Grid transformer and electrical infrastructure TAM is driven by three simultaneous demand waves: AI data center construction, renewable energy grid upgrades, and domestic manufacturing reshoring (CHIPS Act fabs, EV plants). All three require high-voltage transformers and grid connections from the same constrained supply base. The base case reflects sustained AI capex through 2030 with gradual supply-side response as Hitachi Energy, ABB, and Siemens Energy expand manufacturing. Bear case assumes hyperscalers successfully deploy onsite generation (gas turbines, nuclear SMRs) at scale, reducing utility grid dependency and softening transformer demand. Bull case assumes the grid interconnection queue never clears — every incremental data center requires its own dedicated power infrastructure, sustained transformer demand well above current supply.
“Of ~140 large US data center projects representing 12 GW of capacity planned for 2026, only one-third are under construction. The remainder are blocked by grid connection delays and transformer lead times now stretching to 3–5 years — in an industry that deploys in 18-month cycles. More than half of planned 2026 US data center capacity is projected delayed or cancelled.”
“High-voltage transformer lead times have expanded from 24–30 months pre-2020 to 3–5 years in 2026, driven by simultaneous demand from AI data centers, EV grid upgrades, renewable energy interconnection, and domestic manufacturing reshoring. US transformer manufacturing capacity has not grown commensurately — domestic production covers only a fraction of demand.”
“AI data centers draw 60–150 MW each at utility scale. Stargate alone targets 5 GW of total power — equivalent to 5 large power plants. Hyperscalers are now pursuing "bring your own power" strategies: contracting directly with gas turbine operators, nuclear SMR developers, and onsite generation to bypass grid bottlenecks entirely. This creates a secondary boom in power generation hardware.”
“Imports of high-power transformers from China surged from fewer than 1,500 units in 2022 to more than 8,000 in 2025 as US operators turned to international suppliers to address domestic shortages. South Korea, Canada, and Mexico are also supplementing supply. Import dependency has increased but adds months to already stretched timelines.”
“Grid interconnection queues in the US have ballooned to over 2,600 GW of pending projects — more than double the current installed US generation capacity. Average wait time for grid connection is now 5+ years. This is the upstream constraint: even if transformers were available, utilities cannot connect new data centers to the grid fast enough.”
Only one-third of the ~140 large US data center projects planned for 2026 are under construction. The 7 GW stuck in pre-production represents roughly $70B+ in stranded capital waiting for transformers and grid connections. Every month of delay costs hyperscalers competitive positioning in AI infrastructure. This is creating enormous pricing power for any supplier that can deliver electrical infrastructure faster than the 3–5 year standard lead time.
Hyperscalers — Microsoft, Google, Amazon, Meta — are bypassing utility grid connections by contracting directly with power generators: natural gas peakers, nuclear SMR developers (Kairos, X-energy, Oklo), and onsite diesel/hydrogen backup. BYOP creates demand for distribution transformers, switchgear, and power management systems independent of utility timelines. It also creates a new supply chain for modular power generation hardware that barely exists today.
Microsoft (Three Mile Island restart), Google (Kairos Power agreement), Amazon (X-energy deal), and Oracle (SMR campus plan) are all contracting for nuclear power specifically to serve AI data centers. SMR deployments require dedicated electrical infrastructure including step-up transformers, switchgear, and grid-interconnect hardware. If even 5–10 SMR projects reach commissioning by 2030, it creates a new procurement channel for high-voltage electrical equipment outside the standard utility process.
The US currently imports a large fraction of its high-voltage transformers, with dependence on China increasing sharply (1,500 units in 2022 → 8,000+ in 2025). This has become a national security concern. DOE and DHS are pushing domestic manufacturing via CHIPS-Act-style incentives for transformer production. ABB, Hitachi Energy, and smaller players are investing in US transformer factories — but new capacity takes 3–5 years to commission, meaning the shortage deepens before it resolves.