Data center operators could absorb up to 40% of the electrical-equipment market under accelerated AI build-out scenarios, with lead times on large power transformers surpassing 160 weeks by the first quarter of 2026 [S1]. The pull is concentrated in three-phase medium-voltage step-down units that feed 415 V or 480 V IT halls, the same class the broader LV/MV data-center power-distribution market valued at USD 4.32 B in 2026 and growing at a 10.7% CAGR toward USD 6.48 B by 2030 [S7].
Global data-center power demand is projected to roughly double from 415 TWh in 2024 to about 945 TWh by 2030, a load curve that translates directly into transformer MVA orders, not just chip procurement [S6]. Vendor revenue tracks the same line: Market Data Forecast sizes the data-center transformer market at USD 8.92 B (2025), USD 9.48 B (2026), and USD 15.46 B by 2034 at 6.3% CAGR [S2], while Grand View Research reports USD 8.9 B (2026) growing to USD 14.5 B by 2033 at 7.2% CAGR [S3].
Market Size, MV Segment Split, and Voltage-Class Definitions
Market Research Future's data-center transformer taxonomy sets low voltage below 2.5 kV, medium voltage at 2.5 to 36 kV, and high voltage above 36 kV, with the LV class still dominant by revenue but HV the fastest growing sub-segment as utility-side interconnect voltages rise [S5]. The 2.5 to 36 kV band is the working range of nearly every pad-mount and cast-resin dry-type transformer sitting on a data-center yard, and it is the same band pulled hardest by hyperscale switchgear lineups.
Spherical Insights puts the global data-center distribution transformer market at USD 7.1 B in 2025, projected to USD 13 B by 2035 at 6.24% CAGR (2026 to 2035), with North America as the highest-demand region [S4]. Distribution units are the downstream siblings of the MV step-down transformer, typically rated 625 to 3,750 kVA, sized to feed whips into the LV plant room. With AI racks crossing 100 kW and 1 MW, the kVA per rack is climbing fast, which is why unit counts in this 625 to 3,750 kVA band are scaling faster than the headline market CAGR [S4].
How Much of the MV Transformer Pipeline is Data Center Today
Under an accelerated AI scenario, data centers could account for up to 40% of the broader electrical-equipment market, surpassing the 160-week lead-time threshold on large transformers by Q1 2026 [S1]. That is a higher share than the segment's 2024 share, when global data-center power demand was already 415 TWh but transformer queues had not yet spiked to four-figure week counts [S1][S6].
For procurement planning, the practical reading is that MV transformer lead times in the 80 to 160+ week range now cover a disproportionate share of new orders, and the share is moving up, not down, as more hyperscale campuses break ground in 2026 and 2027. The data-center transformer segment sits at roughly USD 9 B in 2026 against a global power-transformer TAM multiple times that figure, which is consistent with the 40% ceiling applying to broader equipment categories (switchgear, breakers, cables) rather than transformers alone [S2][S1]. Operators who treat transformer slot booking as a 24-month critical-path activity, not a 6-month purchase order, are matching the lead-time reality [S8].
Lead Times, Capacity Bottlenecks, and the Procurement Reality
Transformer and switchgear procurement has become a structural constraint on 2026 data-center development, not a routine supply-chain blip, with utilities and hyperscalers competing for the same pool of large-MVA units [S8]. Lead times on 138 kV, 230 kV, and 345 kV generator step-up units now exceed 160 weeks at several Tier-1 OEMs, and 69 kV to 138 kV substation transformers are running 100 to 150 weeks, which means a slot booked in October 2026 lands on a yard in late 2029 or 2030 [S1][S8].
The 2026 build-out is therefore pulling on 2028 and 2029 transformer production, and on the same copper, GOES electrical steel, and pressure transmitter instrument air that downstream skid builders need for the LTC and cooling controls. Operators responding to the crunch are doing three concrete things: ordering transformers before land closing, accepting higher-impedance or k-factor-rated units in exchange for shorter queues, and standardising on 2.5 MVA to 10 MVA dry-type transformer cast-resin designs that ship from a different, less congested production line than oil-immersed units [S8].
Selection Criteria: Oil-Immersed vs Dry-Type for Data-Center Yards
Liquid-immersed transformers lead the data-center transformer revenue mix in 2026, used where the MVA rating and outdoor yard footprint make oil cooling the only practical option; dry-type units are the fastest-growing sub-segment, preferred for indoor substations and edge sites where fire-surpression code and footprint rules push against oil [S2]. A first-pass MV transformer selection for a 2026 hyperscale build should score on five criteria: MVA rating (typically 20 to 250 MVA at the substation edge, 2.5 to 10 MVA at the building MV/LV interface), impedance tolerance (often 5.5% to 8% to limit fault current on 2.5 MVA to 5 MVA units), cooling class (ONAN/ONAF for outdoor oil, AN/AF for indoor dry), sound level (55 dB to 65 dB at 1 m for neighbours), and partial-discharge performance below 10 pC for cast-resin designs at 2.5 to 36 kV [S2][S5].
For indoor LV/MV rooms adjacent to white space, cast-resin dry-type is the default: lower flammability, no oil containment pit, and faster permitting under NFPA 850 and most European equivalents. For outdoor substations stepping down from 69 kV or 138 kV to 13.8 kV or 25 kV, oil-immersed is the default: higher continuous kVA per footprint, lower first cost per MVA, and proven service life past 30 years with proper oil testing. The "right" choice is a function of site, code, and lead time, not a universal preference [S2][S5].
Adjacent Bottlenecks: Switchgear, Cables, and Busbar
The same 2026 demand spike that lifted transformer lead times above 160 weeks also pushed MV switchgear (38 kV, 27 kV, 15 kV class) queues out by 40 to 70 weeks and pulled copper busbar pricing with it, so a transformer booked for 2029 delivery is useless without a matching lineup of medium-voltage VFD-compatible switchgear and bus duct [S1][S8]. Procurement teams that bundled switchgear, transformers, and busbar into a single OEM frame agreement in 2024 to 2025 are now sitting on a materially better delivery position than teams that ran a separate RFP for each subsystem in 2026.
Material inputs are tightening in parallel. GOES (grain-oriented electrical steel) capacity additions are limited to a handful of mills, copper busbar supply is gated by rod and extrusion capacity, and data logger commissioning tools for transformer thermal and dissolved-gas monitoring are seeing demand spikes as more units ship before their associated substation civil works are complete [S1]. The practical downstream effect is that delivery dates on MV transformers are no longer the binding constraint alone; the binding constraint is the slowest item in the full substation bill of materials, which often turns out to be the 138 kV breaker or the station service transformer [S8].
Standards, Sourcing Signals, and Trackable Next Nodes
Data-center transformer specs typically reference IEEE C57.12.00 for general-purpose liquid-immersed units, IEEE C57.12.01 for dry-type, IEEE 519 for harmonic limits on the LV side, and IEC 60076-1 through 60076-5 for international builds, with DOE 10 CFR 431 efficiency tiers and EU Tier-2 (548/2014) governing no-load and load losses [S5]. Voltage class 2.5 to 36 kV is the working definition of MV in the data-center transformer taxonomy used across current market reports, and the same band covers the cast-resin units that indoor LV/MV rooms now standardise on [S5].
Two signals to watch over the next two quarters: any new Tier-1 OEM announcement of a 2027 or 2028 MV transformer production-line expansion at the 50 MVA to 250 MVA rating band, and any utility-side regulatory filing in the U.S. PJM, MISO, or ERCOT footprints that allocates dedicated transformer capacity to data-center loads ahead of general-customer queues. The transformer segment is on track to roughly double by 2030 (USD 14.5 B to USD 15.5 B depending on forecaster) [S2][S3], and 2026 procurement decisions made under 160-week lead times will determine which operators actually build, and which ones sit on shovel-ready sites waiting for a 138 kV unit. Related reading on the rack-density side driving this demand: AI Workloads Push 2026 Data Center Rack Density Past 100 kW, and on the per-MW spend implications: Data Center Power Equipment Spending Per MW of IT Load, 2026.