Franklin Township: how a billion-dollar data center died in the public record

In March 2025, a company called Deep Meadow Ventures LLC filed a rezoning petition for 467.66 acres of farmland in Franklin Township, on the southeast edge of Indianapolis. The filing — petition 2025-CZN-814 — asked to move the parcels from agricultural and small-commercial designations (D-A, C-4, SU-43) to a commercial-special classification (C-S). On August 20, the Metropolitan Development Commission approved it 8-1. On September 22, before the matter ever reached a full City-County Council vote, the petitioner withdrew.

The applicant was Google. The project was a data center campus reported at roughly a billion dollars. And the reason it collapsed is the single most useful thing a VP of Development can study right now, because almost none of it was a surprise. The risks that ended this project were in the public record before a single acre was assembled.

This is what we mean when we say entitlement is no longer local knowledge — it is a live intelligence problem. Here is the Franklin Township record, read the way a screening team should have read it.

The structural risk was visible at filing

Before any opposition organized, three facts about the site and the jurisdiction were already established in public data.

First, the zoning. The parcels carried agricultural and small-commercial designations. A data center campus required a full rezoning to C-S — not a variance, not an administrative sign-off, but a discretionary legislative act that opens the door to public testimony and political review.

Second, the overlays. The zoning designations on the target parcels carried Flood Fringe (FF) and Floodway (FW) suffixes, visible in Marion County GIS data and in the petition filing itself. That put stormwater and flooding on the table as a documented regulatory concern — not a talking point an opponent could be accused of inventing, but a condition written into the parcel's own classification.

Third, and most important, the approval pathway had two veto points, not one. In Indianapolis, the Metropolitan Development Commission's approval is not the last word. The City-County Council holds a "call-down" mechanism: a councilor can pull an MDC-approved petition up for a full council vote. That means MDC approval — the gate most developers treat as the finish line — was structurally the middle of the process here, not the end.

A site read built only on these three structural facts would already have flagged Franklin Township as a two-veto-point jurisdiction with flood exposure on a parcel that needed a discretionary rezoning. That is an elevated-risk profile before anyone has spoken at a podium.

The political risk was building in parallel

Structural risk tells you the shape of the obstacle course. Political risk tells you whether anyone is going to run it against you. In Franklin Township, the answer was visible early.

Indiana already had an active, statewide conversation about data centers and utility costs by 2025. Opponents did not have to invent a message. The framing — that hyperscale power demand drives up household electric bills — was already circulating, carried in part by statewide consumer-advocacy organizing. When the Franklin Township fight started, local opponents could plug into a message that already existed.

Locally, the opposition was not one angry hearing. It formed through residents who organized under the banner Protect Franklin Township, and it drew institutional credibility from the longstanding Franklin Township Civic League, whose land-use committee participated in the hearings and opposition materials. That combination — a fresh resident coalition plus an established civic organization — is the signature of an opposition campaign that will sustain itself across multiple meetings rather than burning out after one.

And the district mattered. The seat covering Franklin Township was held by a councilor who publicly opposed the rezoning and used the call-down process to force the full council vote. Once the district councilor was against it, MDC approval stopped being sufficient. The project was headed into the more political forum, and it was headed there with organized residents waiting.

The anonymity strategy backfired

Hyperscale developers routinely assemble land through shell entities to keep prices down and avoid tipping their hand. Deep Meadow Ventures LLC was that strategy in action. Local reporting later described landowners being approached under confidentiality agreements before the buyer's identity became public.

The strategy is standard, and it is also a liability in a jurisdiction where beneficial-ownership transparency has become a live political issue. When the public learned in mid-2025 that Deep Meadow Ventures was Google, the secrecy did not buy goodwill — it fed distrust. The lesson is not that anonymity is always wrong. It is that the anonymity decision carries its own downside risk, and that risk should be priced before the land is assembled, not discovered after.

The two reads, four months apart

A site read is not a single number frozen at filing. It evolves as the public record changes. Franklin Township shows why that matters.

In March 2025, on the structural facts alone — agricultural zoning requiring full rezoning, flood overlays on the parcels, and a council call-down mechanism as a second veto point — the site already read as elevated risk. No organized opposition was yet visible, but the obstacle course was already steep.

By August 2025, the read had hardened dramatically. The structural risk was now joined by an organized resident coalition, the utility-cost framing, a civic league lending credibility, and a district councilor using the call-down. The political path had become hostile. The petitioner read the same room: rather than absorb a public denial at a full council vote, Google withdrew on September 22, and the petition was formally withdrawn in the record on October 1.

A site read in the public record marked Franklin Township for elevated review at filing. By late summer, the same record pointed at withdrawal. Neither read required a single non-public fact.

What this costs when you find out late

By the time the project was withdrawn, the sunk cost was substantial: land assembly across 467.66 acres of farmland, partner-rate attorney fees, shell-company formation, and the time of a senior development team carried across roughly a year — all on a path that the public record had flagged as steep from the start.

That is the entitlement failure mode RealClear is built to interrupt. Not to replace the land-use counsel, the civil engineers, or the utility leads who do the real work — but to read the jurisdiction's public record before the team commits, so the cheap "no," or the expensive "yes, and here is what it will take," arrives before budget and calendar start compounding.

Franklin Township was not an unlucky deal. It was a legible one. The question for every team screening a data center site in a township-governed jurisdiction is simple: would your process have read this record in March, or in October?