It’s natural, at the end of the year, to think about what the future holds.

Of course, as data center leaders, we do this all the time, gathering information about the future of the data center industry. What do we see?

Whether we’re examining the largest hyperscalers or small, on-premise enterprise facilities, the world of data centers is under unprecedented pressure.

We see that trends are converging, change is needed, and our industry will look very different in ten years – all of us are trying to think ahead, positioning ourselves to be at the forefront of where the industry is going.

The picture is becoming clear. In this article, let’s explore seven likely trends – that could change how all of us do business.

Trend one: IT equipment is getting hotter every 18 months, but data centers are built to last 20 years and this disconnect results in problems that get worse over time

IT equipment is hotter than ever before, the problem gets worse with every product refresh, and there’s no end in sight. To quantify the problem, we examined a decade's worth of off-the-shelf, 2U, 2 socket servers from a Tier 1 server manufacturer to see just how much power consumption (and heat production) increased over time.

  • Servers from 2012 had dual 570-870W power supplies
  • Servers from 2017 had dual 495-1100W power supplies
  • Servers from 2022 have dual 800W to 2400W power supplies.

Power supply capacity has more than doubled over ten years, and if that trend continues, servers will have 7000W power supplies by 2032. Could your data center cope?

As data centers are built to last twenty years or more, designers will have to start thinking ahead – either building in costly unused cooling capacity to support future equipment – or designing data centers to easily accommodate an expensive cooling refit every few years.

Neither approach is appealing due to complexity and costs – but workarounds like living with inadequate cooling by reducing rack density cut into value.

Trend two: Data center construction takes years, and with supply chain uncertainty, might be getting worse

Various vendors claim that they can build a data center in twelve to eighteen months. That may be true, but usually those estimates overlook assessment, planning, site selection, design and other pre-work before construction begins. Some of the largest data centers take five years or more, from initial concept to move-in.

To make matters worse, supply chain issues are affecting data construction costs and schedules. The price of steel has increased by nearly 50 percent in the past twelve months. Chillers, generators, and even lighting are on multi-month lead times.

Addressing these challenges will become the new norm. Already, the industry is beginning to turn toward pre-assembling data center components in factories and then shipping them to sites to accelerate construction.

They’re also ordering essential materials and components well ahead of schedule and stockpiling them. We see these approaches and many others becoming the new norm for the next decade.

Trend three: There's going to be more demand for resource consumption efficiency

Waste, especially of resources provided by municipalities, of all kinds is becoming a serious concern for data center designers, construction companies, and operators.

From initial design to decommissioning, organizations are looking for ways to reduce resource consumption:

  • Innovative designs can be built with fewer raw materials
  • Pre-assembling reduces material waste, cutting costs
  • Improved cooling technologies reduce electricity consumption
  • Corporate ESG initiatives require sustainable energy and water consumption.

This trend won’t go away. Citizens and municipalities, corporate boards and countries will demand better utilization of resources by the data center sector.

Water is becoming a particularly critical concern for the industry, forcing the largest companies to accelerate their efforts to become water-neutral or water-positive.

Focusing on efficiency isn’t just good PR, it’s also good economically – for example, improving PUE can cut operating costs by millions of dollars.

Since legacy data centers are often horribly inefficient, forward-thinking organizations will be scrutinizing their data center investments with efficiency at the forefront.

Organizations that aren’t adapting to both external and business demands for efficiency are not only leaving money on the table, they’re impacting citizens and governments in ways that will cause friction for their operations.

Trend four: People will have to design and build with weather unpredictability in mind

In the past, during site selection, data center designers were told to consider ten years of weather data in their design considerations. However, recent events have exposed the limitations of conventional climate planning.

In 2021, during an extreme freeze in Texas, data centers failed because of inadequately winterized climate control and electrical systems.

A year later, Twitter, Oracle and Google data centers failed in the UK and California due to extreme heat. A disaster related to a weather, climate or water hazard has occurred every day on average over the past 50 years – and the rate of disasters is accelerating.

Weather is becoming increasingly unpredictable. Designers and operators are already assessing the need to retrofit or decommission old data centers because they simply cannot cope with new weather extremes.

Over the next decade, designers and operators will be forced to stay ahead of climate change with innovative designs that make it easier to defend against extreme weather events.

Trend five: There's increasing demand for small data centers ‒ for distributed computing at the Edge

Staying on top of data center construction means that we see announcement after announcement for enormous data centers.

Just in the past month, we’ve seen a 600 MW data center proposed outside London and a 500 MW facility proposed in Virginia.

There’s a narrative, pushed by some data center operators, that larger data centers are better than smaller ones – from an efficiency standpoint, among other things.

However, these data center announcements, though fantastic marketing, obscure reality. Two factors are driving demand for small data centers.

Organizations need data center capacity to serve their customers as soon as possible. Many new digital services are latency-sensitive, need distributed compute to provide near-realtime analytics, or data might need to be stored in a particular country due to data governance regulations.

Those services require data center capacity at the Edge – as close as possible to the customer. Distributed data centers, smaller and more agile, are becoming a new trend in the industry that will only get stronger.

Organizations also know that enormous data centers take enormous resources. Site selection, construction issues, supply chain challenges and many other factors limit the construction of enormous data centers.

Building these largest data centers takes years, and many customers simply aren’t interested in waiting.

We’ll continue to see 500+ MW data centers being announced, but behind the scenes, companies will be building thousands of 1-10 MW data centers.

Trend six: Some industries will be more affected by these trends than others

As we all know, not every industry has its finger on the pulse of data center innovation.

Some industries, whether due to legacy thinking or sunk costs, insist on utilizing on-premise data centers rather than colo or cloud, lagging behind where data center capabilities are concerned.

Two of these verticals are finance and healthcare. Both industries have substantial investments in owned data centers, many of which are outdated and in need of improvement.

These lagging industries are already having problems. We’ve already seen data center failures at the UK’s National Health Service, pointing to a need for data center modernization.

We’ll see financial institutions and healthcare systems evaluating their options as their data centers become more dated, more depreciated, and more dangerous.

These industries will be under pressure to retrofit and modernize their data centers. We’ll also see them accelerating their consideration of alternatives.

Trend seven: More and more organizations will see these challenges and seriously wonder whether they should outsource their data centers to an expert

When it comes to accelerating construction, solving efficiency problems, coping with climate change, and building dozens or hundreds of data centers, some organizations will be more successful than others.

Our view is that the days of the do-it-yourself data center are fading away, replaced by highly expert data center vendors who know how to cope with dangerous risks, unprecedented complexity, and demands for greater efficiency and cost-controls.

Whether organizations are looking for help from colo operators or want to partner with leading data center construction companies, we’ll see a shift toward hiring expertise rather than DIY.

Many of us are well positioned to take advantage of these trends. At Nautilus, we’re leading the way toward a more agile, more efficient, more successful approach to data center design, construction, and operations.

Our data centers cut energy consumption by 70 percent, cut water consumption to zero, and can be built in a year. Click here to learn more about the Nautilus difference.