Undersea bit-barn biz offers 90-day trial of submerged server system Subsea Cloud is offering potential customers the chance to try out its underwater datacenter facilities for up to 90 days before making any further commitments, in a bid to attract new customers to the project. The bit barn biz, which specializes in subsea datacenter deployments, said that its scheme is a change in approach …
It should be cheaper to just build the data center on a container port somewhere and throw cooling hoses into the water. But:
Real estate costs, planning permission, environment impact stuff, finding out that the land, shore, beach and tidal waters all have a different government agency with their own rules/requirements.
Although this does seem like a way of burning VC dollars except in a few locations where you are putting offshore wind infrastructure in anyway
Do you realize how big the oceans are?
Heat is being released on the earth regardless, and heat is transferred between the oceans and the atmosphere all the time via storms. It won't heat up the oceans any more than a land based datacenter would. Less, in fact, since they aren't generating additional heat from the power required to disperse the heat in the atmosphere (via evaporation, which requires huge chillers and lots of outside water pumped in)
"...opportunities with offshore wind farms. This would give it ready access to renewable energy, overcoming challenges related to power transmission..." good luck running a server farm on variable power as the wind blows, or are we going to duplicate energy infrastructure with thermal power generation on shore?
Nope you duplicate the data to different places with different weather. When the 'free' wind power stops in the middle of the day or in summer you switch to the solar data center in the desert and then switch back at night and in winter.
Data centers doing heavy compute are the perfect load balancera
Interesting take. Reminds me of the company that rented servers the size of radiators to heat homes. It was a thing a couple of years ago, but lately not so much in the news anymore. Maybe the fans are too loud, I don't know.
Anyhow, geographically splitting the data centres statistically would help to avoid some of the troughs in power generation, but it is no guarantee. Winter nights can be dead calm and then your servers would still need thermally generated power. And the duplication of equipment would definitely increase the cost.
The whole idea of a submerged data centre seems interesting though as a permanent sea bed structure with regular access. The tides could provide the water flow to dissipate the heat. And northern seas can be quite cold. Would be interesting to hear whether there is a business case for this.
Uncomfortable, Nauga hide couch to try and sleep on, at two in the morning, while freezing to death in the stale aircon atmosphere and staring at the flickering overhead light. The taste of stale, shitty, coffee still lingering in your mouth while you wait for the job to crash.
This post has been deleted by its author
But the reduction in transmission losses is impossible to balance with fighting a marine environment.
The whole ocean is an evil bitch. The north sea is a cut above most in that regard. Eat the smaller electrical and moderate pumping losses and avoid salt corrosion, marine scale and biofilm, and the occasional naval anchor dragging trough the neighborhood.
Boiling crabs in the harbor may have a localized impact, but the drowning data center is going to have a hard fight to beat a standard containerized unit at the surface. Hang a cluster of them off the wind turbine's support tower if you need them to be close, but leave them above water at least. Then it's mostly the seagulls you will need to contend with, and the turbine will sort a few of them out...
Aside from all the maintenance issues mentioned. Has anyone thought about the impact, if the practice took off, of directly warming the sea? A lot of the odd weather this year may be down to the sudden surface sea temp rise about a year ago. Either that or their damn geo-engineering games. Although you'd need a huge amount of power to warm the overall ocean noticeably datacenteres would be sited conveniently not harmelssly so what happens if for example yopu end up with a few big ones in a shallow coastal current?
In 2022, Google's data centres used 22.29 TWh of energy, which is 8.0244 × 10^16 J, including the energy to run the cooling. The global sea water volume is 1,370,000,000 cubic kilometers. Heating a kg of water by 1 K requires 4184 J. Google's energy use would increase the average world sea water temperature approximately by 0.000000014 K. An immeasurably small difference.
When I first read of datacentres under the sea I was imagining staffed underwater facilities where staff would be rotated by submersibles or tunnels or tubes from adjacent land. Voyage to the Bottom of the Sea stuff. :)
Rockall Is would be favourite - tunnel down through the midde of the island to well below sea level and hang the server rooms off the sides of the island - could host all the world's social media there. (Sorry Tim for stealing your plot. :)
More seriously - why is it cheaper or better to pump electricity to a server container underwater than to pump cold sea water through a terrestrial datacentre? (Or coolant cooled by cold seawater?)
I can see real merit in the idea of having a standardized shipping container sized datacentre module that could he forklifted in and out for tenants to populate or maintain. Standard redundant 3-phase AC, cooling or coolant, environmental monitoring, networking etc - plug and pay. :)
I was thinking something like the multistorey yacht/boat storage facilities found in large marinas which are a little like gigantic robotic tape libraries.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
