Every major energy infrastructure player is spending like the grid crisis is already here. Siemens Energy committed €2.3 billion to expand transformer and switchgear manufacturing through 2028. GE Vernova closed a $5.275 billion acquisition of Prolec GE — 10,000 employees, seven factories across the Americas. Utilities have collectively planned $1.4 trillion in infrastructure spending through 2030, double the previous decade.
The spending is driven by real scarcity. According to DOE data cited by Wood Mackenzie, 55% of U.S. distribution transformers — roughly 40 million utility-managed units — are more than 33 years old, beyond their expected service life. Lead times for large power transformers run two to four years. Prices have doubled or more since 2019, with some utilities reporting costs four to six times pre-2022 levels. And the U.S. does not rely in a large pool of domestic producers of the specialized electrical steel these transformers require.
All of this capital is flowing. But it’s flowing toward three competing strategies that rest on very different assumptions about where power infrastructure is heading — and very different exposure to timeline risk.
Strategy 1: Scale the Grid
The first response expands centralized infrastructure for shared use. GE Vernova’s electrification equipment backlog has reached $26 billion. Hitachi Energy is building what will be the nation’s largest large power transformer plant in Virginia — a $457 million facility targeting 2028. Siemens Energy’s $1 billion U.S. expansion adds over 1,500 jobs across six states — a new high-voltage switchgear factory in Mississippi, turbine manufacturing resuming in Charlotte, parts production expanding in Winston-Salem, and grid engineering and R&D growing in Raleigh. The investment alone will increase Siemens Energy’s global gas turbine production capacity by 20%. Since 2023, manufacturers have committed nearly $2 billion to new or expanded domestic transformer capacity.
The assumption: the centralized grid model holds. It just needs to be bigger. The risk: multi-year construction timelines, aging transmission infrastructure, and interconnection queues that are already overwhelmed.
Strategy 2: Bypass the Grid
The fastest-moving operators have stopped waiting for shared infrastructure. Data center developers now secure dedicated generation before site selection, before design, before anything else. The industry calls it “Bring Your Own Power.”
Xcel Energy and GE Vernova formed a strategic alliance in February to co-develop 6 GW of generation, storage, and transmission for data center load. NextEra separately reserved 4 GW of gas turbine slots specifically for behind-the-meter power. Constellation Energy is restarting Three Mile Island‘s reactor under a 20-year power purchase agreement with Microsoft. Meta signed a 1.1 GW nuclear deal with Constellation for its AI data centers in Illinois. FERC has ordered PJM to develop new rules for co-locating large loads at power plants — giving regulatory structure to what was already happening.
The assumption: if the grid can’t deliver on your timeline, secure your own supply. The risk: capital intensity, stranded assets if demand shifts, and the coordination gaps that emerge when everyone builds independently.
Strategy 3: Mobilize Power
A third response is emerging from the maritime sector, built on an assumption the other two don’t consider: power infrastructure doesn’t have to be fixed in place.
Karpowership operates the world’s largest fleet of floating power plants — over 10,000 MW of installed capacity across 45 vessels in 20+ countries, with units that commission in under 30 days. Their subsidiary Kinetics, partnering with Mitsui O.S.K. Lines, is converting vessels into floating data centers — a one-year conversion versus three years for equivalent onshore facilities. Their Intelliship concept integrates power generation and data center operations on a single self-sufficient vessel with no grid connection required.
Floating nuclear power platforms are following a similar trajectory. Samsung Heavy Industries received design certification from the American Bureau of Shipping in December 2025 for a floating SMR platform. Core Power launched its Liberty programme targeting commercial floating nuclear by mid-2030s. But the regulatory pathway remains unbuilt — the IMO’s revised Nuclear Code won’t be adopted before 2030, and the international liability framework for floating nuclear installations doesn’t exist yet. That’s a coordination problem across three institutional frameworks that have never been aligned for this use case.
For now, floating gas and LNG generation operates within existing regulatory structures. Floating nuclear remains under discussion.
The Operating Environment
What makes all three strategies harder to evaluate: the U.S. and EU regulatory environments are moving in opposite directions. The U.S. revoked the EPA Endangerment Finding few days ago and is cutting $15 billion in grid and clean energy infrastructure funding, betting that deregulation and private capital move faster than federal programs. The EU is going the other way — FuelEU Maritime mandates shore power at all core European ports by 2030, but only 20% of required connections are installed and grid expansion timelines run five to fifteen years.
Last April, Spain lost 60% of its generation capacity in five seconds. Sixty million people went dark. Portugal has since committed €400 million in grid hardening. Even well-funded systems discovered that capital alone doesn’t solve architectural gaps.
Operators with assets or trade routes spanning both regulatory regimes are making capital decisions in an environment where the rules themselves are diverging.
The Diagnostic
All three strategies carry timeline risk — construction, supply chain, regulatory. But the underlying demand isn’t going anywhere. Energy infrastructure isn’t discretionary. Whether we like it or not, it’s necessary.
And where there’s capital and urgency in the same sector, the returns go to whoever solves the execution path first.
Comments are closed