How do you underwrite a 2029 delivery when rents that far out cannot be seen?

A 25 basis point spread between development yield and exit cap is the part that breaks first, and it breaks before rents do. At 5.0 untrended yield on cost against a 5.25 exit you have essentially no value creation cushion, so every dollar of cost overrun and every month of delay comes straight out of equity rather than out of margin. Run it at a 5.75 exit and see what happens to your multiple. If the deal only clears at 5.25, you're making a bet on cap rates, not on building apartments.

Second failure point is the debt, specifically the gap between construction loan maturity and stabilization. Your 24 plus 12 assumes leasing goes as planned. If concessions in your submarket are still running one to two months free when you deliver, your effective rents lag face rents and your stabilized NOI arrives late, which is exactly when the loan comes due. Extension options usually carry fees, a rebalancing test, and often a debt yield hurdle you have to hit to exercise. Read the exact test language and model the case where you miss it, because that's the scenario where sponsors lose control of an otherwise fine asset. Confirm current terms in writing with your lender rather than relying on term sheet summaries.

On the supply argument, your logic matches what completions and starts data suggests, but pipeline can restart faster than a hard cost pullback would imply if capital reprices. Also check what's entitled and shovel ready but paused in your submarket, since those projects can start on 90 days notice when rents move.

The input I'd stress hardest is not rent growth. It's insurance and property taxes at stabilization, because reassessment on a newly built 180 unit asset can land well above your pro forma and it never comes back down.