Relocation as an
Energy Strategy

Relocation is the one path to power that no equipment seller can profit from and no local stakeholder wants to hear about, which is why it is almost never priced honestly in either direction. For a minority of capacity-constrained sites it is also the correct answer, and the only way to know is a ledger, not an argument.

Section 01The path nobody is paid to price

The first paper in this series set out eight paths that any large energy user should see priced before capital moves. Seven of them keep the load where it is. The eighth moves the load to where power already exists, and it occupies a strange position in the field: the parties most owners rely on for energy analysis are structurally incapable of recommending it, while the parties most eager to recommend it are compensated for exactly one outcome.

Consider who sits at the table when a constrained site goes looking for answers. The generator dealer, the solar developer, the battery integrator, and the engineering firm attached to any of them earn revenue only if equipment lands at the existing site. The incumbent utility keeps a customer only if the load stays. None of these parties will price the relocation case with enthusiasm, and most will not price it at all. In their studies, the relocation row simply does not appear.

Now consider the other side of the table. Destination states fund economic-development offices whose mandate is recruitment. Site-selection consultants are frequently compensated as a share of the incentives they capture, which means the fee depends on the move happening. Brokers of powered industrial land earn commissions on transactions, not on decisions to stay home. In their materials, relocation is not one path among eight; it is the obvious conclusion awaiting a signature.

Both rooms contain competent professionals. Neither room contains a neutral analysis. The owner who wants one has to construct it deliberately, and this paper is a working method for doing that: what belongs in the relocation ledger, where motivated reasoning hides on each side of it, and a short test to run before anyone books a moving truck.

Section 02Power became a siting criterion

For most of the industrial era, electricity was assumed at the site the way water and roads were assumed. Site selection ran on labor, logistics, land, and tax treatment, with power a line item near the bottom of the matrix. That ordering has inverted for large loads, and the evidence sits in the disclosures of the institutions that deliver power.

Pacific Gas and Electric Company's second-quarter 2026 investor presentation reports more than 12 gigawatts of data-center demand in its service pipeline, against 490 megawatts with executed interconnection construction agreements.¹ Nationally, the Department of Energy's data-center energy report, prepared by Lawrence Berkeley National Laboratory, estimates that data centers consumed about 4.4 percent of U.S. electricity in 2023 and projects a range of 6.7 to 12 percent by 2028.² Demand for delivered capacity is compounding faster than the machinery that delivers it, and not only for compute: electrification of industrial process heat and transport is adding load to the same queues.

When capacity becomes scarce, it starts to price. Industrial land is now marketed with documented power availability as a headline attribute, and parcels that hold it change hands at premiums over otherwise comparable ground, because the premium buys schedule rather than dirt. This is the powered-land market, and its existence reframes the relocation question. An owner is no longer comparing two pieces of geography; the owner is comparing a site that has energy against a site that has everything else.

California sharpens the question with price. In the U.S. Energy Information Administration's most recent monthly data, the average industrial retail electricity price in California was 20.20 cents per kilowatt-hour against a national average of 8.71 cents.³ A standing gap of roughly 2.3 times is a permanent invitation to run relocation math, and boards increasingly accept the invitation. The purpose of this paper is not to argue with the impulse. It is to make sure the math that follows is complete.

Section 03The destination audit

The first discipline of honest relocation math is applying the same evidentiary standard to the destination that you apply to the origin. The reason a move is on the table is that the origin utility gave you a date you cannot live with. The destination deserves the identical question, in writing, before anything else is modeled.

The numbers explain why. The Electric Reliability Council of Texas reported in April 2026 that its large-load interconnection queue had reached 445.8 gigawatts of applications targeting service by 2033. Of that total, 321 gigawatts had submitted no studies, 93.7 gigawatts were under review, and 5.9 gigawatts of large-load capacity was observed energized, with another 3.2 gigawatts approved to energize but not yet operating.⁴ A queue in which roughly three-quarters of the applications have not yet submitted studies contains a great deal of duplication and speculation, and it will thin. But the thinning is the point: the market most often named as the frictionless alternative is processing a connection queue of unprecedented depth, and a relocating project enters at the back of it unless it buys a position that already exists.

Nor is this a Texas peculiarity. Federal regulators opened targeted proceedings in 2026 to speed the integration of large loads onto the bulk power system nationally, which is what regulators do when a constraint is general rather than local.⁵ What genuinely differs across markets is not the existence of a queue but its depth, the process design around it, tariff levels, fuel access, land, water, and air-permitting treatment. Those differences are real and sometimes decisive. They are also facts with dates, not reputations, and they must be established for the specific destination and the specific year, because the conditions that made a market famous for abundance are the same conditions now filling its queue.

Two rules keep the destination audit honest. First, screen on averages but decide on specifics: the state-average tariff that started the conversation is not the rate your load will pay, and the connection cost that matters is the marginal one your project triggers, including any network upgrades assigned to it. Second, treat purchased time as what it is. When a powered site changes hands at a premium, the premium is the market's price for schedule. Paying it can be entirely rational. It should simply enter the model labeled as the cost of time, compared against what the same money buys at the origin.

Section 04The full relocation ledger

Energy-only comparisons flatter the move. A gap between 20 cents and 9 cents per kilowatt-hour, multiplied across a large load and a decade, produces a number big enough to end meetings. The ledger below is what stands between that number and a good decision. Every line is real, every line has been the deciding factor somewhere, and every line hides its own characteristic form of motivated reasoning.

Two lines deserve particular attention because they are the most frequently mispriced. The people line is dominated by attrition: some share of the workforce will not move, and for specialized operations the replacement cost includes recruiting, training, a productivity ramp measured in quarters, and the loss of knowledge that never made it into a manual. These figures are estimates and should be labeled as estimates, but an unpriced line is not a zero. In skilled continuous-process operations, the people line alone can rival the energy savings that motivated the study.

The time line is the second. A relocation is itself a capital project with its own entitlement, design, construction, fit-out, and commissioning schedule, and for regulated products it adds requalification of the new plant. Meanwhile the origin's constraint does not pause while the destination is built. Honest analysis models both clocks, and it frequently finds that the move delivers power later than the best on-site path would have, which is a strange outcome for a decision that began as an escape from waiting.

A relocation analysis that prices the electrons and ignores the people has not priced the move. It has priced the brochure.

Incentives elsewhere are real, and the discipline is to count them the way an auditor would. Texas, for example, exempts state sales and use tax on qualifying equipment for certified data-center projects that meet statutory thresholds, at the entry tier at least 20 qualifying jobs and 200 million dollars of capital investment over five years, for periods running ten to twenty years depending on the scale of the commitment.⁶ That is an enacted program with published terms, and it belongs in the model at written value. The counting rule is categorical: statutory programs and signed agreements enter the ledger net of the obligations they carry, because job commitments, investment thresholds, and clawback provisions are liabilities on the same page. A recruitment presentation, a term sheet under negotiation, or an incentive that requires future legislative action enters at zero until it changes category.

Federal tax treatment deserves one clarifying sentence, because siting models routinely abuse it. Under current federal law as of August 2026, the investment tax credit for qualifying clean-energy property is 30 percent; statutory adders exist but are conditional, and each must be individually qualified rather than assumed.⁷ A model that books unqualified adders on either end of the move is padding its own conclusion. Qualified tax counsel belongs in the loop before any credit line becomes load-bearing.

Section 05The split-load middle

Framed as all-or-nothing, relocation is usually rejected, and usually correctly: the full ledger is heavy, and the position line, the customers, talent, and ecosystem that made the origin valuable, tends to outweigh any single year of energy savings. But all-or-nothing is a false frame. The strongest branch in many real analyses is the split: site the growth where power exists and keep the base where the people are.

The split has several shapes. A manufacturer that cannot get capacity for an expansion can hold its existing lines and place the new line in a powered market. An operator of compute can put the next hall where a written energization date exists while the origin campus continues at its current service level. A processor can move the power-intensive step and keep the labor-intensive steps, accepting freight between them. Each shape converts an existential question into a portfolio question, which is almost always a better question to be answering.

The split is not free, and pricing it honestly means charging it for duplicated overhead, two management structures, inter-site logistics, and the loss of whatever scale economies held the operation together. It also competes directly with the best on-site alternatives at the origin, and it must beat the best of them, not the worst. On-site generation deserves the same neutral treatment here as everywhere in this series. Reciprocating engines are typically the cheapest and fastest machines to deploy, and they face demanding combustion permitting in California's stricter air districts. Turbines reward sites with genuine thermal demand, and they draw scrutiny on efficiency at small scale and part load. Fuel cells permit lightly in strict districts and run quietly at high electrical efficiency, at higher capital cost, with fuel-price exposure and stack-replacement schedules in the lifecycle math. Linear generators are modular and fuel-flexible, with the diligence burden shifted to fleet operating history. Solar with storage buys no fuel, and it rarely carries a continuous load alone. A relocation case that beats a strawman of staying, grid-or-nothing with no alternatives priced, has proven nothing. It has to beat the best stay-and-solve portfolio the origin site can support.

Section 06Running the comparison clean

The method is the one this series applies to every path: requirements first, then evidence, then arithmetic, then a decision memo that names the conditions under which the answer flips. Requirements first means the deadline, the measured load shape, and the growth case are written down before any destination enters the room, because a destination chosen first will quietly rewrite the requirements to fit itself.

Then both ends are priced on documents. The origin's constraint is a dated, written utility answer, not a recollection of a phone call. The origin's alternatives, flexible service, phased service, bridging, and on-site generation, are priced at their best credible configuration. The destination's power answer is an application-specific schedule and cost allocation in writing, not an average or an advertisement. The full ledger is assembled with every line either priced from a source or labeled as an estimate, and the sensitivity table names the variables that actually move the answer. In relocation cases there are usually three: the origin energization date, the destination energization date, and the people line.

Before capital moves in either direction, six questions.

1. Is the problem you are fleeing a documented fact?A dated, written capacity and timeline answer from the origin utility. A stale phone call is a mood, not a relocation case.
2. Is the destination's power answer written, dated, and specific to your application?A schedule and cost allocation for your load, from the destination utility or the seller of a powered site. Economic-development marketing is not a queue position.
3. Does the ledger carry all eight lines, priced or labeled?Energy, people, logistics, entitlements, incentives, time, exit, position. An unpriced line is not zero, and an estimate dressed as a sourced figure is worse than either.
4. Are incentives counted as enacted commitments, net of their obligations?Written statutory terms and signed agreements only, with job thresholds and clawbacks carried as liabilities, and tax counsel qualifying every credit line before it bears weight.
5. Was a split-load case priced alongside full relocation and full stay?Moving the increment while keeping the base is often the strongest branch. Its absence usually marks an analysis built to reach a predetermined end.
6. How is the analyst compensated if the load moves?A fee contingent on the move argues for moving; an equipment margin at the origin argues for staying. Neither is neutral, and the question applies to every party in the room, including us.

A comparison that survives all six can still be wrong; siting decisions carry irreducible uncertainty. It will at least be wrong on the merits, with the reasoning on paper, rather than wrong because the only analysis in the room was written by the party that profits from its conclusion.

Section 07The row that proves the study

Relocation earns a permanent row in every constrained site's analysis for one reason above all: it is the row no seller can afford to include. Its presence, priced with the same rigor as everything else, is the fastest way to tell a decision document from a sales document. Its absence tells you what you are holding.

The row cuts in both directions. We are a California firm, and a meaningful share of relocation analyses will conclude that staying and solving power on-site wins once the full ledger is visible, because the people, exit, and position lines are heavier than energy prices for most operations that have been anywhere for a decade. Some will conclude the opposite. The owner is entitled to the number either way, from a party with no stake in which way it falls. That is the entire method of this series: all the paths, sourced numbers, dated facts, one decision.