Quick Read
The original global buildout report has the broadest source base and useful charts, but it is a research synthesis rather than an automated monthly database pipeline. The aggregate MVP gauge exists and runs end-to-end, but today its regime call is low-confidence because the build-rate history is thin/interpolated and one source used a fixture fallback. The company-level research is more audit-heavy but earlier: ORCL, MSFT, and AMZN are accepted best-proxy datasets; GOOGL is under repair; no company has accepted direct powered-on GW rows yet. The biggest missing piece is true GW coverage from companies that actually disclose MW/GW, then a portfolio chart that does not mix incompatible bases.
Progress
What Is Left
- Repair GOOGL provenance so every forecast number is supported by its cited snippet.
- Promote the original global buildout report into a repeatable data pipeline or explicitly keep it as a manual synthesis.
- Prioritize direct-GW companies next: DLR, EQIX, IRM, APLD, CRWV, CORZ, IREN, WULF, CIFR, HUT, NBIS.
- Keep bases separate: powered-on GW, under-construction GW, contracted MW, capex, revenue, and grid demand.
- Build a portfolio chart only after basis reconciliation; do not average GW and capex proxies.
- Backtest forecast-vs-actual once enough delivered outcomes exist.
- Raise the aggregate gauge confidence by replacing fallbacks, adding vintages, and reducing interpolation.
Attempt 1: Original Global Buildout Report
This is the broad research report at the datacenters root. It estimates global transparent-market capacity, pipeline depth, annual additions, and forecast revisions. It currently has 502 normalized figure rows, 89 sources, 73 forecast curve rows, and 7 explicit revision rows. Caveat: this track is strongest as a sourced synthesis; it is not yet the same automated, provenance-enforced monthly pipeline as the gauge MVP.
Operational Trend
Build Pipeline
Forecast Revisions
Annual Additions
Attempt 2: Aggregate Gauge Rows Written
Attempt 2: Aggregate Series By Basis
Attempt 3: Company-Level Coverage
| Company | Status | Actual GW rows | Proxy rows | Forecast vintages / as-of periods | Meaning |
|---|---|---|---|---|---|
| ORCL | Accepted proxy | 0 | 8 | 16 / 8 | Usable best-proxy dataset; direct GW is blank. |
| MSFT | Accepted proxy | 0 | 8 | 16 / 8 | Usable best-proxy dataset; direct GW is blank. |
| AMZN | Accepted proxy | 0 | 9 | 17 / 9 | Usable best-proxy dataset; direct GW is blank. |
| GOOGL | Accepted proxy | 0 | 9 | 20 / 9 | Usable best-proxy dataset; direct GW is blank. |
| META | Accepted proxy | 0 | 0 | 13 / 12 | Usable best-proxy dataset; direct GW is blank. |
| CRWV | Accepted GW | 7 | 0 | 12 / 8 | Usable direct-GW dataset. |
| DLR | Fallback-contaminated | 0 | 0 | 0 / 0 | Do not count yet; artifacts are fallback/cross-ticker contaminated. |
| EQIX | Fallback-contaminated | 0 | 0 | 0 / 0 | Do not count yet; artifacts are fallback/cross-ticker contaminated. |
| IRM | Accepted GW | 9 | 0 | 9 / 9 | Usable direct-GW dataset. |
| APLD | Accepted GW | 11 | 0 | 11 / 11 | Usable direct-GW dataset. |
| CORZ | Accepted GW | 8 | 0 | 12 / 8 | Usable direct-GW dataset. |
| IREN | Accepted GW | 9 | 0 | 15 / 9 | Usable direct-GW dataset. |
| WULF | Accepted GW | 11 | 0 | 16 / 11 | Usable direct-GW dataset. |
| CIFR | Accepted GW | 13 | 0 | 13 / 13 | Usable direct-GW dataset. |
| HUT | Accepted GW | 9 | 0 | 12 / 9 | Usable direct-GW dataset. |
| NBIS | Accepted GW | 0 | 0 | 17 / 9 | Usable direct-GW dataset. |
| NVDA | Accepted proxy | 0 | 0 | 9 / 9 | Usable best-proxy dataset; direct GW is blank. |
Regime Gauge
Indicator Ladder
Lead/Lag Check
Track Record
Aggregate Source Status
| Source | Family | Basis | Status |
|---|---|---|---|
| amzn_capex | capital_intent | capex_usd | live fetch HTTP 200 |
| googl_capex | capital_intent | capex_usd | live fetch HTTP 200 |
| meta_capex | capital_intent | capex_usd | live fetch HTTP 200 |
| msft_capex | capital_intent | capex_usd | live fetch HTTP 200 |
| orcl_capex | capital_intent | capex_usd | live fetch HTTP 200 |
| vertiv_b2b | equipment | grid_demand | live fetch HTTP 200 |
| goldman_us | market_tracker | grid_demand | live fetch HTTP 200 |
| jll | market_tracker | it_load | live fetch HTTP 200 |
| knight_frank | market_tracker | it_load | live fetch HTTP 200 |
| sightline | market_tracker | it_load | live fetch HTTP 200 |
| ercot_queue | power_queue | grid_demand | fixture fallback used: HTTP 403 |
| pjm_queue | power_queue | grid_demand | live fetch HTTP 200 |
| nvda_dc | target | revenue_usd | live fetch HTTP 200 |
Links
Generated 2026-06-08T20:25:00+00:00. Active company-level agent: gw-forecast-production-001-loop07. Aggregate lead/lag currently reports 4 quarters, but remains low-confidence.