Datacenter Buildout Hiring Signals — May 2026

Volume & Trend

• Total open roles (last 90d): **30** (vs previous 90d: **no baseline** — `total_roles_prev_90d = 0`, `delta_pct = null`; May 2026 is the first tracked 90-day window for `datacenter-buildout`, so trend is establish-only, not growth).
• Distribution by level: **100% unspecified (30 of 30)** — every requisition omits a structured seniority level. Seniority is buried in prose instead ("8+ years mission-critical power", "PE license preferred", "12+ years in data center real estate"), so the level field is non-discriminating this period.

Top 5 Specialties (last 90d)

1. **Medium voltage distribution**: 6 jobs at 6 companies
2. **Switchgear**: 6 jobs at 6 companies
3. **Grid interconnect**: 6 jobs at 6 companies
4. **Liquid cooling design**: 5 jobs at 5 companies
5. **Thermal management**: 4 jobs at 4 companies

These five specialties account for 27 of 30 roles. Power-path skills (MV distribution + switchgear + grid interconnect = 18 roles, each spread one-per-company across 6 distinct firms) outnumber heat-rejection skills (liquid cooling + thermal = 9 roles) by 2:1.

Top 5 Hiring Companies (last 90d)

| Company | New Roles | Top Specialty | Avg Level | |---------|-----------|---------------|-----------| | Digital Realty | 0 | n/a | unspecified | | Equinix | 0 | n/a | unspecified | | Iron Mountain | 0 | n/a | unspecified | | Compass Datacenters | 0 | n/a | unspecified | | Stack Infrastructure | 0 | n/a | unspecified |

(Values copied verbatim from `stats.top_companies`. Note the contradiction worth flagging to data-eng: aggregate `total_roles_90d = 30` while per-company `new_roles = 0` for all five — the company-level rollup is not populated this period, so company ranking below relies only on the JD sample, not on countable stats.)

Business-Talent Correlation Insights (3 most striking)

**No correlation insights can be produced this period.** `correlation_links` is an empty array — zero business signals were ingested for `datacenter-buildout` in the May 2026 window. Per the numeric-integrity constraint, no correlation pattern, business-signal date, or confidence score can be asserted without source links, and none will be fabricated. The interpretation: the talent layer is live but the public-business-signal layer for this topic is not yet wired, which is itself the headline finding — see next section.

Hiring as Leading Indicator

Because `correlation_links` is empty, **every hiring cluster below precedes any captured public business signal by definition** — this topic is currently pure leading indicator with no lagging confirmation. Per-company counts are not in the precomputed stats (`new_roles = 0` for all firms), so each bullet is anchored to the specialty-level counts that *are* in `stats.top_specialties`, with company names drawn from the JD sample:

• **Grid-interconnect cluster (6 roles / 6 companies, 0 public capacity signals):** Digital Realty ("Manager, Utility & Grid Interconnection" — "multi-hundred-megawatt power capacity for new campus"), Iron Mountain ("Director, Energy and Grid Strategy" — "PPAs… multi-hundred-MW"), Compass ("Director of Site Selection and Land Development" — "large-load power feasibility"), Equinix ("Director, Capacity Planning and Site Acquisition" — EMEA greenfield). Six interconnection/PPA hires across six firms with no announced capacity deals — expect multi-hundred-MW interconnection or PPA announcements in **Q3–Q4 2026**.
• **Liquid-cooling cluster (5 roles / 5 companies, 0 public deployment signals):** Digital Realty, Equinix, Iron Mountain, Compass all staffing "direct-to-chip and rear-door heat exchanger" / ">100 kW/rack" roles. Reference-design hiring with no announced high-density product tier — expect high-density / AI-rack colocation product launches **Q4 2026–Q1 2027**.
• **Switchgear + MV distribution cluster (12 roles / 6 companies):** uniform one-per-company spread of MV switchgear electrical engineers with no announced new builds in the signal layer — expect shell/groundbreaking announcements to trail this requisition wave by one to two quarters.

**Why now (specific capability shift, not "AI got better"):** the JD language ">100 kW/rack", "direct-to-chip", and "rear-door heat exchanger" maps directly to the 2024–2025 arrival of GB200 NVL72-class racks (~120 kW+ per rack), which broke air-cooling economics and forced facility-water/CDU retrofits. The hiring surge in liquid-cooling + grid-interconnect roles is the staffing echo of that 2024–2025 rack-power step-change, not generic growth.

Reverse-Hype Warnings

**Over-marketing detected:** notable for its near-absence. Across all 20 sampled JDs there is **zero** "rockstar", "world-class", "ninja", or "moonshot" language. The only mild inflation is the repeated "hyperscale" / "mission-critical" framing (Digital Realty, Iron Mountain, Compass campus PM roles) and recurring superlatives like "end-to-end delivery" and "multi-hundred-megawatt" — buzzwords that signal scale ambition but are operationally vague. The relevant reverse-hype read here is inverted: these are unusually sober, credential-gated JDs ("PE license", "single-line diagrams", "short-circuit/coordination studies", "NEC/IEC standards", "Uptime Tier III/IV"), which means the talent layer is real construction demand, not narrative inflation. Sober JDs in a hyped sector are a *positive* signal of genuine buildout.

**What's under-reported:** First, **compensation** — not one of the 20 JDs states a salary band despite requiring 8–12+ years and PE licensure, hiding the true cost-of-talent in a constrained labor pool. Second, **the grid-interconnection bottleneck** — multiple roles casually require securing "multi-hundred-MW" capacity and "managing substation delivery timelines" / "PPAs", but none acknowledge the multi-year utility queue that gates these projects; the timeline risk that most threatens the buildout is the least described. Third, **liquid-cooling staffing depth** — only 5 liquid-cooling and 4 thermal roles versus 18 power-path roles, despite JDs themselves citing >100 kW/rack AI density; heat-rejection headcount is structurally under-built relative to the power capacity these same companies are racing to secure.