First thing I'd do is get on a call with the engineering lead to understand exactly what changed and why they need the extra time. Once I have the full picture, I'd loop in my program manager and the stakeholders to make sure everyone is aligned on the situation. I'd put together a summary of the delay and present the options to leadership — things like descoping, slipping the date, or launching without that feature — and let them decide how they want to proceed. Throughout all of this I'd make sure communication is clear and consistent so there are no surprises.
As soon as I hear that number, I'm not scheduling a stakeholder update — I'm getting 30 minutes with the engineering lead to separate the stated ask from the real blocker. Nine weeks of that gap is usually not all critical path. I'd map which of the remaining work is launch-blocking versus what can ship in a fast-follow, then stress-test a minimum viable scope that could still meet the committed date. I'd go to leadership with three options — each with a clear trade-off on scope, date, and customer impact — and a specific recommendation. My recommendation would be the partial launch: ship the core flow on the original date, document the known gaps, and commit to a 3-week follow-on. I own that recommendation. Leadership decides, but I walk in with a position.