An AI data center can be physically close to finished and still be waiting for electricity.
The building may be ready. The servers may already be ordered. But the project may still need a substation, a large transformer, high-voltage cable, protection equipment, and trained crews to install and test everything.
This is why grid delays can hold up a data center even when the data-center construction itself is moving quickly.
The difficult part is not only making the equipment. The equipment also has to be installed, tested, and connected before the site can receive power.
After reading this article, you should be able to explain why transformers and cables can take years to arrive, how to tell whether a factory expansion will really increase supply, and why a large order book does not automatically mean faster delivery or higher profit.
Why Can the Grid Take Longer Than the Data Center?
A large data center is not connected to the grid by simply running one cable from the street.
The utility may need to study the new load, decide where it can connect safely, build or expand a substation, install transformers and switchgear, add high-voltage cable, and sometimes reinforce the transmission network.
A substation is a facility where electricity is switched, protected, and changed to the voltage needed for another part of the grid.
A transformer changes electrical voltage. Large power transformers are especially important because they connect different voltage levels across the grid and at large industrial sites.
Switchgear is equipment that turns electrical circuits on and off and protects them when faults occur.
After the equipment is installed, the project still needs commissioning. Commissioning means testing the installed system to make sure the equipment works safely together before the site begins full operation.
The International Energy Agency reports that some cables can take roughly two to three years to procure, large power transformers can take up to about four years, and some direct-current cables can take more than five years.[1]
Those timelines help explain why a data-center building can be ready before the grid connection is.
Why Can’t Manufacturers Simply Make More Transformers?
Large power transformers are not mass-market products made in millions of identical units.
They are often designed for a specific voltage, capacity, site, cooling method, and utility standard. That means the factory needs specialized materials, specialized machinery, testing equipment, and experienced workers.
One example is a bushing. A bushing is an insulated component that lets a high-voltage conductor pass safely through the transformer tank without creating an electrical fault.
Manufacturers also need electrical steel, copper or aluminum conductors, insulation systems, cooling equipment, and large testing bays.
Even transportation can be difficult. A large transformer may weigh hundreds of tons, so moving it to the site can require special railcars, trucks, routes, and permits.
For this reason, adding a factory building is only one part of increasing supply. The company also needs trained workers, qualified suppliers, testing equipment, and enough customer demand to keep the new factory busy.
A Simple Way to Follow the Bottleneck
Orders → Factory Capacity → Equipment Delivery → Installation → Testing → Power On
This is the one compressed chain worth remembering.
Orders tell us that customers want the equipment. Factory capacity tells us how much can actually be produced. Delivery tells us whether the equipment reaches the site. Installation and testing determine when the equipment becomes usable.
Why Are Developers Ordering Equipment So Early?
When delivery times stretch into several years, developers may try to reserve transformer or switchgear production slots long before a data center opens.
This is easier for a very large developer with strong financing. It can commit money early, reserve a place in the factory schedule, or buy equipment before the rest of the project is finished.
A smaller developer may not be able to do that. It may still be waiting for permits, financing, or a final utility agreement.
So long lead times can create a second advantage for large buyers: not only can they buy more equipment, they may also be able to secure it earlier.
How to Tell Whether a Supplier Can Benefit From the Shortage
A shortage can create strong demand, but that alone does not tell us whether a supplier will earn more money.
Four questions are useful.
- Is the product genuinely hard to get?
Check actual lead times, not only management language about “strong demand.” - Is the supplier adding real capacity?
Look for a new factory, production line, test bay, cable plant, or workforce program with a measurable output target. - Is delivery improving?
Revenue should start to rise as the new capacity begins producing equipment. - Is the business becoming more profitable?
Margins and cash flow should improve or at least remain healthy as more equipment is delivered.
If only the order book is growing while delivery times keep getting longer, the shortage may still be a problem for the supplier as well as for the customer.
GE Vernova: Large Orders, but the Scope Matters
GE Vernova sells power-generation equipment and grid equipment. Its grid-related products include transformers, substations, switchgear, software, and services.
In Q2 2026, GE Vernova reported total orders of $24.2 billion, total backlog of $176 billion, revenue of $11.1 billion, and free cash flow of $5.1 billion. It also said data-center orders in its Electrification business exceeded $5 billion during the first half of 2026.[2]
Electrification is GE Vernova’s business area that sells grid-related electrical equipment and systems.
The important detail is scope. The $176 billion backlog belongs to the whole company. It should not be described as $176 billion of data-center backlog.
The more useful question is whether strong data-center demand is helping the company increase revenue, improve margins, and generate cash while it expands production.
Siemens Energy: More Factory Capacity Is Beginning to Support More Revenue
Siemens Energy’s Grid Technologies business sells transformers, switchgear, substations, and HVDC systems.
HVDC means high-voltage direct current. It is used to move large amounts of electricity efficiently over long distances, including through subsea cables.
In Q3 FY2026, Grid Technologies reported orders of €5.367 billion, revenue of €3.624 billion, a profit margin of 19.8%, and an order backlog of €51 billion.[3]
The segment’s book-to-bill ratio was 1.48. Book-to-bill means new orders divided by revenue during the same period. A ratio above 1.0 means the company received new orders faster than it recognized revenue.
Siemens Energy also said that expanded production capacity supported revenue growth.
That is useful because it connects the factory expansion with an operating result. The company is not only announcing more capacity; it is beginning to deliver more equipment from that capacity.
Hitachi Energy: A Factory Announcement Becomes More Useful When the Output Is Clear
Hitachi Energy is part of Hitachi Ltd. It supplies transformers, high-voltage switchgear, grid automation, and HVDC systems.
Hitachi reported that Hitachi Energy’s backlog had grown from about $11 billion in FY2020 to about $60 billion in FY2025.[4]
The company is responding with a large manufacturing program. In September 2026, Hitachi Energy announced a new $528 million transformer factory in Mississippi. The company expects the new plant to more than double transformer production capacity compared with its nearby Crystal Springs facility and to create more than 700 jobs.[5]
This is a more useful announcement than simply saying “we are expanding.” It tells the reader where the plant will be, how much the company plans to spend, how much production should increase, and how many workers will be needed.
The next step is to check whether construction finishes on time and whether commercial production starts when the company expects.
Prysmian: Cables Are Needed for Both Electricity and Data
Prysmian supplies power cables as well as fiber and optical cable.
That matters because a data center needs two kinds of connection. It needs electricity, and it also needs high-capacity data connections.
In Q2 2026, Prysmian reported organic growth of 14.3% in Transmission, 13.0% in Power Grid, 9.1% in Industrial & Construction, and 18.0% in Digital Solutions. The company also reported an adjusted EBITDA margin of 15.4%.[6]
Utilization means the share of a factory’s available production capacity that is actually being used.
A new cable factory does not automatically create profit. The plant has to open on time, receive enough orders, run at a good utilization rate, and sell products at prices that cover its costs.
Quanta Services: Sometimes the Scarce Capacity Is People
Quanta Services does not mainly solve the bottleneck by manufacturing transformers. It designs, builds, upgrades, and maintains transmission lines, substations, and other large energy infrastructure.
Its important capacity is therefore partly human: engineers, lineworkers, electricians, supervisors, and project managers.
In Q2 2026, Quanta reported revenue of $9.56 billion, free cash flow of about $0.9 billion, $33.6 billion of remaining performance obligations, and $53.4 billion of total backlog.[7]
Remaining performance obligations, or RPO, are contracted work that the company still has to perform.
Quanta’s total backlog is broader than RPO. It also includes estimated future work under master service agreements, or MSAs. An MSA sets the rules and commercial terms for repeated work over time. It does not necessarily guarantee that every estimated future project will happen.
This is why one company’s backlog number should not be compared mechanically with another company’s backlog number.
Can Skilled Labor Become the Next Bottleneck?
Yes. A grid project needs people as well as equipment.
Utilities and contractors need workers who can install substations, connect medium-voltage and high-voltage systems, test protection equipment, operate heavy equipment, and commission the completed system.
Those skills take time to develop.
A transformer sitting at the construction site does not help the data center until trained crews install it, connect it, test it, and place it into service.
How to Judge a Factory Expansion
When a supplier announces a new factory, these questions make the announcement easier to evaluate.
- What exactly is being added?
A new factory, one new line, more test bays, a cable vessel, or a workforce program? - How much more can the company produce?
Look for an annual output target or a percentage increase. - When will commercial production begin?
The announcement date and the production date can be years apart. - Are customers already committing money?
Orders, deposits, reservation fees, and advance payments can reduce demand uncertainty. - Are critical materials and components available?
Copper, electrical steel, insulation, bushings, and specialized machinery can still slow production. - Will there be enough trained workers?
New buildings and equipment need people to operate them. - Can the company protect its profit margin?
Older fixed-price contracts can become less profitable if material and labor costs rise. - Does delivery improve after the expansion opens?
Check revenue, delivery times, margins, working capital, and cash flow.
Why Other Infrastructure Projects Can Feel the Same Shortage
Transformers, switchgear, cables, and electrical construction crews are not used only by AI data centers.
Water systems, factories, renewable-energy projects, transportation systems, and ordinary utility upgrades may need the same equipment and the same workers.
This means AI demand can add pressure to an already tight supply chain, but it should not be treated as the only cause of long lead times.
Grid replacement, electrification, new generation, material shortages, factory limits, approved-vendor rules, permitting, and logistics can all contribute.
A better question is therefore: Which part of the shared supply chain is actually limiting this project?
How to Check the Claims Yourself
You can verify many of these claims without a paid database.
Start with the supplier’s Investor Relations page. Open the latest earnings release, investor presentation, annual report, and earnings-call transcript.
Search for words such as capacity, factory, lead time, delivery, backlog conversion, advance payment, utilization, and skilled labor.
For each expansion, record four dates separately:
- announcement date
- construction start
- expected completion
- commercial production
Do not assume that a ribbon cutting means the plant has reached full production.
Also record the scope of every financial number. Is it company-wide, one segment, one product line, or data-center-specific?
Finally, check working capital. Working capital is the short-term money tied up in inventory, customer receivables, and supplier payables. A fast factory expansion can use a lot of cash before the additional production starts generating revenue.
What a Healthy Pattern Looks Like
- orders are growing across more than one customer;
- the capacity expansion has a measurable output target;
- commercial production starts close to the promised date;
- revenue grows as the new capacity comes online;
- profit margins remain stable or improve; and
- cash flow improves as deliveries increase.
What Should Make You More Cautious?
- the company announces “more capacity” but gives no output target;
- factory completion dates keep moving later;
- backlog rises while lead times get worse;
- revenue grows but margins fall sharply;
- inventory and capital spending consume most of the cash; or
- one very large customer or project accounts for most of the growth.
The Main Idea
The grid bottleneck is a real physical problem. Large transformers, cable systems, substations, and skilled construction crews all take time to add.
For a supplier, strong demand is only the beginning of the story. The company still has to expand production, deliver the equipment, install it successfully, and earn an acceptable return on that work.
A shortage becomes a business opportunity only when the supplier can turn it into reliable delivery, healthy margins, and cash.
Continue Reading
- How to Research AI Infrastructure Companies: From Orders to Cash Flow
- Who Gets Paid in the AI Data Center Boom?
- What Must Be Built to Power the AI Data Center Boom?
Key Terms
- lead time: the time between placing an order and receiving usable equipment
- substation: a facility that switches, protects, and changes the voltage of electricity
- transformer: equipment that changes electrical voltage
- bushing: an insulated component that safely carries a high-voltage conductor through grounded equipment
- switchgear: equipment that switches, protects, and isolates electrical circuits
- HVDC: high-voltage direct current, used for efficient long-distance or subsea transmission
- commissioning: final testing and verification before a system enters full operation
- backlog: accepted or contracted work that has not yet been fully delivered
- book-to-bill: new orders divided by revenue in the same period
- RPO: remaining performance obligations, contracted future work still to be performed
- MSA: master service agreement, a framework for repeated work under agreed terms
- utilization: the share of available production capacity actually being used
- working capital: short-term operating money tied to receivables, inventory, and payables
Sources
- IEA — Building the Future Transmission Grid
- GE Vernova — Q2 2026 results
- Siemens Energy — Q3 FY2026 results
- Hitachi Investor Day 2026 — Energy Business Strategy
- Hitachi Energy — Mississippi transformer factory, September 2026
- Prysmian — Q2 2026 results
- Quanta Services — Q2 2026 results
Status checked October 1, 2026. Grid-equipment shortages can have several causes, including demand growth, factory capacity, materials, approved suppliers, permitting, logistics, and skilled labor. Company backlog definitions also differ, so figures should be compared only after checking their scope.