FT has an article about a $3Tn build for AI (and not without the usual doom and gloom).
But if anyone is interested in the shovels and picks play of the AI, below is your starting point.
Grid equipment & power components - ETN (Eaton) — switchgear, UPS, busways, modular enclosures.
Sensitivities: Power H / Financing L–M. AI DC demand directly pulls orders; less balance-sheet risk. Notables: acquired Fibrebond (modular DC power enclosures) 4/1/25; partnering with NVIDIA on DC “grid-to-chip” power mgmt. - HUBB (Hubbell) — utility/grid components & DC gear.
Sensitivities: Power H / Financing L. Mgmt flagged ~30% DC sales growth in CY25. - GEV (GE Vernova) — HVDC, transformers/switchgear, gas turbines.
Sensitivities: Power H / Financing L–M. Backlog +$5.2B QoQ in Q2’25; electrification (grid) strength. - ABB (ABB) — electrification leader; DC demand a growing mix.
Sensitivities: Power H / Financing L. Orders/backlog rising in 2025; grids tight. - ENR.DE / ADR: SMNEY (Siemens Energy) — transformers/HV, gas turbines.
Sensitivities: Power H / Financing L–M. Record ~€136B backlog; ~60% of 2025 gas-turbine orders tied to DCs. - POWL (Powell Industries) — medium/low-voltage switchgear & E-houses (benefits from LNG + DC builds).
Sensitivities: Power H / Financing L. $1.3B backlog (Mar-25). - Optional (EU): SU.PA / ADR: SBGSY (Schneider Electric) — end-to-end DC electrification & cooling.
Sensitivities: Power H / Financing L. H1/Q2’25 beat with DC-driven growth. EPC / T&D builders - PWR (Quanta Services) — T&D lines, substations, grid interconnects.
Sensitivities: Power H / Financing L–M. Record ~$35.8B total backlog (Q2’25). Rule-of-thumb: a $1B T&D revenue timing shift ˜ $100–120M EBITDA timing (assumes 10–12% margin; for timing, not value). - MTZ (MasTec) — clean energy/T&D/comm EPC (incl. DC campuses).
Sensitivities: Power H / Financing L–M. 18-mo backlog +23% YoY; raised FY25 guide. - MYRG (MYR Group) — T&D + C&I electrical (data centers).
Sensitivities: Power H / Financing L. Backlog $2.64B (Jun-25); C&I (incl. DC) strong. Data-center developers / operators (REITs & managers) - DLR (Digital Realty) — wholesale/hyperscale.
Sensitivities: Power H (interconnect/utility) / Financing H (rate/WACC). Data points: delivered 96 MW in Q2; expected stabilized yield ~12.2% on dev pipeline; net debt $18.45B (Q2’25). Quant:
• Power delay: Using $10–14M/MW capex norms, 100 MW at 12% NOI yield ? ~$144M/yr NOI; a 2-quarter slip defers ~$72M NOI/FFO.
• Rates: +100 bps on $18.45B gross debt ? ~$185M/yr extra interest (˜ $0.53/sh using ~346.6M shares/units), before hedges. - EQIX (Equinix) — interconnect-heavy retail colocation; scaling to 30–60MW & multi-100MW campuses; pursuing >1 GW advanced-nuclear/off-grid options.
Sensitivities: Power H / Financing M. Raised FY25 guide; actively securing alternative power. - DBRG (DigitalBridge) — developer/asset-manager (Vantage, Switch, etc.).
Sensitivities: Power H / Financing M. Management: “running out of power in 18–24 months”; 2Q25 deck: “powered land/on-site gen” strategy. Utilities & IPPs (power gating the build-out) - DUK (Duke Energy) — Carolinas large-load growth.
Sensitivities: Power H / Financing M–H (equity needs). Lifted 5-yr capex to ~$83B; plans $6.5B equity ’25–’29; adding ~5 GW gas by 2029. - D (Dominion Energy) — NoVA DC epicenter.
Sensitivities: Power H / Financing M (rate design). 2024 IRP ties DC load to ~$22B incremental system costs over 15 yrs; new high-load rate class proposal. - VST (Vistra) — merchant gen retail + growth via gas assets.
Sensitivities: Power H / Financing M. Buying ~2.6 GW gas fleet for $1.9B to meet surging demand; exploring bespoke DC power deals. - CEG (Constellation) — nuclear-heavy baseload (24/7/clean PPAs prized by AI).
Sensitivities: Power H / Financing L–M. (Context: nuclear’s premium for 24/7 credits & AI PPAs; pair with DC offtakers.) General note; company-specific deal terms vary. - NEE (NextEra) — renewables + regulated FPL; levered to PPA appetite & interconnect queues.
Sensitivities: Power H / Financing M. (Watch: queue timelines, curtailment, storage tie-ins.) What to model (fast) Grid OEMs (ETN/HUBB/GEV/ABB/ENR/POWL/SU):
- Unit growth tied to transformer/switchgear lead times and HV equipment capacity; NEMA survey: HV transformers often ~3-year lead times (pricing power tailwind while tight).
- Backlog burn vs. capacity adds (GEV +$5B QoQ; Siemens Energy record backlog). Sensitivity: ±10% shift in HV delivery slots ? proportional revenue timing; margins less volatile due to pricing locked in backlog.
EPCs (PWR/MTZ/MYRG):
- Interconnect queue/utility-readiness drives revenue recognition. Sensitivity: for PWR, $35.8B backlog; if 10% of T&D work slips 1 quarter, revenue timing ? ˜ $0.9B; at ~10–12% EBITDA margin ? $90–110M EBITDA timing shift (illustrative).
Developers (DLR/EQIX/DBRG):
- MW delivered on schedule is the KPI. Use: cost/MW $10–14M and ~10–12% stabilized yields to convert MW timing into NOI/FFO timing (example above for DLR).
- Debt sensitivity: map net debt and maturities to +/-100 bps scenarios (DLR example: +$185M/yr interest per +100 bps gross on Q2’25 debt).
- Alternate power optionality (EQIX/DBRG): on-site gen & advanced nuclear cut grid-constraint risk but add project/deployment risk.
Utilities/IPPs (DUK/D/VST):
- Capex cadence & cost allocation. DUK: bigger plan (˜$83B/5y) implies more equity and rate cases; D: IRP shows $22B DC-driven costs driving new rate designs; siting & rate-class outcomes set earnings path.
- Merchant leverage (VST): asset buys (~2.6 GW) + bespoke DC contracts = earnings torque to realized spark spreads & contract spreads.
Quick screens / alerts - Transformers/HV gear: track lead-time updates (Utility Dive / OEM calls). Slippage down = potential price pressure; up = capacity scarcity (pricing power).
- DLR/EQIX dev math: MW under construction, expected yields, power commitments disclosed each quarter; convert MW timing ? NOI delta using the $/MW and yield inputs above.
- Policy drift: VA/NC proceedings (rate classes, IRPs) and PJM load forecasts; these move utility capex, cost sharing, and DC siting.
|
