In February 2025, OpenAI and Oracle announced Port Washington, Wisconsin as a Stargate data-center site. Fourteen months later, we pulled Redfin's monthly data for the city and three 7-to-15-mile-away neighbors. Home prices went up everywhere. The city at the site gave up the most at the negotiating table.
The sale-to-list ratio is the final sale price divided by the home's original list price. If a home lists at $350,000 and sells for $357,000, the ratio is 102% — the buyer bid above asking and the seller had the leverage. If that same home lists at $350,000 and sells for $340,000, the ratio is 97% — the seller accepted a concession and the buyer had the leverage.
Above 100% is a seller's market. Below 100% is a buyer's market. It's the clearest single measure of who holds negotiating power in a local housing market — and it can shift even when absolute sale prices are still rising.
The Y-axis is the sale-to-list ratio by month. The dashed line at 100% is the break-even. Above, sellers are winning; below, buyers are winning. The orange vertical line marks when OpenAI and Oracle announced Stargate in February 2025. The red line marks when Vantage broke ground in December 2025. The green line marks April 7, 2026 — when Port Washington voters passed the first anti-data-center ballot measure in US history with 66.4% support.
Port Washington is the bold red line. The three grey lines (Grafton, Cedarburg, Mequon) are the control cities in the same county, 7 to 15 miles away.
Something changed in Port Washington after the Stargate announcement that did not change in the three nearest cities in the same county. Inventory rose 36%. Sellers went from getting above-asking prices in 2024 to giving concessions in 2026. November 2025 was the worst single month in the eleven years of available data for the city — 45 active listings, 5 homes sold, 95.1% sale-to-list. A month later, Vantage broke ground. Four months after that, 66.4% of voters passed the first anti-data-center ballot measure in US history.
Absolute sale prices kept rising — up 28% in Port Washington from 2024 to early 2026, roughly in line with regional appreciation. Homeowners who sold gained with the market. What shifted was the concession sellers made at closing: more of what they asked for, less of what they got.
Meta is building a 1,500-acre, 1 GW campus called Project Domino in Lebanon's LEAP Research and Innovation District — roughly the same scale as Stargate. We ran the same check against three Indianapolis-metro controls (Brownsburg, Whitestown, Zionsville).
Lebanon's sale-to-list ratio did not move. 98.0% in 2024, 98.0% through the post-announcement period, 99.0% in early 2026. Port Washington dropped 2.8 points. Lebanon did not drop at all.
But Lebanon's inventory jumped +75% — from 26.8 homes on the market in 2024 to 47.0 post-announcement. That's a bigger inventory shift than Port Washington. And Lebanon's median price held up at +2.4% while the three control cities fell 6-16%.
Our working interpretation: the same underlying behavior may be showing up in both cities, just in different metrics. The sale-to-list ratio is most sensitive when a market was previously tight. Lebanon was already a buyer's market at 98% in 2024 — there was no further downside for sale-to-list to register. Inventory is the cleaner signal in soft markets.
There are a number of factors it could be, and we are continuing to research. Other plausible explanations: Lebanon's LEAP district was a state-designated innovation zone, broadcasting industrial use in advance and reducing announcement shock. Lebanon has no ballot measure, no organized ballot campaign, and less national news coverage than “Stargate” got. Lebanon's housing market is deeper and more liquid than Port Washington's, which dampens signal in any one metric. The Indianapolis metro has been a stronger housing market than the Milwaukee metro, absorbing some of the pressure.
The point is not that every data center announcement moves every metric the same way. The point is that the data to run this check is already public, the workflow is cheap, and the result may tell you something worth acting on. This sort of study may be helpful in your data center fight. If you live near a proposed project, send us your zip code and the announcement date — we'll send back the same table we built for Port Washington, and you can bring it to your next planning commission meeting.
For the complete methodology, data caveats, and 1,581-parcel residential-exposure analysis, read the full story on our site.
Raw data available on request: george@poweredbywho.com.