DC Escalation Calculator escalate your own base cost by the DC Build index — and see exactly which packages moved it
Drafting a contract clause? The price-adjustment clause kit settles on a single official BLS series, with deadband, share, cap/floor and a named vintage.
your own $/MW, or the whole project — the math is a ratio, so the unit is yours
Escalated to 2026-08
$10,390,762
from $9,000,000 in 2024-08 · DC Build index
Total escalation
+15.5%
24 months · index 147.7754 → 170.6110
Annualized rate
7.45%/yr
compound, over the window you chose
Escalation dollars
$1,390,762
the delta the bridge below decomposes
What drove it rows rounded to 2dp for display — compare TOTAL to Headline below
Contribution is priced against the headline's base index (the KPI card above) — not each component's own. So weight × "Its own escalation" will not reproduce Contribution, except where every component happens to start at 100 — the Index column below lets you verify Contribution directly: 100 × weight × (end − base) ÷ headline base index. That is a different formula from
contribution_pp on /datacenter (weight × the component's own YoY) — don't carry that shortcut over here.| Component | Weight | Its own escalation | Index (base → end) | Contribution | Of your delta |
|---|---|---|---|---|---|
| Switchgear & switchboard | 14.0% | +23.8% | 155.9112 → 193.0224 | +3.52pp | $316,427 |
| Steel mill products | 6.5% | +38.5% | 147.0513 → 203.6122 | +2.49pp | $223,908 |
| Copper wire & cable | 5.5% | +47.1% | 140.2439 → 206.3461 | +2.46pp | $221,421 |
| Power & distribution transformers | 12.0% | +10.3% | 180.9807 → 199.6751 | +1.52pp | $136,626 |
| Aluminum mill shapes | 3.0% | +56.3% | 125.4619 → 196.0408 | +1.43pp | $128,955 |
| Construction wages | 15.0% | +8.6% | 130.4969 → 141.7291 | +1.14pp | $102,612 |
| AC & refrigeration equipment | 10.0% | +7.6% | 159.6304 → 171.7715 | +0.82pp | $73,943 |
| Generator sets & turbines | 9.0% | +7.6% | 132.6215 → 142.7496 | +0.62pp | $55,515 |
| Industrial pumps | 5.0% | +11.3% | 145.0640 → 161.3976 | +0.55pp | $49,738 |
| Electrical contractors | 9.0% | +5.7% | 142.2914 → 150.4496 | +0.50pp | $44,717 |
| Plumbing/HVAC contractors | 6.0% | +5.2% | 140.0441 → 147.3084 | +0.29pp | $26,545 |
| Ready-mix concrete | 5.0% | +2.3% | 145.1786 → 148.5793 | +0.12pp | $10,356 |
| TOTAL | 100.0% | — | — | +15.46pp | — |
| Headline | — | — | 147.7754 → 170.6110 | +15.5% | — |
TOTAL adds up the rows above as printed (each rounded to 2dp); Headline is the true figure, rounded to 1dp. The gap between the two is that rounding — nothing is missing; the underlying, unrounded numbers already sum with no residual.
What you could carry realized regimes, measured to 2026-08 — not a forecast
| Basis | Window | Annualized | Cumulative |
|---|---|---|---|
| Long-run every month in the sample | 2007-12 → 2026-08 (224mo) | +3.8%/yr | +99.6% |
| Downturn regime (GFC) the post-crisis construction downturn | 2008-12 → 2011-12 (36mo) | +2.5%/yr | +7.6% |
| Trailing 3yr the last three complete years | 2023-08 → 2026-08 (36mo) | +5.8%/yr | +18.5% |
| Current momentum carry the latest 12-month rate — the naive answer | 2025-08 → 2026-08 (12mo) | +9.3%/yr | +9.3% |
| Peak regime (COVID) the 2021–23 spike | 2021-04 → 2023-12 (32mo) | +8.6%/yr | +24.6% |
Cite
Cite
MacroGauge DC Build Index (escalation), 2026-10-02, 2018-01=100, +8.4% YoY — https://macrogauge.vercel.app/escalationMethodology
This escalates your number. We publish an input-price index (2018-01=100), not a turnkey $/MW quote — so the base cost is yours to supply, and the calculator applies the ratio between two months of the DC Build index, and optionally carries a rate you choose past the last print. Because the index is a fixed-weight Laspeyres aggregate, it is linear in its components: each row's contribution is
weight × (component index change) ÷ the headline's base index — not the component's own base index (see the note above the bridge table). In the underlying numbers, the contributions sum to the headline escalation with no residual — but each row in the table is rounded to 2 decimal places for display, so the printed rows can land a few hundredths of a point away from the headline. The TOTAL row under the table adds up what's printed; Headline shows the real figure — the difference between them is that rounding, nothing more. Every month here is sampled on its last daily-grid day, not the first — "base month 2024-03" means 2024-03-31 — except the window's end, which stops at the index's latest observation (2026-10-02) and is therefore a partial month. Escalation is national — state parity multipliers on /datacenter are level multipliers (cost relative to the national average), not escalation rates; your base cost for a real site already embeds its location, so applying them here would count location twice. The measured leg above is history: it stops at the index's last print, and nothing past that print is asserted. The deliver by leg is not a forecast either — it carries forward a rate you choose from regimes that have actually occurred (the long-run average, the post-2008 downturn, the last three years, the latest twelve months, or the 2021–23 spike), each shown with the exact window it was measured over. Those windows measure to the last month every Build component actually reports a full print for — not to the partial month noted above — because only 2 of the 12 components (Copper wire & cable and Aluminum mill shapes, 8.5% of the index's weight) have moved since then; anchoring a rate on a 2-component move would misstate it as a basket-wide one. We do not predict which regime will obtain, and we publish no central path. The realized band underneath the table is a count of what happened across overlapping historical windows of the same length as yours, reported alongside the number of independent draws behind it — a small number, over a sample containing one downturn and one spike, best read as a range of precedents rather than a probability. Component sources and weights are documented on /datacenter. Index choice. The default is the live DC Build grid, whose trailing month carries a futures-proxy tail for copper and aluminium until the next PPI replaces it — so its latest month restates. Official prints only is the same Laspeyres basket on the agencies' own prints: every month is complete, and a month once published moves only when BLS revises its print. That is the basis to cite in a price-adjustment clause; its monthly series downloads as CSV on /datacenter. P80. The P80 line is the 80th percentile of the same realized windows as the band: carried over your forward leg, it is the allowance that would have covered four in five of the historical windows of your length — with the same small-sample caveat.