data centre

Data centre management requires flood of ideas

Can data centres get better through liquid cooling? Billy MacInnes finds some companies thinking bigger and deeper
Image: Stockfresh

28 June 2024

Here’s one to file under the “you learn something new everyday” heading – or at the very least in the “I learn something knew everyday” category. What am I talking about, you might ask? I haven’t exactly narrowed it down that much so far. But to bowdlerise Shakespeare, it’s to be devoutly wished that we learn something new everyday.

So what did I learn today? I learned about underwater data centres. Let me repeat that, underwater data centres.

Here in Ireland, we’re all very familiar with the out-of-water ones, aka land-based data centres, primarily because this little island seems to be engaged in a bizarre strategy of siting as many of them here as possible to the point where we have more than anywhere else. According to some estimates, there are more than 70 data centres in Ireland and Dublin hosts around 25% of all data centres in Europe. We don’t know for sure, however, because the government doesn’t collect figures for them.

We all know the intense power requirements that data centres bring with them. In 2022, they accounted for 18% of total power consumption in Ireland, the same as used by all households in the country’s urban areas. It’s estimated that data centres could account for 30% of Ireland’s electricity consumption by 2030.

And yet here we are, on an island, surrounded by water (which is the definition of an island, to be fair). And that brings us back to underwater data centres, although the one I learned about was actually based off the coast of Scotland.

Project Natick, as it’s known, is a submersible data centre capsule that was deployed 120 miles off the coast of Scotland in 2018 by Microsoft. Filled with 12 racks of servers, it was brought back to the surface in 2020. According to Microsoft, six of the 855 servers in the capsule during its time underwater failed. By comparison, a simultaneous experiment on dry land lost eight out of 135 servers.

When it retrieved the capsule, Microsoft reported that it proved the concept of underwater data centres was feasible, as well as logistically, environmentally and economically practical.

Ben Cutler, a project manager in Microsoft’s Special Projects research group who leads Project Natick said: “Our failure rate in the water is one-eighth of what we see on land. I have an economic model that says if I lose so many servers per unit of time, I’m at least at parity with land. We are considerably better than that.”


According to this report in Datacenter Dynamics (DCD), Microsoft cited the steady external temperature of the seawater as a factor for the success of the project. The submersible capsule was also filled with inert nitrogen gas rather than the reactive oxygen gas found in the land-based data centre.

Sadly, Noelle Walsh, head of Microsoft’s cloud operations and innovation (CO+I) division, confirmed that it had no plans to build subsea data centres anywhere in the world. “My team worked on it, and it worked,” she told DCD. “We learned a lot about operations below sea level and vibration and impacts on the server. So we’ll apply those learnings to other cases.”

Security seems to be one of the strongest arguments against deploying data centres underwater. You only have to consider the vulnerability of underwater sea cables to appreciate the threat posed to submersible data centres. Although it’s probably worth pointing out that land-based versions aren’t invulnerable to security threats either.

Elsewhere, however, DCD notes that a company called Highlander has deployed a commercial 1,300 tonne facility in the sea near Hainan island, China. The system, designed to be used in shallow water, is submerged 35 meters underwater and the company claims it can process more than four million high-definition images within 30 seconds. Highlander aims to eventually deploy 100 modules at the site, which it claims would save 68,000 square meters of land, 122 million kilowatt-hours of electricity and 105,000 tons of freshwater per year.

Perhaps the scepticism around underwater data centres is to be expected. As land-based creatures, we’re not completely comfortable with the notion of moving things that are above ground under water. It will be interesting, however, once offshore wind farms become widely adopted, whether we might be more amenable to siting data centres at sea as well.

Project Natick’s existence shows that we are prepared to think outside the box, but Microsoft’s reluctance to use the results to consider the deployment of submersible data centres more widely is a failure of imagination. What we consider the safe option is often nothing more than inertia, an unwillingness to move beyond doing something the way we’ve always done it.

It puts me in mind of the phrase people often use to try and convince reluctant would-be swimmers to join them: “Come on in, the water’s lovely.” Who knows? Maybe someday, they might.

Read More:

Back to Top ↑