" and releases no discharge into the surrounding waterway"
And what about heat?
Like extra heat into already warming up oceans?
Plans for floating datacenters in the US have advanced beyond the handshake stage, with Samsung Heavy Industries and Mousterian Corporation signing an engineering contract. Dallas-based Mousterian, which describes itself quaintly as "a developer of water-adjacent datacenter infrastructure," says the contract covers its first …
The amount of heat directly released is trivial compared to the mass of the ocean.
The emissions from all the gas, diesel and/or coal burned to power the ship, would lead to far more warming up of the ocean, due to extra heat trapped in the atmosphere.
"And what about heat?
Like extra heat into already warming up oceans?"
You can work it out. Some assumptions:
10 of these 50MW data centres, all off the coast of Texas in the gulf of Mexico. Let's assume all the electricity consumed goes into heating the gulf's water.
10x50MW is equal to 1.6x10¹⁶ J per year.
Volume of the gulf is 2.4 million cubic kms, or about 2.4x10¹⁸ kgs of water.
Specific heat capacity of water is about 4200 J/kg/°C
So the temperature increase of the gulf over a year is 1.6x10¹⁶ / (2.4x10¹⁸ x 4200).
It's about 0.0000016°C per year.
To put it in some context, if we assume that the gulf receives the same solar radiation per unit area as the rest of the world (averaged out for day/night cycles) then the 500MW of electrical power is only about one millionth that of the solar radiation.
Of course, that energy will be localised, not spread out across the whole gulf. But if you want to think about the effect on weather, do the same calculation over say, just the area of a weather region. But that's getting into meteorology and I'm way out of my depth now.
"salt water environment"
Cruise ships (for example) generally have a large number of computers on board and in my experience you generally don't notice the salt at all. There are even open air swimming pools 15 odd decks up on these beasts. A modern cruise ship can be of the order of 200,000 tonnes or even more. I've been on a P&O beast that was 180,000t and the usual Atlantic swell of 5 metres was mostly unnoticeable. Even whizzing across the Bay of Biscay with a swell that would dwarf my house only sent a few people off for some sea sick pills.
However, the seas are able to destroy the biggest ships if fortune decides so. There are "rogue waves" and other phenomena that laugh at our paltry efforts to master nature.
It is not beyond the whit of man (think drilling rigs too) to deal with salt water corrosion.
Have you considered that the use of ocean water for cooling is very old tech. Just look at the number of nuclear power plants located on the side of the ocean. How do you think that nuclear submarines keep the reactors cool.
Even in Australia there is a coal fired power station (Newport Power Station) that uses estuarine water for cooling. When the tide is in that uses ocean water.
"If there's one thing I picture thinking about Texas, it's scarce development land"
Drive or take a train across Texas and there's land as far as one can see. Days of it at highway speeds and that's not even a sliver of what's there. One could lose a 300 hectare solar farm in the state and have to zoom way into a satellite photo to find it again.
The comment about "critical IT infrastructure" doesn't work. xAI is renting out their computers to competitors due to no need for the capacity. Other computing centers are also under-utilized so the critical part is road apples.
There are also more plans to build facilities than there are under construction already and most of those likely won't get built.
I could imagine an oil drilling rig repurposed as a data center vs. putting them in space, but not a ship. The cost is vastly more expensive than having one on land and there is tons of land in the US away from nice people a data center could be put.
What could possibly go wrong?
Govenor Hotwheels makes very few decisions not politically motivated or designed to disenfranchise voters, but it appears he may have run into reality and had to make an actual decision to prevent brining the Texas electrical grid crashing down. For those not familiar, Texas deregulated it's electricity market in 1995 using a bit of "lectric legerdemain" promising Texans competition in the electricity market to lower the price consumers pay. (the cost per KWh in Texas has tripled since, surprise!)
Slipped into the misguided deregulation, Texas removed the maintenance requirement requiring the Transmission & Distribution Service Provider (TDSP) replace equipment based on the "engineering life" of the component (e.g. a pole has an engineering life of 25 years, the TDSP has to replace poles within a 25 year cycle). Deregulation left maintenance to the discretion of the TDSP, which, in reality, allowed them to avoid regular maintenance altogether, preferring a just replace it when it breaks philosophy, helps keep profits high.
30 years after deregulation, you can imagine the condition of the Texas grid. Now somebody wants to make another 474 GW of new connections? This won't end well, for reasons that are all too readily apparent. I suspect that is probably what forced the good governor's hand, as his Texas is open-for business philosophy -- no matter how bad it hurts Texans far outweighs any slight political benefit from the anti-datacenter stance.
Adding anything to the Texas grid and it starts to sound like a bowl of Rice Krispies "Snap, Crackle ... Pop".
Called Rice Bubbles here. But the TX grid sounds like it is going to go Bang before the the AI bubble goes Pop.
I imagine a major weather event or other natural disaster could do it. When a substantial proportion of the unmaintained infrastructure is damaged just sourcing replacement components can be a major headache as there would not be a significant inventory usually on hand for routine maintenance.
"e.g. a pole has an engineering life of 25 years,"
I've owned my home for 11 years and it was built a bunch of years before that. I've only seen one pole on my line replaced in that time and that was because somebody hit it. I have a feeling that most poles last longer than 25 years. My take would be to only replace them when necessary unless they start seeing data that shows they might be better off doing the replacements preemptively.
There's been some good papers done on the ROI for going solar vs staying grid connected. Texas is middling, but some areas could make more sense. A big set of data centers going in would be a big concern as the power company might push the transmission closer to the max. It's not so much neglect, but insufficient investment to have proper margins. The data centers can go in much faster than the power companies can order and receive transformers and other gear so there's a temptation to squeeze the margins until the rest of the equipment can be installed.
F'em. Do what you need to make sure your needs are met and then you don't have to stress about those data centers breaking the grid mid-summer and turning off your HVAC.
“ Texas governor Greg Abbott this week paused approvals for datacenter projects seeking access to the Electric Reliability Council of Texas (ERCOT) grid until state regulators audit their connection requests.”
Is he auditing the size of the envelopes they’re offering?