This is true. The generating capacity doesn't exist to support these. I've been consulting with a group that represents some venture capitalists on solutions and opportunities. We've been looking at things like the actual data center design incorporating "hot aisle/cold aisle" data racks (not really that new) to on-site, off grid power generation utilizing different fuel sources such as hydrogen, nat gas turbine generators and small-scale nukes. Hydrogen is the most plentiful fuel source on the planet, the problem is how to economically separate it from water or nat. gas. Currently electrolysis for water and a process called steam reforming are used to generate H2 on an industrial scale. The latter relies on nat. gas and/or Petrochemical by products as feedstock so these H2 plants are normally sited next to a refinery. This is why the tech industry is booming here in Texas. It's not becuase they love our barbeque or want to follow Musk. It's our oil/gas/refinery wealth. That and lots of relatively cheap land. Bottom line, if you can't bring the power to the data center, bring the data center to the power source.
As for nuclear, interesting story. About eight years ago I was introduced by a friend/business associate who worked for the late Paul Allen's Vulcan Real Estate company, to a couple of brilliant guys who were working on a new small scale, breeder reactor concept. Breeders actually create more fissionable material as they operate (think sour dough starter). My friend knew about my nuclear background from the years I spent at the Hanford Reservation in Eastern WA. When I was at Hanford, they (the Fed old Atomic Energy Commission) had built a large scale breeder called FFTF (Fast Flux Test Facility) and I at the time worked for Exxon Nuclear on building fuel bundles for another project called LOFT for the Navy.
Anyhow, one of the big challenges with FFTF and breeders like it is they are cooled via liquid NaK (sodium potassium) which is extremely volatile when exposed to the atmosphere as the mere humidity will cause it to ignite and cannot be extinguished. So removing the fuel cores must be done in a constant atmosphere of inert gas, in this case Argon.
At any rate, the two gents I met with had developed some new tech concept that was much safer. Unfortunately, as far as I know this enterprise died when Mr. Allen passed away.
It's all fascinating stuff and there is a lot of research going on at the power generation level because yes, AI is a fricking freight train coming down the tracks with no brakes.
For any investors out there, my recommendation is to look at the power generation piece and not the data./IT end of it as that is where some entity is going to break through.
Sorry for the long post and here endeth todays lesson
www.eesi.org
www.houstonchronicle.com
