How a shelved 96 unit development gets rebuilt until it pencils

A 96 unit deal that died in early 2024 is a good case to take apart. Land tied up, plans 60 percent done, yield on cost 5.4 percent against exit cap assumptions in the low fives. Dead. The usual ending is letting the contract lapse and eating about 300k of predevelopment. The original budget on a deal like that runs 25.7M total, 268k a unit, land at 4.2M, podium parking under four stories of wood, and a unit mix heavy on two bedrooms because the sponsor liked them. Land comes first. A seller who has held at 4.2M for two years with no other buyer is negotiable whether or not he admits it. An offer of 2.6M cash or 2.9M with 1.1M of it carried on a subordinate note for five years, interest only, is the kind of structure that gets taken, and the 2.9M version usually wins. That carry sits subordinate to the construction loan, so the lender needs a standstill agreement to permit it. Expect six weeks and expect it to be the most annoying part of the whole restructure. Parking is the load bearing change. Podium out, surface and tuck under in, which costs 14 units and brings the project to 82. Hard cost per unit falls from about 212k to 174k. Losing 14 units of revenue against saving 38k a unit on 96 units of cost is the entire trade, and it is close enough to be worth running four separate ways. Mix follows from that. Shifting from 55 percent twos to 62 percent ones and studios drops rent per unit, lifts rent per square foot by about 11 percent in a submarket like this one, and raises unit count inside the same building envelope. Where it lands: 19.4M total, 237k a unit, stabilized yield on cost 6.5 percent. Against the same exit cap assumption that is 125 basis points of spread instead of 15. That is a deal. The part that nearly breaks it is the parking ratio. Surface and tuck under gets to 1.15 spaces per unit against a code requirement of 1.5, which means a variance, and neighborhood associations turn out against those. A 4 to 3 vote after a transit contribution and a bike room nobody asked for is a realistic outcome, and a vote the other way collapses the restructure, because the podium coming out is what every other change rests on. The move worth copying is going back to a seller who has had no other offer in two years. A 1.3M price reduction is worth more than every other change combined, and it costs one uncomfortable phone call that sponsors routinely do not want to make. The thing to watch on a project delivering in late 2027 is that the timing depends on the supply pipeline staying thin, and every developer running this restructure is making that same bet.