A new data centre might have everything it needs to get started. The land's been secured, construction is underway, and customers are already making plans to use its computing capacity. But there's one problem. The facility still doesn't have the electricity connection it needs to operate.
Data centre grid connection delays can leave planned cloud, AI and colocation capacity unavailable long after a project was expected to open. And with businesses increasingly depending on these services, the consequences can reach well beyond the companies building the facilities.
In November 2026, Norway's electricity grid operator introduced temporary restrictions on new data-centre connections. It's one of several developments showing how difficult it can be to turn planned infrastructure into something customers can actually use.
For enterprises planning future workloads, there's an important difference between a provider having electricity capacity reserved and having a working data centre ready to deliver services.
Why Are Data Centres Struggling to Get Grid Connections?
Getting a data centre connected to the electricity grid isn't quite as simple as asking for power and waiting for someone to switch it on. The local network needs enough capacity to handle the facility's electricity use without affecting other customers or the stability of the wider grid.
If the network can't support that demand, new cables, substations or other upgrades may be needed before the connection can go ahead. Those projects take time, and the data-centre developer doesn't necessarily control when they'll be finished.
The International Energy Agency (IEA) estimated in its 2025 Energy and AI report that grid constraints could delay around 20% of global data-centre capacity planned for construction by 2030. That's a risk estimate rather than a prediction that all those projects will be delayed. But grid connection capacity isn't the only problem.
Operators also need to know whether new facilities can connect safely and how they'll behave when something goes wrong with the electricity network. Norway offers a recent example. On 9 November 2026, grid operator Statnett announced a temporary halt to new data-centre capacity reservations, alongside restrictions on certain technical approval applications.
Statnett explained that data centres can place unusual demands on the power system, particularly when their electricity use changes quickly. It expects to lift the restrictions no later than 1 August 2027, although existing connections and some applications aren't affected.
So even a well-funded project with a clear construction schedule can run into delays because the electricity network isn't ready to support it.
What Does a Data Centre's Grid Connection Status Actually Mean?
When a provider announces plans for a new facility, it may mention the amount of electricity capacity it's secured. But that figure doesn't necessarily tell customers how close the data centre is to opening. There are several stages between requesting electricity and delivering computing services.
The exact process depends on the country and grid operator, but the differences are useful when assessing a project's progress.
Requested and reserved grid capacity
A grid connection request is usually one of the first steps. The developer tells the network operator how much electricity it needs and where it wants to connect. The operator then assesses whether the network can support the request and what work may be needed. Being in the connection queue doesn't mean the developer has secured electricity.
It means the project has entered the process of asking for it. A capacity reservation is a step further. The grid operator has set aside a certain amount of network capacity for the project, usually under specific conditions. However, the developer may still need technical approvals, network upgrades or other work before the facility can connect.
Statnett's November figures show how different these stages can be. Norway had around 800 MW of data-centre capacity already connected to the grid, compared with 4,200 MW reserved and another 9,100 MW waiting in the capacity queue. Those figures describe very different levels of readiness. They certainly don't represent 14,100 MW of operational data centres.
Approved connections and energised supply
Once a project has progressed through the relevant checks, the developer may receive a connection approval or enter a formal agreement with the network operator. This can establish important details about the connection, including its planned capacity, technical conditions and delivery arrangements.
But it doesn't mean the facility can start using that electricity immediately. There may still be physical work to finish. The developer might be waiting for a substation upgrade, for example, or for equipment to be installed and tested. An energised grid connection means the electrical connection has been made live.
It's a much stronger sign of progress because electricity can now be supplied through the connection. Even then, the amount of power available may depend on the connection agreement and whether the facility is being brought online in stages.
Operational power versus available computing capacity
There's another step that can be easy to overlook. A data centre having electricity doesn't automatically mean its computing infrastructure is ready. The facility still needs working cooling systems, power distribution equipment, servers, networking and storage.
These systems need to be installed, tested and commissioned before the operator can deliver the services customers are expecting. And operational computing capacity isn't measured in quite the same way as electricity capacity. Some of the power entering a data centre is used to run cooling and other building systems, rather than the IT equipment itself.
So a facility with a live electricity connection and a stated power capacity may still have less computing capacity available than its long-term development plans suggest. For an enterprise buyer, the useful question isn't simply how much electricity a provider has secured.
It's how much of the promised computing capacity is actually ready to use, and when the rest is expected to become available.
What Happens When Grid Connections Are Delayed?
The consequences of a delayed connection depend on how far the project has progressed and whether customers have already made plans around its expected opening date. A delay during the early planning stages might change where a developer decides to build.
But if businesses have already committed to using the facility, the same problem can affect their own infrastructure projects.
Construction and location decisions
A developer might have found an ideal location for a data centre, with suitable land, good network connectivity and access to customers. But if the local electricity grid can't support the facility, the project could face years of additional work. Sometimes the developer may choose to build somewhere else.
That isn't always an easy decision, particularly when a project needs to meet specific requirements around latency, data residency or access to existing infrastructure. Ireland's December 2025 data-centre connection policy shows how location can influence the process. Its energy regulator introduced conditions for new connections that include assessments of local grid constraints.
A project that's suitable for one part of the country may therefore face different connection requirements somewhere else.
Cloud, AI and colocation availability
For customers, data centre construction delays can mean waiting longer for the infrastructure they've planned to use. An enterprise preparing to move workloads into a new cloud region might need to keep its existing systems running.
A company reserving colocation space could find that the building is progressing, but the power needed for its equipment won't be available on time. The same problem applies to planned AI capacity. Businesses may have deployment schedules built around access to computing resources that a provider expects to bring online in the future.
If the connection is delayed, those plans may need to change. And the closer an enterprise is to its intended deployment date, the fewer alternatives it may have. That's why cloud capacity planning needs to account for the development stage of the facilities behind future service commitments, particularly when those services depend on infrastructure that hasn't opened yet.
How Can Enterprises Assess Grid Connection Readiness?
Enterprises can't control how quickly a grid operator connects a data centre. But they can ask for better evidence before making commitments that depend on a provider's future capacity. The starting point is understanding what the provider has already secured, what's still being worked on and which parts of the delivery schedule depend on someone outside its business.
Verify the provider's connection milestones
If a provider is offering capacity in a facility that's still being developed, ask what stage its electricity connection has reached. Has the developer submitted an application, received a reservation, signed a connection agreement or already energised the site? These are different milestones, and each tells you something about how much work remains.
It's also worth asking which grid operator is responsible for the connection and whether the provider can share evidence confirming its status. For larger commitments, infrastructure teams should establish:
- Whether the required electricity capacity has been formally reserved or contracted.
- Which technical approvals and network upgrades are still outstanding.
- Whether the connection date has been confirmed by the grid operator.
- Whether the facility's planned opening depends on work that hasn't started or been completed.
Not every document will be available to customers, particularly where commercial agreements are confidential. But providers should still be able to explain their connection status and the main dependencies affecting delivery.
Check what the delivery commitment covers
A promised opening date isn't necessarily the same as a commitment to deliver a particular service. For example, a provider might expect to open the first part of a data centre in 2027, with additional capacity becoming available later. If your workloads depend on that second phase, the first opening date won't tell you much about when you can deploy them.
The same applies when a facility has enough electricity for its initial operations but needs a further connection upgrade to support its full planned capacity. When reviewing colocation contracts or future cloud capacity agreements, establish exactly what will be available on the agreed date.
That includes the amount of usable computing capacity, the services covered and any conditions that could affect delivery. It's also worth checking what happens if the provider misses that date. Contractual protections will differ, but the agreement should make clear how delays affect the customer's commitments and available options.
Plan for capacity that arrives late
Even with a well-developed project, connection dates can change. So enterprises should consider how much flexibility they have if their preferred infrastructure isn't ready when expected. That could mean keeping existing systems running for longer, splitting a deployment into stages or identifying another region that can support the workload.
The right choice depends on what the application needs. Some workloads can move between locations fairly easily, while others have stricter requirements around performance, security or where data can be stored. For infrastructure procurement, the important thing is to make those decisions while there's still time to act.
If a planned data centre is essential to a major migration or AI deployment, its outstanding grid connection work should be part of the project's risk assessment. Otherwise, a delay outside the enterprise's control could end up determining its own delivery schedule.
Are Grid Connection Rules Changing?
Grid operators are also changing how they deal with the growing number of data-centre connection requests. In Britain, Ofgem proposed a commitment fee and new progress milestones in July 2026. Its consultation on speculative data-centre projects followed a rise in electricity-demand connection applications from 41 GW to 125 GW in less than a year.
The proposals are intended to stop projects that aren't progressing from holding capacity that others could use. Norway is taking a different approach. Alongside Statnett's temporary restrictions, the government announced on 10 November that it's developing a new data-centre licensing regime.
The licensing system is still being developed, so it shouldn't be treated as an existing requirement. Both developments point towards closer checks on which projects are ready to proceed and what their connections could mean for the electricity network.
For enterprises, this adds another reason to pay attention to grid connection requirements when evaluating future infrastructure. A provider's development plans may need to change as the rules governing its connection change.
Final Thoughts: Planned Capacity Isn't Guaranteed Capacity
Data centre grid connection delays aren't just a problem for the companies building the facilities. They can affect when businesses get access to the cloud, AI and colocation services they're planning to use. The challenge is knowing how much confidence to place in a provider's promised delivery date.
A connection request, a reservation and an operational facility are very different things, even when they're all described as future capacity. As grid operators tighten their requirements, enterprises may need to ask for stronger evidence before committing to infrastructure that hasn't been built.
And as those decisions become more closely tied to electricity networks, understanding the wider changes shaping enterprise infrastructure will be just as important. EM360Tech helps technology leaders keep up with those changes and make more informed decisions about the systems their businesses depend on.
Comments ( 0 )