SemiAnalysis tracked every enacted local data-center moratorium in the US and found that the instrument that looks like a wall is, in most cases, a detour sign.
SemiAnalysis Tracking — Ep. 033
300+
Enacted local moratoriums across the US (SemiAnalysis)
In episode 033 of its programme — titled “300 Data Center Bans, 3 Projects Delayed: Moratoriums Explained” — SemiAnalysis set its own map of enacted local data-center moratoriums against the projects those instruments have actually stopped. The accompanying newsletter piece carries the headline “Everyone Says Datacenter Moratoriums Are Killing the US Buildout. We Mapped All 300 of Them.” The finding embedded in both titles is the finding worth examining: the count of instruments and the count of effects are not the same number, and they are not close.
Host Jordan Nanos stated the gap directly: “there’s 300 plus total moratoriums across the US… that are enacted as we speak… they’re affecting three projects that you can find. All of this stuff on the chart is actively affecting three projects.” The screening that produced that result was described by Maya Barkin: “74 of the 77 fell out and only three remained.” Those three, as identified on the programme, are an AWSproject in Ohio, the North Point campus in Pennsylvania, and a CoreSite expansion in Colorado. Nothing beyond their appearance on SemiAnalysis’s list is established or claimed here about any of those projects’ status, timing, or prospects.
On the question of how some sites sidestep the objection altogether, Jeremie Eliahou Ontiveros offered a structural argument: “if you do offgrid and you bring your own power and you don’t sort of interact at all with the system, then there’s no risk whatsoever that power prices go up because of you.” Ontiveros also referred to a letter from the governor of Texas and to the residential-price objection raised against grid-connected projects; no contents, date, legal effect or scope of any state action is described here, and no official is named.
The key insight: A moratorium binds a jurisdiction. It binds a project only if that project must be inside that jurisdiction. The instrument is real and locally effective — but the property doing the work is the substitutability of the site, and that substitutability is what the screening reveals.
Being inside a restricted boundary is a necessary condition for being stopped. It is nowhere near a sufficient one, which is why a count of instruments and a count of effects come apart.
The Structural Read
The gap between 300-plus and three is not a measurement error. It is the gap between two legitimately different quantities: instruments enacted and effects produced. SemiAnalysis’s screening sequence makes that gap legible by working through it step by step. The 300-plus figure counts enacted instruments — each one real, each one locally effective on the day it passes. The 77 figure counts campuses sitting inside at least one restricted jurisdiction. The three figure counts projects for which sitting inside that boundary was not just a necessary condition but, after full screening, a sufficient one to establish active effect.
Sitting inside a restricted boundary is necessary to be affected by a restriction. It is nowhere near sufficient. Everything between necessary and sufficient — whether the boundary covers the relevant parcel, whether the project needed the specific approval in question, whether the restriction reaches work already permitted — is where seventy-four of the seventy-seven went. That is not a quirk of this dataset. It is a general property of local prohibitions applied to substitutable sites.
A data-center campus needs power, connectivity, land, and water on terms it can accept. It does not need a specific county. Compare this to a mine, which needs the ore body, or a port, which needs the water. Where the required inputs exist at only one location, a local prohibition at that location is decisive. Where a broad set of locations can satisfy the required conditions, the same prohibition redirects rather than prevents. No claim is made here that any project moved, relocated, or was abandoned — none of that is established. The observation is structural: the substitutability of the site is the property that determines whether a local instrument becomes a binding constraint or a route-change.
Permission Layer — Business Engineer Framework
Consent vs. Financing: Two Different Constraint Structures
Ontiveros’s argument is that the objection raised against grid-connected projects concerns residential power prices, and that a project generating its own supply and not interacting with the grid leaves that objection nothing to attach to. The structural reading is that this converts a political constraint into a capital one. A developer who must be permitted needs consent — granted by parties who do not answer to it and can withdraw it. A developer who brings its own generation needs financing — a different kind of problem, answerable to different parties. Which path is easier, cheaper, or more likely is not assessed here, and no cost or prediction appears.
Three Implications
IMPLICATION 1 — Instruments Measure Political Activity, Not Effect
An instrument announces itself on the day it passes. Its effect becomes visible only later, and only if somebody goes to look for it. That asymmetry is structural rather than anybody’s fault. It is why counts of enacted instruments and counts of affected projects drift apart in every domain where a rule is easy to enact and its consequences are diffuse. This is an observation about what each count measures — not a criticism of any publication, reporter, or outlet. The counting lesson travels well beyond data centers.
IMPLICATION 2 — The Screening Gap Is the Analytical Work
The value in SemiAnalysis’s exercise is not the 300-plus figure — that figure is the starting point. The value is the screening sequence that reduces it: from enacted instruments, to campuses inside restricted boundaries, to projects for which that boundary constitutes an active, material constraint. Each step applies a tighter filter. The gap between each pair of numbers is where the analytical work lives. For any regulatory instrument applied to substitutable sites, performing that sequence is the prerequisite for understanding whether the rule bites.
IMPLICATION 3 — The Interconnection Is the Chokepoint, Not the Zoning
Ontiveros’s framing points toward a deeper structural observation: the political objection to grid-connected data centers is specific, and it attaches to a specific technical arrangement. Removing that arrangement — by bringing generation behind the meter — removes the attachment point for the objection. That is a different problem than navigating zoning, answerable to different parties and solvable with different tools. Whether that path is preferable is not assessed here. What is clear is that the constraint structure changes when the technical architecture changes.
Business Engineer Framework
The Permission Layer
The Permission Layer framework maps how regulatory and political consent functions as a layer in any infrastructure stack — who grants it, who can withdraw it, and what happens when a developer removes the technical dependency that made that consent necessary in the first place. SemiAnalysis’s moratorium mapping is a precise illustration of the layer in action: the instrument is real, the layer is real, and the substitutability of the site is what determines whether the layer is load-bearing. Understand which of your inputs are substitutable and which are not — that is the question the Permission Layer forces.
SemiAnalysis’s exercise in Ep. 033 is a methodological demonstration as much as a regulatory one: start with instruments, screen for actual contact, screen again for binding effect, and report what remains. What remained, by SemiAnalysis’s tracking, was three — an AWSproject in Ohio, the North Point campus in Pennsylvania, and a CoreSite expansion in Colorado — out of 300-plus enacted moratoriums. The distance between those two numbers is not the story of moratoriums failing; it is the story of substitutability doing its work quietly, before anyone has to fight a zoning board. The instrument is real. The chokepoint is elsewhere.
This is businessanalysis. It is not investment advice, no view is expressed on any security, and no recommendation is made.
The count of enacted local moratoriums, the number of campuses inside restricted jurisdictions and the number of projects affected are all SemiAnalysis’s own tracking figures as stated on its programme, and none is independently verified here. The three named projects appear on SemiAnalysis’s list; nothing above claims anything about their status, timing, cost or prospects beyond that, and nothing above claims that any project moved, was relocated or was abandoned. Where a letter from the governor of Texas is referred to, that reference is reported only as something said on the programme: no contents, date, legal effect or scope of any state action is described, no official is named, and no government’s position is characterised. Nothing above takes a position on whether moratoriums are good, bad, justified or effective, or characterises anyone’s motives. The observation that instrument counts and effect counts measure different things is not a criticism of any publication, reporter or outlet, and nothing above claims that coverage of this subject has been wrong, overstated or misleading. No capacity, cost, timeline, delay length or power price appears above, and nothing is predicted. This is businessanalysis. It is not investment advice, no view is expressed on any security, and no recommendation is made.
Gennaro is the creator of FourWeekMBA, which reached about four million business people, comprising C-level executives, investors, analysts, product managers, and aspiring digital entrepreneurs in 2022 alone | He is also Director of Sales for a high-tech scaleup in the AI Industry | In 2012, Gennaro earned an International MBA with emphasis on Corporate Finance and Business Strategy.
Scroll to Top
Discover more from FourWeekMBA
Subscribe now to keep reading and get access to the full archive.
