DraftKings cuts 2026 earnings forecast by $625m for AI predictions

DraftKings slashes 2026 earnings forecast by $625m to fund predictions push

Jason Robins made a decision on Friday that will cost DraftKings somewhere between $650 million and $700 million before the year is out, and he made it anyway. The Q3 earnings call laid out the arithmetic plainly: a 16% revenue increase, $1.329 billion for the quarter, monthly unique payers up to 4.1 million. A strong report by any conventional measure. Then the guidance revision that swallowed it.

The company cut its full-year 2026 adjusted EBITDA forecast from a range of $700 million to $900 million down to $175 million to $275 million. That is not rounding error. That is a $625 million reduction at the midpoint, nearly all of it deliberately deployed into Predictions. The net loss for the third quarter alone reached $468.7 million, nearly double the same period last year.

Robins' argument on the call was that this is recoverable. He said customer acquisition costs can be earned back even in a scenario where sports event contracts are ruled unenforceable by the Supreme Court in 2028. The logic runs: bring users onto the platform through Predictions, retain them through the core sportsbook and iGaming products, let the cross-sell do the work. CFO Alan Ellingson framed it as deploying free cash flow from an established business at attractive payback periods. That framing requires the core business to stay healthy. For now it does — the sportsbook and iGaming operations are projected to generate $875 million to $925 million in adjusted EBITDA for the full year, which is what funds the bet.

The 2027 guidance tells you what Robins actually believes. DraftKings projects Predictions to contribute $650 million to $750 million in revenue next year, against an additional $250 million to $350 million in investment. The consolidated EBITDA target of $800 million to $1 billion depends on that revenue materializing in a legal environment that remains genuinely unresolved. Five federal circuits have not sided with the CFTC preemption argument. The Supreme Court has not yet agreed to hear the case. State attorneys general are still filing.

The reporting frames this as a bold bet under uncertainty, and structurally that is correct. What I don't think the consensus has fully absorbed is the sequencing problem. DraftKings is spending at scale in 2026 and 2027 on a product whose legal foundation will not be decided before the spend is complete. Robins is not hedging the timing — he is accepting that the legal question resolves after the capital is out the door. That is a different risk posture than "we'll expand once we have clarity." It is expansion as a fait accompli strategy: be large enough by 2028 that the political cost of unwinding the product exceeds the appetite of any court to do it.

I have seen this logic work once and fail twice in adjacent industries. The variable that determined the outcome each time was not the legal question. It was whether the customer base proved sticky before the ruling arrived. Robins has 1.6 million people who have engaged with the Predictions platform since January. Whether that number is large enough to anchor the argument is what 2027 will establish.