026. A federal hearing on February 12 will address whether CFTC authority can override state gaming bans.
The 91 percent number will appear in every brief. State regulators will argue that a platform generating 91 percent of its revenue from sports is a sportsbook by definition. Kalshi will argue that the classification of the instrument, not the subject matter, determines jurisdiction. Both arguments are defensible. The outcome will determine whether an industry clearing more than $40 billion annually continues to exist in its current form.
What the 91 percent means
The information finance thesis is not wrong. Polymarket's election markets outperformed every major polling aggregate in 2024. Fed funds rate contracts provide genuine price discovery. Geopolitical event contracts surface probability estimates available nowhere else. The thesis is real. It is also commercially irrelevant to the business as it actually operates.
The 91 percent problem is that the industry's commercial reality and its intellectual justification are misaligned, and the gap between them is the exact space where regulatory risk lives. The prediction market industry in early 2026 is a structure that can only survive if both halves hold: the sports revenue that funds it and the financial classification that protects it. The sports revenue is the primary evidence against the financial classification.
The market's success is the input to its own potential destruction. Whether the loop resolves in the industry's favor depends on courts, regulators, and legislators who are themselves watching the volume numbers to decide what these markets are.