The Americas' data center boom may soon face a critical power crunch, Wärtsilä warns - and the industry is moving too fast to see it coming

Wärtsilä warns of a “speed to powerless” risk as AI data-center buildouts outpace grid expansion across the Americas, driven by interconnection queues, transmission constraints, equipment lead times, and permitting delays. Reported constraints include ~2,600GW of US generation/storage seeking interconnection (end-2025) and a 60% jump in Brazil’s projected 2030 data-center load to 30.3TWh/year. The company also models power-planning implications: RICE shows a 25% levelized cost of electricity advantage ($86/MWh vs $111/MWh), implying ~$178M annual savings for a 1GW data center, but stresses a resilient hybrid (on-site plus grid/contracting) approach.

Analysis

The market is likely underestimating where the economic rents migrate when grid access becomes the bottleneck: away from pure data-center real estate and toward the vendors that can deliver modular, on-site power with the shortest path to commissioning. That favors engine OEMs, electrical balance-of-plant, and EPC execution more than the hyperscalers themselves, because the latter may still build revenue capacity but at a slower, more capital-intensive cadence.

Over the next 1-3 months, the important signal is not the report itself but whether hyperscalers and colo operators start attaching hybrid power language to RFPs and earnings commentary. If they do, the likely second-order effect is a repricing of the power stack: higher backlog visibility for industrial power names, but also a valuation haircut for data-center growth narratives that assume unconstrained lease-up and rapid load addition.

The contrarian risk is that this is partly a bottleneck reallocation, not an outright demand destroyer. If interconnection approvals accelerate, utility buildouts catch up, or alternative firm power solutions gain traction, the urgency premium for reciprocating-engine solutions fades quickly. The structural takeaway over 6-18 months is that the winners are the picks-and-shovels names with service revenue and fleet replacement potential, while the losers are the projects whose IRR depends on cheap, fast grid power that may not exist on schedule.

Key Decisions for Investors

• Long CMI on a 3-6 month horizon as the cleanest public-market proxy for on-site reciprocating-engine adoption; add on weakness if hyperscaler RFPs start emphasizing hybrid power. Falsifier: no order-book inflection or commentary that data-center power demand is being solved through grid-only projects.
• Pair trade: long CMI / short GEV for 3-12 months. Thesis is that data-center buyers will increasingly prefer engine-based modular power over aeroderivative-heavy solutions; stop if GEV announces material data-center turbine wins or if CMI backlog fails to accelerate.
• Buy ETN or PWR on pullbacks as the bottleneck shifts spending toward switchgear, transformers, and EPC execution. This is a lower-beta way to own the capex migration regardless of whether final generation is grid-tied or on-site.
• Watch DLR and EQIX rather than short them immediately; if quarterly leasing commentary starts flagging power-delivery delays or pushouts, then consider a tactical short against CMI/ETN. Falsifier: continued strong absorption and no change in delivery schedules.
• Set an alert on utility/ISO queue data in PJM, ERCOT, and Brazil; if interconnection timelines improve meaningfully over the next 1-2 quarters, reduce the urgency premium on the on-site power trade.