A Few Practical Foundation Tips for Data Center Projects

One thing I have seen come up again and again in data center projects is that the foundation strategy can look simple at the beginning, then become one of the biggest cost surprises later.

The site is usually selected for power, fibre, water, access, climate, zoning, or commercial reasons. All of that matters. But once the land is chosen, the project has to deal with the soil that is actually there.

That is where many assumptions start to break.

For a normal building, some settlement or variation in ground behaviour may be manageable. For a data center, the tolerance is much lower. Rack halls carry high sustained loads. Plant areas, generators, UPS rooms, and cooling equipment create heavy concentrated loads. Even small differential settlement can affect raised floors, equipment alignment, cooling distribution, and long-term operation.

So before deciding between soil improvement and piling, I think teams should slow down and ask a few practical questions.

1. Is the ground investigation good enough for the decision being made?

A few boreholes may not tell the full story of a large data center site. Boreholes give information at specific points. The real risk is what happens between those points.

A good geotechnical report should not only provide logs. It should give a clear design soil profile that the structural engineer can actually work with.

Where the ground is variable, adding CPT testing can be very useful. It helps fill the gaps between boreholes and gives a better picture of soil strength, layering, and pore pressure.

Saving money on investigation at the beginning can be very expensive later.

2. Are we comparing the full foundation system, or just the headline price?

Soil improvement may look cheaper than piling at first. Sometimes it is. But the full comparison should include more than the installation cost.

You need to look at transfer layers, footing size, concrete volume, reinforcement, spoil disposal, dewatering, groundwater treatment, approval risk, and programme impact.

A cheaper method can become less attractive once all supporting works are included.

This is especially true when the site has contaminated soil or a high groundwater table. Excavated material may not just be normal spoil. Water may need treatment before discharge. These items can change the cost picture quickly.

3. Is the method suitable for concentrated loads?

Soil improvement can work well under more uniform loads, depending on the ground and method used. But data centers also have heavy point loads from plant zones and structural columns.

If the system needs a thick gravel layer or larger footing to spread those loads, that should be checked carefully. The load still has to go somewhere. The question is whether the system transfers it predictably enough.

In some cases, piling may give a clearer load path and better settlement control, especially under heavily loaded columns.

4. Is the solution clearly covered by recognised codes?

Some ground improvement systems are technically useful, but can sit in an approval grey area. They may be called ground improvement, but in construction they can look very similar to piles.

If the authority later treats the method as piling, the design may need to be checked against pile design requirements. That can create delay or redesign risk.

This does not mean those systems should never be used. It means the approval route should be understood before the method is locked in.

5. Are all foundation elements designed for their real loads?

One shortcut is to take the worst column load and apply it across too much of the foundation. It feels safe, but on a large site it can waste a lot of concrete, steel, drilling time, and disposal cost.

A better approach is to design pile groups or foundation elements based on the loads they actually carry, with a sensible allowance for uncertainty.

6. Review the foundation strategy before it becomes fixed

The best time to challenge the foundation concept is early. Once the contractor’s method is priced, scheduled, and built into the programme, changing it becomes much harder.

At concept stage, a review can still improve the decision. During construction, the same issue becomes a delay.

I came across an interesting article that goes deeper into this topic from a data center foundation design perspective. Worth a read here: Data center foundations: a developer's guide to ground risk

Curious to hear from others here.

• In your experience, do project teams usually review foundation strategy early enough?
• Have you seen soil improvement look cheaper at first, but become more expensive once transfer layers, groundwater, spoil disposal, or approval risk were included?
• For data centers, where do you see more risk: choosing piling too quickly, or pushing soil improvement too far?