Data-center capital now prices a Senate vote

The data-center trade has a new bid-ask: $25 billion for income, and a Senate vote for everything else. The income half surfaced in exclusive talks over Stack Infrastructure's Asia Pacific portfolio, where the $25 billion bid sits against a build-out ask that is not yet a number; the vote happened elsewhere, as senators moved to strip data centers from the rural Opportunity Zone tier, deciding whether the tripled step-up in basis was a housing benefit instead of the data-center benefit it was sold as. Markets that had priced rural land on a tax-advantaged development envelope are now repricing it on a calendar of committee votes.

The rural tier carried a tripled step-up in basis, an incentive that magnifies the tax-free portion of a long-term exit and was marketed, in part, on the premise that data centers would anchor rural development districts. The Senate is now asking whether the tier was a housing program all along, and rural land comps will register the answer before any guidance does, because an entitlement buyer cannot underwrite a tax benefit that a floor vote can remove. Prices paid for raw acreage in Opportunity Zone counties now move with Senate procedure, not with power line cost.

PWD's tracking caught both moves in the same cycle, and they belong to a single story: one prices existing income, the other prices the land that new income would occupy. GI Partners paid $14.7 million for a south Phoenix data center parcel that clears on a different kind of political clarity, its value set by the substation next door and the city's data center determination. The buyer is not speculating on whether a statehouse will preserve a tax break; it is paying for a site where the municipal box is checked, and the underwriting can start from power, fiber, and cooling rather than a senator's stance.

Rural Opportunity Zone land now behaves like a policy option, while the Phoenix site behaves like a municipal decision. One carries a Senate vote between it and a data center campus; the other carries a substation. Capital has begun to pay for that difference, which is why the $14.7 million matters less as a milestone price for Phoenix land than as the cost of buying a site where the political question has already been settled at the city level and the underwriting can start from the obligation to build rather than the obligation to lobby.

Policy risk lands on land comps before data center values because the asset class has two markets with different clearing speeds. Income-producing portfolios trade on yield and lease term and can absorb a policy shock by adjusting discount rates, while raw land trades on optionality—exactly what a legislative review kills. A tripled tax basis is an option on future taxes; remove it, and the land reverts to its pre-incentive comp, minus the carrying cost the buyer has already paid. That reversion is the market marking the policy's probability.

Blue Owl's $25 billion bid is the income side of the bid-ask, an enormous but legible price for a portfolio of operating data centers in Asia Pacific that produces a stream of rent that can be securitized, levered, or held. The build-out ask is the other side, not a number anyone can print with the same confidence, because building new data centers now means land, power, permits, and a tax code that can change mid-campaign. The spread between the two is where the Senate vote lives; capital that wants the higher number must buy not only megawatts and fiber but a political calendar.

GI Partners' $14.7 million south Phoenix parcel is the market pricing municipal clarity. What separates the site from a rural Opportunity Zone parcel is documentation, not dirt; the buyer paid for the substation next door and the city's data center determination, both of which shorten the path to a building. In a repricing cycle, the site with the shortest political path gets the bid, and that is exactly what the purchase was: an entitlement arbitrage.

The Phoenix trade also shows what capital is leaving behind. A year ago, a buyer looking at rural land could underwrite the Opportunity Zone benefit as a durable feature of the parcel; today the same buyer has to value that benefit at the Senate's discretion. The discount now applied to rural land reflects legislative risk rather than power cost or distance to fiber, which is why the Phoenix site, with no such discount, clears at $14.7 million while rural sites will clear lower or not at all—the difference is the market's estimate of what Washington might do.

The data center capital stack now splits in two: income trades like the Blue Owl negotiations price existing stock, while land trades like the GI Partners purchase price the permission to build. The two no longer use the same discount rate—an income bid can be defended with a lease maturity schedule, but a land bid now has to defend a tax opinion. When that tax opinion moves from a law firm to a Senate floor, the bid-ask widens.

The build-out ask was always difficult to estimate, because power availability, equipment lead times, land entitlements, and construction labor all move—but those are private-sector constraints a sponsor can price with a contract. A Senate vote is different, a public-sector constraint with no counterparty, and a developer cannot hedge it by paying more for the site. If the rural Opportunity Zone tier is stripped for data centers, the price of every eligible parcel reprices immediately, whether or not the current owner intended to use the benefit. That is the new binding constraint, and it comes from the statehouse rather than the dirt.

This shift explains why capital concentrates on sites with political clarity even when they cost more per square foot. GI Partners paid $14.7 million in Phoenix rather than finding cheaper rural acres, because the marginal dollar now buys certainty that the Phoenix parcel delivers but rural Opportunity Zone land cannot. A developer who buys the Phoenix parcel can underwrite the building to completion; a developer who buys the rural parcel must underwrite a floor vote, leaving one with a construction schedule and the other with a lobbying problem.

The market is still quoting data center land on power and fiber when the binding constraint is a Senate calendar. The $25 billion income bid prices existing assets; the build-out ask prices the world the Senate will permit. Until that vote is resolved, rural land comps will trade at a discount while city-approved sites like Phoenix clear at premiums that look irrational only if one ignores the political option embedded in the dirt. The next rural Opportunity Zone land sale in a data center market after the Senate vote will state the market's estimate of the policy's survival, just as the Phoenix closing schedule will state how fast political clarity converts to groundbreakings. They are now the same trade.