DraftKings bets $300m on predictions as core sportsbook stalls

Jason Robins used the word "pivotal" to describe the third quarter. His CFO used the word "investment" to describe what is functionally a $300 million loss. Both are correct, and that is the tension worth examining.

DraftKings reported $1.3 billion in Q3 revenue, up 16% year on year. The core sportsbook is healthy enough — 4.1 million monthly unique payers, average revenue per user at $109. These are real numbers, and they are good numbers. But the guidance Robins released alongside them is where the story actually lives: $7.35 to $7.75 billion projected for 2027, absorbing $300 million in prediction market losses at the midpoint of EBITDA guidance. The core business is expected to generate between $875 million and $925 million in EBITDA next year. The predictions business will consume roughly a third of that before it generates a dollar of net income.

That is a structural bet, not a product launch.

The consensus read on this is that DraftKings is buying market share while the regulatory environment is still unsettled, counting on the CFTC framework to stabilize before the spending curve turns. I don't think that's where this lands. The $300 million figure is not a preemption hedge — it is a signal that Robins believes the prediction market customer is structurally different from the sportsbook customer, and that acquiring them now costs less than acquiring them after the category consolidates. The 1.6 million customers who have engaged with predictions year-to-date are not sportsbook customers who wandered in. They are a separate cohort, and DraftKings is pricing the cost of owning them before Flutter or Sporttrade or a Kalshi-with-a-distribution-deal does.

Flutter's CEO has already called prediction markets a structural threat to FanDuel. He said that publicly, which means his board has known it privately for longer. FanDuel earned $3 million in its last reported quarter. DraftKings is about to spend one hundred times that acquiring prediction market users in a single fiscal year.

I have watched companies absorb losses at this scale twice in my career: once when it worked, and once when the regulatory floor they were counting on turned into a ceiling. The difference was not the size of the bet. It was whether the product margin at stabilization justified the acquisition cost. DraftKings' prediction market gross revenue is being almost entirely offset by promotional spending this fiscal year — the company said so directly. That is not unusual for a launch phase. What would concern me is if the same sentence appears in the Q3 2027 release.

Sporttrade's CFTC registrations, finalized in late October, clarify part of the picture: a DCM and DCO structure exists that can serve institutional and broker-routed flow, which is exactly the infrastructure layer DraftKings needs if its predictions business is going to run at scale without being geofenced out of key states the way Kalshi has been. Whether DraftKings routes through Sporttrade or builds its own regulatory stack is not on the public record. But the infrastructure question is now separable from the legal question, and that changes the investment calculus.

The Kalshi circuit losses keep accumulating — Ninth Circuit, Sixth Circuit, Utah criminal charges still pending in the Tenth — and each one makes the DraftKings model look slightly better by comparison, because DraftKings is building inside the regulatory perimeter rather than fighting to move it.