Amazon has had fulfillment robots in their warehouses for over ten years. And that has only increased. This is not something that has been a secret or is anything new. You can start by going back to look up "Kiva systems" for the original style of fulfilment robot. And I'm sure that newer styles of robots that do more specialized tasks have been developed.

This is Yahoo talking out of their ass in the hopes that they sound smart.

It is a micro generator. This has the usual issue of "Science reporting" trying to dumb it down so much that it gets it wrong.

Science Fiction has always been woke. And that is because it has always been looking forward into the possibilities of what Mankind was possible of.

Hint for you: Star Trek was woke from the beginning.

So which disgruntled staffer dumped their email to the press? Its makes a lot more sense than what ever crap the Trump campaign is trying to spin. And if their had been a breach, the FBI would have been on it.

Apple already knows the EU is going this way.

Its John Deere that is going to try to fight this tooth and claw.

The issue is scale. One GPU with a remote cooler is one thing. Thousands of them requires an entirely different approach to cooling.

sure not a problem, how are you going to extract the energy out of the warm water to drive this deslinization plant? if someone had that technology they would have made a fortune off it already.

Well I don't think you understand that scale you are talking about.
For example: The city of Toronto uses a great lake for cooling for its NBA arena.
But if you wanted to use city water that you pull in and then reject back to the city to be used elsewhere. Lets assume you have a "small" data center with a 100 tons of cooling load (12,000 BTUs/hour *100) with a total of 12,000,000 BTU/hr. Water increases temperature at 1 degree F per pound of water, so (very roughly) 8 btus per gallon per degree. So lets assume we heat city water by ten degrees (that keeps us well below legionella). (insert a bit of math here I am not going to try to show in HTML): That means this "small" data center needs to pull 2,400 GPM, all the time. And the new water has to be reject its heat someplace or it is going to be pulled back into the data center having been warmed up. So at 4am when everyone is asleep and this data center is still operating, it is still going to be pulling 2,400 GPM and the water in those lines is going to get warm because no one is using the warm water.
And remember that all of this water you are piping around has to be piped as if it is potable water - because you are claiming it is still potable. That is likely to stomp on this right there. I suspect most city water managers are not going to allow anyone to inject untested water back into the drinking water supply for a city.

Okay here we go: 1. chips and electronics normally have an upper operating temperature abound 100F.
2. Water boils at 212F (before adjusting for elevation).
3. In order to boil the water the water has to be heated to 212F, so the chips either have to be able to operate at 212F (which I don't think is a design constraint right now) or the water temperature has to be raised with a mechanical process. That means high temperature chillers that can provide the needed cooling, while raising the water temperature to a level where it will start to evaporate. And if you are at a temperature that is that high on the discharge side, just go to Direct expansion for the heat rejection part of the show.
4. and salt water? If you want to increase the operating cost of your cooling plant, please feel free. Everything would have to see a seawater resistance grade of stainless steel. So you have just spend an incredible amount of money. That includes your seawater grade chiller as well. And remember that the stainless steel is not immune to seawater, it is very resistant to seawater - So it will last 5-10 years before rotting out.

The problem is the energy is diffuse. In order to boil the water in order to desalinate, you need a way to concentrate that energy. That is the "non trivia" part of the problem.

The problem is, if you want it to run in a desert, it is going to have to reliably operate in 140F conditions. If the outside air temperature is (OAT) 120F (not unheard of in a desert), that is the lowest temperature the chips are going to see, so they have to be able to operate in conditions hotter than that.

And that means having huge fans that can draw enough hot air over those chips, and having racks optimized for air flow.

What you are talking about is referred to as a "dry cooler". In that case the water has to be hotter than the outdoors. So the water is going to have to be 120F, possibly hotter than that. And that means you have to have chips that can withstand temperatures of 120F (or hotter).

As the other commenter noted: Evaporate cooling in cooling towers is reliable and dependable. And it is doubly dependable in extremely dry climates. The only trade off is you need water equal to your heat generation. And that water is on a one way trip. And this is easy to set up, you inject city water into the loop, cycle it through the cooling tower, let it evaporate and refill it with more city water.

The only way I could see "Dry coolers" working is to get water-to-water chillers operating that can discharge the heat rejection side to over 120F. But at that point, you might as well just go directly to direct expansion cooling. And the reason data centers don't use that is it is much more power intensive. Cooling towers are cheap to operate as long as you have an infinite supply of water.

And part of the issue is the objections can be done sequentially.

So you file your first complaint, let the developer work through that issue over the course of a couple months. Then file your second complaint, and that takes a couple of months.

Wash, Rinse, Repeat.

Suddenly very few developers have the stomach to develop in that area. And people wonder why there isn't any development in that area of the country.

Every engineering and science discipline has known that they are both the solution and the problem.

Mechanical engineering has built bows, cross bows, trebuchet.
Civil and structural engineering has been building and destroying fortifications since before AD,
Chemical engineering has built gunpowder and plastic explosives.
Biological engineering has biological weapons
Electrical engineering has guidance systems for missiles
Nuclear engineers have THE BOMB
CompSci has had rose colored glasses on and continues to refuse to acknowledge all the ways their power can be used for evil and warfare. For any of them who say "Who us? Who would have thought of it?" I suggest they go back and read Huxley, Bradbury and Orwell who imagined variations on the problems now a hundred years ago.

I would prefer to see the following:

The content creator can hold the copyright indefinitely. The moment the copyright is transferred to anyone else - person or corporation, the clock starts with a hard ten year run out. This gives the content creator an income as long as they are alive, and then there is a ten year window after they die while someone else holds the copyright.

And yes, that means for big companies like Disney - they have to make the choice - does the copyright stay with the person who created it, or do they take then and start the ten year countdown.