Sustainable Gaps LLC · Preconstruction Planning ThesisSG-PFB-2026-08 · Rev 1.1

A substation package priced at $30M today
could reach $38–41M by 2031.

This is a planning forecast, not a quoted price or guaranteed outcome. Current equipment queues, escalation, sequencing constraints, and temporary-power requirements can move cost and schedule exposure into the field long after a bid is complete. This brief models that exposure for a reference project and shows where earlier planning can preserve options before procurement and construction begin.[1][2]

Evidence key

[e]Planning estimate or forecast. Assumptions are stated and the result is not presented as measured fact.
Field observationFirsthand professional experience, clearly separated from published evidence.
Source-backedTraced to published data. Calculated means computed from that record; professional judgment identifies an informed planning assumption.
● / ○Solid chart points show measured history. Hollow points, dashed lines, and shaded regions show projections.

Contents

§1 · What current signals suggest

A field pattern worth testing against current evidence

This work began as a five-question check of a pattern Erik Herring experienced firsthand during the 2021 supply-chain disruption. Sustainable Gaps then compared that experience with current public procurement, energy-sector, and market records. The records do not prove the forecast; they provide evidence for testing it. A sample of the signals collected:

Signal (as collected)Date collected
Sources Sought — Power Generation with Microgrid at Joint Base McGuire-Dix-Lakehurst, NJ2026-07-07
Solicitation — 178th Wing Base Microgrid Construction2026-07-10
$797,068,317 — Weston Solutions — TEMPORARY power generation, Palo Seco plant, PR2026-04-19
$593,899,755 — Weston Solutions — TEMPORARY power generation, San Juan plant, PR2026-04-19
Commercial electricity sales soaring in Virginia, driven by data centers2026-05-05
FERC order — PJM Interconnection / AEP transmission companies2026-06-24

Dates are SG collection dates; the two Weston awards originate from the 2022–2025 FEMA/USACE Puerto Rico mission and remained active records at collection. Read those rows together: $1.39 billion of federally contracted "temporary" power at just two plants[3][4]. Military bases building their own microgrids rather than waiting on the grid. Data-center load bending a state's electricity sales curve. Interconnection dockets churning. Each row is ordinary alone. Together they are the same pattern we watched break projects five years ago — arriving again, at a scale with three more zeros.

§2 · The questions experience forces

Before the data: what a builder who lived 2021 would ask

1 · Who actually owns your delivery date — you, or a factory's order book?

In 2021 I watched finished buildings stand dark because a breaker plant in Mexico ran out of springs testimonial. The GC owned the schedule on paper; a component supplier owned it in fact. If you can't name your transformer's production slot, you don't own your date.

→ answered by the lead-time data, §4

2 · Your bid assumes 12-week electrical gear. Who eats weeks 13 through 160?

Fixed-price doctrine is settled: the contractor eats it, unless an escalation clause says otherwise. In 2021 the firms with clauses negotiated; the firms without them ate 25% quarterly price spikes or went to the claims boards.

→ answered by the price ratchet, §4, and the budget impact, §6

3 · If the grid can't feed the site on time, what is your bridge — and is it priced, permitted, and dated?

The 2021 answer was rented switchgear to win Certificates of Occupancy — improvised, expensive, and it worked. The 2026 version is measured in megawatts. Improvising it at need costs multiples of planning it in precon.

→ answered by the workaround menu, §6

4 · Which schedule is real — the conference-room Gantt, or the one taped up in the trailer?

Every project has two schedules. The baseline assumes the market of five years ago; the field discovers the market of today. The distance between them is measured in change orders.

→ drawn, to scale, in §5

5 · When the "temporary" power arrives, what ever makes it leave?

Our own engine logged $1.39B of temporary generation at two Puerto Rico plants — deployed as emergency measures, extended for years, and ultimately resolved by the utility moving to purchase the "temporary" units. Temporary is a budget class, not a duration. Plan its exit on day one or fund it forever.

→ the discipline test, §6

§3 · How it snuck up before — and the symptoms now

Nobody rang a bell in 2021 either

The last time, the warning signs were all visible in hindsight: plants down, distributors quoting "call us," prices moving between bid and buyout. Nobody assembled them until buildings were standing dark. We wrote the after-action then. The symptom checklist from that report — every box now checked again:

  1. Quote validity collapsing — transformer quotes now hold 30–60 days against 160-week deliveries testimonial — distributor-channel practice, not survey data.
  2. The workaround industry booming before the crisis peaks — data centers went from 2% to 11% of a global rental fleet's revenue in four years.[6]
  3. Buyers changing behavior quietly — utilities pre-buying three years ahead; cities bidding for used equipment.[7][8]
  4. The substitute becoming scarce too — gas turbines, the escape valve, now backlogged 116 GW into 2029.[9]
  5. Escalation language moving from exotic to standard — the #1 downloaded construction-law topic of 2021 — and a top priority every year since.[10]

Five for five. The difference this time: the window is knowable. New US transformer capacity comes online 2027–2028. The advantage belongs to whoever prices reality before then — because after, everyone's done the homework.

§4 · The data — verified, multi-source

The record, measured — and the next five years, projected

Equipment lead times by class, 2021–2030

Weeks from order to delivery · 2027+ projections [e] · tap any point

Distribution transformers Power / substation transformers Largest transformers (LPT/GSU) Medium-voltage switchgear
FORECAST 0100200 20212024202620282030 205 wk (≈4 yrs) 144 wk 30 wk · easing
Series plot constrained top-of-class quotes; the class average for power transformers is 128 weeks. Some cells interpolated between dated survey points. Sources: Wood Mackenzie quarterly surveys, NERC reliability assessments, DOE, trade reporting.[5][11][12] The split matters: small units are easing; the units that energize substations are not.

The price ratchet — equipment prices vs 2019

Computed from federal producer-price data (BLS series PCU335311335311 / PCU335313335313, retrieved 2026-08-10; raw-index figures run above single-quote trade estimates)[2] · tap any point

Transformers Switchgear
0%+40%+80% 2019202120232025 +81% +83%
The planning implication: distribution lead times fell substantially from their 2023 peak while the cited producer-price measures remained elevated. That divergence makes price normalization an assumption to test, not a condition to build into a schedule without review.

The decade, on one line — 2021 → 2031

Measured left of the marker · projected right [e]

YOU ARE HERE 2021springs · CO rentals 2022DPA invoked 2023100-wk peak 2024xAI · NIAC alarm 2025Stargate · tariffs 2026116-GW backlog 20272 US plants online 2028window narrows 2029turbine slots open 2030queues may ease 2031forecast reviewed
Left of the line: the record. Right of it: announced factory dates, sold-out order books, and this brief's calculated expectations — checkable in public, on schedule.

§5 · Conference room vs jobsite trailer

Same project. Two schedules. The gap can become added cost.

The Gantt shifts right — 24-month plan vs queue reality

$30M substation scope, transformer ordered at month 4 from approved drawings · reality per current 160-week quotes DERIVED

THE CONFERENCE ROOM · 24 months 0mo 12mo 24mo 40 Design Procure gear Civil & foundations Install & test Energize · mo 24 THE TRAILER · 40+ months Design Transformer queue — 160 weeks Civil & foundations crews demobilized · overhead burns $125K/mo Re-mob · install · test Energize · mo 42+ Bridge power (if load arrives) · $200–240K/MW-mo
Every bar that shifts right is an operating cost: extended field overhead, re-mobilization, repriced trades — and if the customer's load shows up on the original date, the amber bar triggers. This is the picture that belongs in the precon meeting, not the claims meeting.[5][13]

§6 · The budget impact

Timing changes how project exposure can be managed.

The table compares planning allowances with modeled late-stage exposure for the same reference deliverable. These are scenarios, not supplier quotations, and actual results will depend on project scope, market timing, and procurement strategy.

WorkaroundPriced in preconArriving as a change order
Early equipment buy (owner-furnished, slot reserved)Carry cost ~$0.5–0.9M [e]Re-quote +8–12%/yr + queue reset: $1.1–1.6M+
Re-sequenced phases (civil proceeds; staged energization)~1–2% of scope, designed inRe-mob + idle overhead: $1.8–3.4M
Bridge power (dated, permitted, exit-planned)Priced option; runs only if neededLate rental at modeled spot rates: $12–14.4M for 5 MW-yr[13]
Escalation protectionIndex-tied rider, negotiated calmlyClaims, default risk, fear-priced next bid
Total modeled exposure~$0.8–1.5M considered early, plus the dated bridge option (unspent unless triggered)+$3.9–7.4M — or +$15.9–21.8M if late bridge power is required

The same $30M scope, bought in 2026 vs 2031

Cost stack for an identical substation deliverable DERIVED from PPI trajectory + forecast run-rates [e]

$30M $38–41M [e] 20262031 ■ Equipment 13.5 → 17 ■ Labor/civil 12 → 14 ■ Overhead 3 → 4.5 ■ Queue & bridge allowance 1.5 → 4
Reference assumptions: engineered-equipment escalation +4–5%/yr through 2028, moderating afterward without assuming a full price reversal; labor +3.5%/yr; longer baseline schedules; and a queue/bridge allowance. These assumptions should be replaced with project-specific supplier, labor, and schedule evidence before a decision is made.[1][2]

Preparing the customer — the preconstruction conversation

Equipment queues are easier to address before schedule pressure narrows the available choices. A practical kickoff record can state: the current lead-time evidence, the allowance carried in the plan, the dated bridge option if the load arrives early, and the trigger that releases a contingency if market conditions improve. That conversation helps the project team distinguish a known planning decision from a late surprise.

The discipline test rides along: any bridge power carries a purchase order for the permanent fix, a contracted end date, and a cumulative-spend tripwire at ~60–70% of the capital alternative. Our own signal data shows what happens without it — $1.39B of "temporary" generation that ran for years before purchase was the exit.[4]

§7 · The bottom line

Put the exposure into preconstruction planning.

  1. The underlying constraint is supported by current evidence. Twenty-four primary-source sets were reviewed on 2026-08-10. No core claim was refuted, and two claims were narrowed or corrected.[14]
  2. Each workaround carries financial exposure. Earlier review allows the team to compare, price, defer, or reject an option while alternatives remain available.
  3. Customers can discuss the options early or confront them after schedule pressure builds. A documented contingency table makes the assumptions and decision points visible.
  4. Earlier planning may improve bid credibility and protect project options. It does not guarantee an award, supplier slot, final price, or delivery date.
Transformational Approach — the SG precon engagement
Sustainable Gaps compares announced project dates, equipment availability, public procurement activity, and cost indicators to help teams examine preconstruction exposure. The reference model in this brief is refreshed quarterly; a client assessment replaces its assumptions with project-specific records. Next step: review the interactive model · discuss a project · erik.herring@sustainablegaps.com.

References

Evidence and sources

The public sources supporting this planning thesis are linked below where a stable source page or document is available. Sustainable Gaps records the retrieval date used for each model review because public records and market data can change.

  1. Wood Mackenzie, quarterly transformer and T&D supply-chain work, Q2 2025, summarized in the POWER executive summary — lead-time and supply-deficit context.
  2. U.S. Bureau of Labor Statistics, Producer Price Index series PCU335311335311 (power and specialty transformers) and PCU335313335313 (switchgear), via FRED, retrieved Aug 10, 2026 — the price curves computed from the federal series.
  3. Sustainable Gaps Signal Engine, collection records Apr–Jul 2026 — the §1 table rows retained with collection dates; underlying notices are available through SAM.gov Opportunities and USASpending.gov.
  4. FEMA / U.S. Army Corps of Engineers, Puerto Rico temporary emergency power mission records, 2022–2025; Weston Solutions task-order awards for Palo Seco and San Juan, recorded through USASpending.gov.
  5. NERC, 2025 Long-Term Reliability Assessment and earlier LTRA records — supply-chain, interconnection, approval, and infrastructure-pace constraints on load growth.
  6. Aggreko annual results as reported by Power Progress in 2026 — data centers increased from 2% of revenue in 2021 to 11% in 2025.
  7. Duke Energy, quarterly results and investor materials — supply-chain procurement and grid-investment context.
  8. Traverse City Light & Power, used substation transformer RFP; City of Harbor Springs, MI, standalone transformer procurement posted Apr 2024.
  9. GE Vernova, Q3 2025 results filed with the SEC — gas-power backlog and slot-reservation evidence.
  10. AGC of America / ConsensusDocs, material price-escalation guidance — contract and commercial considerations for volatile material pricing.
  11. U.S. Department of Energy, Large Power Transformer Resilience Report to Congress (Jul 2024) — acquisition lead times, resilience, and domestic supply-chain context.
  12. NIAC / CISA, Addressing the Critical Shortage of Power Transformers (Jun 2024).
  13. United Rentals, Guide to Planning Bridge Power Systems for Data Centers; the reference model's bridge-power cost remains an estimate with its fuel and duration assumptions stated separately.
  14. Sustainable Gaps, Grid Supply-Chain Intel Report & Thesis Governance Report (SG-GST-2026-08, Aug 10, 2026) — the two separate evidence-review passes behind this brief; available on request.