Your IT needs to grow. But the power grid is at capacity.

Your IT needs to grow, but you can't get any extra power. The grid operator or your data center provider has no more capacity. In more and more places across the Netherlands, getting a larger connection to the power grid can take years. That suddenly makes expanding your IT infrastructure a serious business risk. Because what happens if your organization needs more capacity, but the power grid can't deliver it?

Just how dire the situation is becomes clear from a recent ruling by the preliminary relief judge in Arnhem. In 2021, the Australian company Goodman applied for a power connection for its data centers and received a commitment from TenneT. Yet that connection does not have to be provided for the time being. Goodman may have to wait until 2033 or even 2035.

There is intense debate about the causes of grid congestion and who is responsible for it. But that doesn't make a single extra kilowatt available. The reality is that companies must plan their growth within the limits of the power capacity they already have.

And that is precisely where an opportunity lies that is often overlooked.

Data centers are evaluated, among other things, based on their Power Usage Effectiveness (PUE). This metric measures how much energy a data center consumes in total relative to the energy consumption of its IT equipment. The lower the PUE, the more efficient the supporting infrastructure, such as cooling and power supply.

But the biggest waste often isn't found in that supporting infrastructure. It's in the IT equipment itself.

A server that's barely doing anything still consumes about 50 to 67 percent of the power it needs at full load. Modern processors can scale back significantly, but memory, power supplies, and other components continue to consume energy — even when the server is largely idle.

This means that many IT environments consume a significant portion of their available power capacity without receiving commensurate performance in return.

At Sciante, we measure the actual energy consumption of IT environments. In doing so, we regularly find that the same tasks can be performed using significantly less power.

So the question isn't just how to get more power, but how much extra IT capacity you can extract from the power you already have.

In this blog, I'll show you where that potential lies and how you can expand your IT without needing a higher-capacity power connection.

You're paying for capacity that's actually holding your growth back.

A large portion of your IT capacity sits unused in the rack. Yet you're still paying for it — not just for the hardware, but also for electricity, cooling, and software licenses. Meanwhile, it's precisely the lack of available power that can make further growth impossible.

This isn't a matter of poor procurement. On-premises and colocation infrastructure is typically purchased for a five-year term. The capacity must still be sufficient at the end of that period. Expected business growth, new functionality, and more resource-intensive software are therefore factored in upfront. A sensible choice, since expanding mid-term is costly.

But what does that capacity actually look like after five years?

When we measure IT environments, an average utilization of just 15 to 20 percent turns out to be quite normal. That means the majority of the available computing power remains unused, while the equipment still consumes energy.

Suppose you have ten servers that are utilized at an average of 20 percent. If you securely consolidate the same workloads onto four servers, their average utilization rises to 50 percent. Depending on the configuration, this can roughly halve energy consumption.

And that creates something you can't order from your utility provider: immediately available power capacity.

You can use that to expand your existing business processes — without an additional connection and without waiting years.

But the benefits go far beyond electricity. When it's time for the next round of hardware replacements, you'll need to purchase significantly less hardware. Furthermore, with virtualization, Windows Server, and SQL Server, you often pay for licenses based on licensed capacity. Reducing the required capacity can therefore also significantly lower your licensing costs when renewing your licenses.

The savings on electricity are therefore likely the smallest benefit. Less hardware and fewer licenses can save you a fortune. But the greatest value may lie elsewhere.

Your business can continue to grow, even if the utility provider cannot supply you with additional power in the coming years.

A few minutes of peak load determine your hardware budget for years.

A few minutes of peak load can determine how much hardware you'll need for years to come — and how much power that equipment continuously consumes. Yet those peaks are often completely unnecessary. The result: capacity that remains unused most of the time but, when your power connection is maxed out, prevents you from growing further.

Consider an SQL Server cluster where all virtual servers start their maintenance jobs exactly on the hour. The load on processors, storage, and the network spikes simultaneously. The system must be able to handle that peak, even if it lasts only a few minutes.

By intelligently spreading those jobs out over the hour, the combined peak load can drop dramatically — without a single job being skipped or taking longer to complete.

We see the same thing with backups, ETL processes, and copying production environments to acceptance environments. Resource-intensive tasks compete for the same resources and slow each other down. The solution may seem to be more capacity, but a different schedule is often sufficient.

But there's also a reverse problem.

Workloads that are primarily active during business hours often run on different physical hardware than those that operate at night or on weekends. Both environments have their own capacity, which remains largely idle outside their peak hours.

Why reserve hardware, power, and licenses twice for workloads that barely overlap in time?

The solution involves two steps: spreading workloads across the same hardware and, conversely, combining workloads across different hardware.

This allows you to utilize available capacity much more evenly throughout all 168 hours of the week. The total hardware capacity required decreases, as does energy consumption, without users having to wait for their applications.

This requires accurate measurements of actual resource consumption, including peaks, throughput times, and interdependencies — not only to determine the optimal configuration once, but also to adjust it as workloads change.

This way, you create room for growth within your existing power connection, without sacrificing performance or waiting years for additional grid capacity.