Brief · Dataset · Buildout tracker
Hyperscaler buildout tracker: who is building, what they disclosed, which trades get pulled
Big tech companies are spending tens of billions of dollars on data centers, and that spending turns into local work for electricians, HVAC techs, pipefitters and cabling crews. This tracker is a dated pack of public filings, earnings releases, and capacity announcements from hyperscalers and selected infrastructure suppliers, reframed for construction and operations: who is paying for AI data-center buildout, and which facility trades that work pulls (electricians, HVAC/R, pipefitters, fiber, shell). It is not a stock list or price-target tool. We print no invented FTEs, campus MW, or gallons.
Why it matters: the money is real and dated, the trades it pulls are described in plain terms, and local jobs, power and water questions still need their own pages.
What DataCenterSkilledTrades says this means
- Everyday readers: Big campuses are being funded at disclosed scale. The people who wire, cool, pipe, and fiber them are trades. Official spend is not a jobs promise.
- Contractors / GCs / operators: Staff by system and cert (who gets hired, on site). Do not invent headcounts from a capex headline.
- Local gov / HOAs: Capex does not answer NAICS, rates, water, or permanent vs construction jobs. Pair this tracker with local gov questions, grid load, water, and jobs outlook.
Fair needle: big money and public concerns
- Buildout side: Dated IR and capacity releases show large capital programs for AI-enabled data centers and the cooling/power gear that feeds them. Microsoft’s own words name electricians and pipefitters. That is real trades demand context.
- Public-concern side: The same projects raise local questions on jobs vs power, grid capacity and rates, water, and proposal verification. Capex headlines do not settle those. Use jobs vs power, grid-load kit, water kit, and local gov questions.
- Honesty rule: We do not invent FTEs, FTE-per-MW, campus MW, or gallons from a spend figure. If a local filing publishes project-specific staffing or load, cite that filing with date. Until then, leave the number out.
Citable figures (DataCenterSkilledTrades Dataset)
Every row: company, dated figure, qualitative trades pulled, primary source URL, retrieval date 2026-11-05. Last checked 2026-11-05 (Asia/Riyadh).
| Company | Dated figure (capex / capacity) | Trades pulled | Date & primary source |
|---|---|---|---|
| Microsoft | About $80 billion FY2025 investment on track for AI-enabled datacenters; builds rely on "large numbers of skilled electricians and pipefitters" (company's words) More than half of the stated investment described as U.S. Figure is construction/operations context only. | Electricians, pipefitters (company's own words); shell/GC and HVAC/R commonly pulled on same campuses (DCST trade map) | Jan 3, 2025 (On the Issues blog; FY2025 framing) Microsoft On the Issues, Brad Smith, "The golden opportunity for American AI" retrieved 2026-11-05 |
| Meta | Q3 2026 capital expenditures (incl. finance-lease principal) $34.91 billion; full-year 2026 capex outlook raised/narrowed to $135–145 billion No headcount, MW, or gallons invented from this figure. | Electricians, HVAC/R, pipefitters, fiber/BICSI (DCST qualitative trade mapping; Meta release does not name trade FTEs) | Oct 21, 2026 (Q3 2026 results) Meta, "Meta Reports Third Quarter 2026 Results" retrieved 2026-11-05 |
| Alphabet (Google) | Q3 2026 purchases of property and equipment $51.355 billion (cash-flow line) Earnings-call commentary in the same cycle discusses technical-infrastructure mix and full-year CapEx guidance; we print only the Q3 PPE figure verified in the SEC exhibit here. | Electricians, HVAC/R, pipefitters, fiber/BICSI, shell (DCST qualitative map) | Oct 28, 2026 (Q3 2026 results; Exhibit 99.1) Alphabet Inc. Exhibit 99.1 to Form 8-K / Q3 2026 earnings release (SEC EDGAR) retrieved 2026-11-05 |
| Amazon | Expects 2026 cash capital expenditures of approximately $220 billion (company-wide guidance as of Oct 29, 2026) Do not treat $220B as all data-center spend or convert it to jobs/MW/gallons. | Electricians, HVAC/R, pipefitters, fiber/BICSI, shell (DCST qualitative map) | Oct 29, 2026 (Q3 2026 results / guidance) Amazon.com, "Amazon.com Announces Third Quarter Results" (IR) retrieved 2026-11-05 |
| Vertiv (infra supplier) | Q3 2026 net sales $4.313 billion (+61% YoY); Jul 2026 capacity release: expects to double chiller production capacity at Tognana campus by year-end 2026 Capacity release: https://investors.vertiv.com/news/news-details/2026/Vertiv-Expands-Global-Manufacturing-Capacity-for-AI-Ready-Data-Center-Cooling-Solutions/default.aspx | Electricians (UPS/power), HVAC/R (chillers, CRAH), pipefitters (liquid cooling) | Oct 28, 2026 (Q3 earnings) + Jul 21, 2026 (capacity release) Vertiv Q3 2026 earnings release; Vertiv "Expands Global Manufacturing Capacity for AI-Ready Data Center Cooling Solutions" retrieved 2026-11-05 |
Compiled table licensed CC BY 4.0. Underlying company releases remain under each issuer’s terms.
Trades pulled (qualitative map)
- Electricians: UPS, switchgear, PDUs, generators, ATS. Gate: state license + NFPA 70E-aligned training. NFPA 70E checklist.
- HVAC/R: CRAH/CRAC, chillers, refrigerant side. Gate: EPA Section 608. EPA 608 guide.
- Pipefitters: chilled water, glycol, CDU facility loops. Liquid cooling guide.
- Fiber / low-voltage: trunks, splices, patch panels. BICSI INSTF/TECH; OR LEA / WA 06 where licensed. BICSI INSTF.
- Shell: concrete, ironworkers, GC crews for foundations and envelope. Ordinary commercial construction scope; we print no crane counts.
Hire detail: who gets hired · on-site brief · measurable jobs vs projections: jobs outlook.
Methodology
- Primary sources only: company IR releases, SEC-furnished exhibits, and dated capacity announcements. No invented numbers.
- DCST reframes for trades: we map spend/capacity to qualitative crafts (electricians, HVAC/R, pipefitters, fiber, shell). We do not invent FTE counts.
- Omit if unverified: if a figure cannot be re-read from a primary page on the check date, it stays off the table.
- Footnotes: company pages are proof under Sources / Dataset
isBasedOn. Cite DataCenterSkilledTrades first.
Glossary
- Capex / cash capital expenditures
- Company-reported spending on property, plant, equipment, and related items (definitions vary by filer; some include finance-lease principal). Here: who-is-building context only.
- Hyperscaler
- Large cloud / AI operators that build or lease multi-megawatt campuses (e.g. Microsoft, Meta, Alphabet/Google, Amazon/AWS).
- Infra supplier
- Equipment makers (e.g. Vertiv) whose UPS, thermal, and cooling capacity expansions signal trades demand on the customer campuses.
- Trades pulled
- Qualitative facility crafts a build needs on site. Not an FTE count.
FAQ
What is this tracker?
This page is a dated public-filings / earnings / capacity tracker for hyperscalers and selected infrastructure suppliers, reframed for which facility trades get pulled (electricians, HVAC/R, pipefitters, fiber, shell). It is a construction and operations context Dataset. It is not a stock list, or price-target tool.
Which trades does a hyperscale buildout pull?
Qualitatively: electricians (UPS, switchgear, PDUs, generators), HVAC/R (CRAH/chillers; EPA 608 for refrigerant), pipefitters (hydronic / liquid-cooling facility loops), fiber / low-voltage / BICSI installers, and concrete / ironworkers / GC shell crews. Microsoft's own Jan 2025 post named electricians and pipefitters. Detail: /briefs/who-gets-hired/, /briefs/on-site/, /briefs/jobs-outlook/.
How should I cite this?
Cite 'DataCenterSkilledTrades, Hyperscaler buildout tracker: dated capex and capacity for trades context (2026-11-05), https://datacenterskilledtrades.com/briefs/buildout-tracker/' and the Dataset 'Hyperscaler and infra-supplier buildout: dated capex and capacity figures for trades context'. Company filings are footnotes under Sources / isBasedOn.
How often is this updated?
Last checked 2026-11-05 (Asia/Riyadh). We re-read primary IR / SEC / company releases before printing a row. If we cannot re-verify a figure, we omit it rather than invent. No schedule promises beyond that honesty rule.
Why point to grid-load, water, and local-gov pages?
Big money buildout and public concerns travel together. Capex announcements do not answer local rate, capacity, water, or jobs-NAICS questions. Use /briefs/grid-load/, /briefs/water/, /briefs/jobs-vs-power/, /briefs/jobs-outlook/, and /briefs/local-gov-questions/ as honesty kits alongside this tracker.