Data centers are pulling hard on the electrical equipment supply chain, and the order books show it. A new plant takes years and a large check. Most manufacturers have meaningful capacity hiding in the plants they already run, lost to waiting, rework, slow engineering, and schedules nobody trusts. This post shows you how to find it and get it back.
The short answer
- Demand is real and documented: the IEA expects global data-center electricity use to more than double by 2030, and power transformer lead times averaged 128 weeks in mid-2025.
- Your output is set by one constraint at a time. Find it, protect it, and stop feeding it work that isn't ready.
- Upstream speed matters as much as the shop floor: slow quoting and engineering starve the constraint or flood it with late changes.
What the demand picture actually looks like
The IEA's Energy and AI report puts data centers at about 1.5% of world electricity use in 2024, or 415 TWh, and projects roughly 945 TWh by 2030. In the United States, it says data centers account for nearly half of electricity demand growth through 2030. It also notes that wait times for transformers and cables have doubled in the past three years.
US government numbers point the same way. A Lawrence Berkeley National Laboratory report released by DOE found data centers used about 4.4% of US electricity in 2023 and could reach 6.7% to 12% by 2028.
On the equipment side, Wood Mackenzie reported in August 2025 that US power transformer demand had risen 116% since 2019 and distribution transformer demand 41%, with expected 2025 supply shortfalls of 30% and 10%. Its second-quarter 2025 survey put average lead times at 128 weeks for power transformers, 143 weeks for generator step-up units, and 44 weeks for switchgear.
| Equipment | Average lead time, Q2 2025 |
|---|---|
| Generator step-up transformers | 143 weeks |
| Power transformers | 128 weeks |
| Switchgear | 44 weeks |
Source: Wood Mackenzie survey averages. Quotes for specific ratings and data-center specifications can differ widely.
Manufacturers have responded with capital. Wood Mackenzie counts about $1.8 billion in announced North American capacity investment since 2023. But new lines take time, and the GAO found utilities citing limited manufacturing capacity and labor and material shortages as key constraints. Capital alone won't close the gap fast enough.
Source: Wood Mackenzie supply chain survey, Q2 2025.
View the data
| Equipment | Average lead time |
|---|---|
| Generator step-up transformers | 143 weeks |
| Power transformers | 128 weeks |
| Switchgear | 44 weeks |
Step 1: Find the real constraint
Every plant has one resource that sets its pace. In transformer and switchgear work it's often a test bay, a winding cell, a vapor-phase or drying oven, a tanking or welding station, or a scarce group of design engineers. Gaining an hour anywhere else adds nothing to output. Gaining an hour at the constraint adds an hour of plant output.
Three quick ways to find it:
- Follow the queue. Walk the floor and note where work sits waiting. The longest queue usually sits just in front of the constraint.
- Compare load to capacity. For each major work center, divide planned hours for the next 13 weeks by available hours. The one closest to or above 100% is your candidate.
- Ask what expedites. Whatever resource planners argue over in the weekly meeting is usually the one that matters.
Don't forget engineering. In engineer-to-order work, the constraint is often a design or drafting team, and the floor sits idle waiting on drawings and bills of materials.
Step 2: Recover lost time at the constraint
Once you know the constraint, measure how its available hours are spent. Typical losses include waiting for material or paperwork, setups, failed tests and retests, missing documentation, and breaks with no relief cover.
An illustrative example shows the scale. Say two test bays have 5,000 available hours each, or 10,000 in total, and each unit needs 80 bay hours. Maximum output is 10,000 ÷ 80 = 125 units a year. If 15% of bay time, or 1,500 hours, is lost to waiting and retests, and you recover half of it, you gain 750 hours. That's 750 ÷ 80 ≈ 9 more units a year, about 7.5% more output, with no new building.
Common fixes:
- Stage everything a unit needs before it reaches the constraint: parts, drawings, test plans, and customer witness dates.
- Cover breaks and shift changes so the constraint doesn't stop.
- Put quality checks before the constraint, so it never spends time on work that will fail.
- Track the constraint's hours daily, not monthly.
Step 3: Speed up quoting and engineering
When lead times run past two years for some products, customers lock in slots early. The manufacturer that answers an RFQ with a credible price and a realistic delivery date first has an edge. Slow engineering also hurts the floor: late drawings and changes after release cause the waiting and rework you just worked to remove.
Focus on three things. Reuse past designs instead of starting from a blank sheet. Write down the configuration and pricing rules your senior estimators carry in their heads. And check promised dates against actual constraint load before you quote them. We cover this in more depth in our post on engineer-to-order quoting.
Step 4: Schedule to the constraint
A schedule that loads every work center to its maximum creates piles of work in progress and hides the real bottleneck. Instead, schedule the constraint first, then release work to the floor only as fast as the constraint can consume it.
Little's Law makes the payoff concrete: lead time equals work in progress divided by throughput. Illustrative example: if a plant finishes 10 units a week with 120 units in progress, average lead time is 120 ÷ 10 = 12 weeks. Cut work in progress to 80 at the same output, and lead time falls to 8 weeks, which frees cash and floor space.
Step 5: Standardize, then automate
Standardization means fewer one-off designs, common sub-assemblies, and standard test procedures. It shortens engineering, cuts setups, and makes work more predictable. Offer standard options and price true custom work to reflect what it costs you.
Automation works best after the process is stable. Good early targets are the information flows around production: material readiness, test records, document packages, and status reporting. Physical automation, such as winding or cutting equipment, can help at the constraint, but it costs more and takes longer to install.
Before you pick a starting point, it helps to see where your capacity is being lost across quoting, engineering, scheduling, and supply. A quick, honest self-assessment across those areas usually points to the first move.
Questions to ask your team this month
- Which resource sets our output today? Can every manager name the same one?
- How many hours did the constraint lose last week, and to what? If nobody knows, start measuring.
- How long does it take us to return a priced quote with a delivery date? Measure it from RFQ receipt to sent quote.
- How much work in progress are we carrying? Is it rising or falling against output?
- What share of orders are one-off designs? Which could become standard options?
- Do we check delivery promises against real capacity before we quote?
Where to go from here
Start by naming your constraint and measuring its lost hours for two weeks. If you want help finding where capacity is leaking across commercial, engineering, and production, tell us about your operation. We offer this work, so weigh our view accordingly. You can also see how we approach the sector on our Energy Infrastructure page.
Sources
- IEA: Energy and AI, executive summary
- U.S. Department of Energy: DOE releases new report evaluating increase in electricity demand from data centers
- Wood Mackenzie: Power and distribution transformers will face supply deficits of 30% and 10% in 2025
- Wood Mackenzie: Mind the gap, tackling supply-chain challenges in the electric T&D sector
- U.S. GAO: Electricity grid, DOE could better support industry efforts to ensure adequate transformer reserves