Construction workers are earning up to 30% more and some are nabbing six-figure salaries in the data center boom

Hyperscaler-driven AI demand is triggering a major U.S. data center buildout with Raul Martynek estimating upwards of $100 billion of U.S. investment in 2026 and Bank of America projecting global hyperscale spending of $611 billion across 2025–26 (up 67% in 2025 and 31% in 2026). The boom is inflating construction labor costs—average pay on Skillit rises from $62,000 ($29.80/hr) for non-data-center projects to $81,800 ($39.33/hr) for data-center builds (≈32% uplift)—while some top contractors have doubled revenue and timelines have compressed (projects finished in as little as six months). The scale and speed of spending imply persistent demand for data-center real estate, skilled construction labor and related contractors, creating potential upside for infrastructure and industrial real-estate exposures but presenting margin and labor-cost considerations for developers and operators.

Analysis

Market structure: Hyperscalers (GOOGL/GOOG, AMZN, META, MSFT) and data‑center real‑estate/operating specialists (DLR, EQIX, CONE) are the primary beneficiaries as $611B+ of hyperscale capex (Bank of America 2025–26) forces premium pay for specialized labor and equipment. Large national contractors and heavy‑equipment suppliers (CAT, FLR) gain pricing power; small regional contractors and non‑hyperscale commercial projects face margin squeeze and delayed timelines as labor reallocates and timelines compress to 6–12 months.

Risk assessment: Tail risks include a policy/regulatory clamp on compute‑intensive AI (25–40% capex cut scenario), energy/permits bottlenecks, or a macro shock where 10‑year yield >4.5% increases WACC and deflates ROIC on long datacenter projects. Near term (days–weeks) watch contractor bookings and hourly wage prints; 3–12 months monitor hyperscaler capex guidance and utility interconnection lead times; 2–4 years could see normalization if model efficiency reduces incremental compute growth.

Trade implications: Preferred plays are data‑center REITs and hyperscalers via defined‑risk option spreads and selective equipment/utility exposure; pair trades: long DLR/EQIX vs short office REITs (SLG/O) to capture structural reallocation. Use 6–12 month call spreads on AMZN/GOOG to express upside while hedging volatility; overweight industrials/utilities, underweight traditional office/hospitality real estate over the next 6–18 months.

Contrarian angles: The market underestimates dependencies—grid upgrades, permitting, and semiconductor shortages—that can pause projects; labour‑inflation is partially transitory as training and automation slowly flow in, so wage premium could compress if capex growth falls below ~20% YoY. Historical parallel: telecom tower/fiber booms saw multi‑year overhangs; crowding into REITs could produce yield compression then repricing if funding costs rise.

Key Decisions for Investors

• Establish a 2–3% portfolio long split between DLR (Digital Realty) and EQIX (Equinix) — 1.25–1.5% each — within 2–6 weeks; target +12–18% return in 12 months, stop‑loss 12% if company FFO guidance misses by >10% or interconnection backlog shrinks materially.
• Deploy 1–1.5% capital into 9–12 month call spreads on AMZN and GOOGL (0.5–0.75% each): buy ATM calls and sell 20% OTM calls to cap cost; unwind if either issues capex guidance down >20% YoY or 10Y Treasury rises above 4.5% and tech index drops >8%.
• Reduce exposure to office/legacy commercial REITs (example: SLG, O) by 40–60% over next 3 months and reallocate to industrials/data‑center infrastructure, anticipating multi‑quarter tenant and capex reallocation risks.
• Buy a tactical 0.5–1% exposure to heavy‑equipment providers (CAT) or copper producers (FCX) via 6–12 month call spreads to capture material/equipment demand; trim if producer margins fail to improve within two consecutive quarters or if copper spot falls >15% from entry.