The cost of a surprise: what late discovery does to IRR

Fourteen million dollars. That is what Related Digital ended up committing to Saline Township, Michigan, in farmland trust funding, community investment, and fire-department upgrades, to resolve a rezoning fight the public record had already flagged before the company ever filed. A-1 agricultural zoning with no industrial precedent. A five-member township board where two or three votes were plausibly opposed. A parallel utility approval running through Michigan's Public Service Commission on its own separate clock. Every one of those facts was knowable in early 2025. Our own pre-filing read on that site scored it 35 out of 100 before a single rezoning application was submitted.

The board denied the Stargate rezoning 4-1 on September 10, 2025. RD Michigan Property Owner I LLC and the landowners sued two days later. By October 15, a consent judgment locked in the $14 million package plus a 55-decibel noise cap, restricted access, and a ban on high-water-use cooling. In calendar time, that is fast: five weeks from denial to settlement. In cost, it is the whole point of this post. The same $14 million commitment, if built into the original proposal instead of extracted through litigation, buys a friendlier board, a shorter timeline, and no public record of a denial for the state attorney general to later cite when she moved, in February 2026, to reopen the separate DTE power contracts on ratepayer-protection grounds. Five months after the settlement closed, the site was still generating new legal exposure: a resident's motion to intervene, denied February 20; construction proceeding underneath an unresolved utility fight the following month. The surprise from September was still billing the project in the spring.

The same risk, priced early versus discovered late, is not the same number

This is the part every development pro forma underweights: the dollar cost of a given entitlement risk is not fixed. It is a function of when you find it. A risk priced before you file gets negotiated. A risk discovered after you are denied gets litigated, and litigation's real cost is rarely the legal fees. It is the time, and time is where the actual damage happens, because every month a project sits in a contested posture is a month of carrying cost, a month your equity is not earning what your investment committee modeled, and a month closer to a rate lock or a construction bid expiring.

Braintree, Massachusetts is the clean lab version of this, because the risk in question was not a subjective political judgment. It was arithmetic. Braintree calculated its Chapter 40B safe-harbor share at 1.65 percent of qualifying land and denied the 383 Washington Street comprehensive permit on that basis in February 2020. The Housing Appeals Committee recalculated the same figure at 1.39 percent, because a 244-acre conservation parcel belonged in the denominator and Braintree had left it out. Superior Court affirmed the recalculation. The Massachusetts Appeals Court affirmed unanimously on June 18, 2025.

Read that timeline again. A denial in 2020, over a number that was checkable against the same conservation-acreage records in 2020, was not fully resolved until the middle of 2025. Five years. Not five years of active construction delay, five years of a project sitting in appellate limbo because nobody priced the actual denominator before relying on it. Whatever the project's return looked like in an underwriting model built in 2020, run that same model with five extra years of carrying cost, legal fees on both sides, and zero productive construction, and the number is not a rounding error. It is the difference between a deal that clears a fund's hurdle rate and one that does not.

Thirty years is what happens when nobody has to answer for the delay

Saline compressed a lawsuit into five weeks and still left cost exposure running for a year and a half. Braintree took five years to confirm an answer that was available on day one. Morrow County, Oregon shows what happens when there is no forcing mechanism at all, and the gap between discovery and reckoning runs for decades instead of years.

State agencies acknowledged the Lower Umatilla Basin needed cleanup more than 30 years before the Oregon Health Authority formally identified the crisis in 2024: at least 634 domestic wells testing above the federal nitrate limit, some close to ten times the threshold. Amazon settled the resulting suit for $20.5 million in March 2026, denying it caused the contamination and saying it preferred to spend its resources on the community rather than the case. Sixteen other defendants, most with decades of agricultural operations in the same basin, remain in active litigation. Hermiston annexed 800 more acres to attract additional data-center investment in 2025, in the same basin, with the case already filed. Nobody paid the thirty-year bill up front. Somebody is paying it now, in a settlement whose size nobody can actually attribute to any single decade of the accumulation.

What a normal timeline already costs, before anything goes wrong

None of this requires a lawsuit to be expensive. NAHB's most recent regulatory-cost survey puts regulatory costs at $131,734 of the average new home price, 26.4 percent of the sales price, up from $93,870 five years earlier, an increase of more than 40 percent while disposable income rose only 18.3 percent over the same span. Berkeley's Terner Center found development fees ranging from $12,000 per unit in Los Angeles to $75,000 in Fremont for multifamily housing, money committed long before anyone finds out whether the entitlement itself survives.

That is the tax on a project that encounters no surprise at all.

Every one of Saline's five weeks, Braintree's five years, and Morrow County's three decades is additional time stacked on top of a baseline that was never free to begin with. For data-center teams underwriting the current wave of hyperscale sites against a 30,000-jurisdiction map, that stacking is not a footnote. It is the difference between a screening discipline that reads for the surprise before the term sheet and one that discovers it, expensively, on the other side of a denial.

The multiple is bigger than anyone wants to admit

My bet is that the actual multiple between a risk priced before filing and the identical risk discovered through denial and litigation is not the twenty or thirty percent a conservative pro forma might reserve for contingency. It is closer to three to five times the base cost, once you add the concessions a court or a settlement extracts on top of what negotiation would have produced, the legal fees on both sides of a fight nobody had to have, and the months or years of carrying cost that a priced-in risk simply does not generate. Nobody publishes that multiple, because publishing it means admitting how much of the number was avoidable, not unlucky.

The entitlement-to-dollars bridge is not a report you generate once. It is the discipline of asking, before you file, whether the risk in front of you is one you are pricing now or one you are agreeing to discover later, at a markup, in a forum you do not control.