On Tuesday, Oracle announced that it was designing a data center that would be powered by three small modular reactors. These SMRs will provide a reliable source of clean baseload power that you simply cannot get from traditional clean energy sources like wind and solar.
The new data center will require more than a gigawatt of electricity, and Oracle has already selected the necessary building permits for the SMRs.
At our Houston luncheon on Tuesday, September 17 we will be discussing how AI data centers will be impacting demand for copper.
AI Electricity Demand - Sept 13
AI Electricity Demand - Sept 13
Dan Steffens
Energy Prospectus Group
Energy Prospectus Group
Re: AI Electricity Demand - Sept 13
The current forecast for the SMR reactors is 2030. Will enough uranium be available at that time?
That is a long time to wait if you want electricity for data centers in the near term.
We all know gas powered power plants have the shortest timeline to generate electric power.
Joel Salatin, the most famous farmer in North America and maybe the world. has a solution for the elimination of COo2 from power plants and methane from cows.
If you feed cows 100% on grass they are a positive for removing carbon. His farm does not use vaccines or any chemicals for the past 50 years.
He has stated if all the cows were raised like him in ten years they would remove all the carbon since the industrial age.
He has been farming this way for decades.
You can look him up at Polyphase farms, there are dozens of videos on what and how he does it.
A super smart and entering farmer. The solution is not factory farms for the eco system and healthy foods like most of us are currently consuming.
I will stop here.
That is a long time to wait if you want electricity for data centers in the near term.
We all know gas powered power plants have the shortest timeline to generate electric power.
Joel Salatin, the most famous farmer in North America and maybe the world. has a solution for the elimination of COo2 from power plants and methane from cows.
If you feed cows 100% on grass they are a positive for removing carbon. His farm does not use vaccines or any chemicals for the past 50 years.
He has stated if all the cows were raised like him in ten years they would remove all the carbon since the industrial age.
He has been farming this way for decades.
You can look him up at Polyphase farms, there are dozens of videos on what and how he does it.
A super smart and entering farmer. The solution is not factory farms for the eco system and healthy foods like most of us are currently consuming.
I will stop here.
Re: AI Electricity Demand - Sept 13
In addition to there not being enough reliable electricity, AI has a bigger problem.
Thermodynamics: The End of Moore's Law?
The problem with expecting AI to simply scale up, at the same rate it has been, is simply ignorant to the laws of thermodynamics.
The computing power necessary to turn our chaotic world into a mechanism of trillions of thinking moving parts will make what we have today look like the early industrial revolution by comparison.
Processors will not only need to become more numerous, but smaller, as more and more of them will need to be packed into smaller spaces.
That’s where Moore’s law comes in.
Most are familiar with the general premise: That chip density will double every year or so, but what most people might not know is that Gordon Moore, who authored this hypothesis in 1965, didn’t envision it progressing far beyond ten years.
Here is the actual quote published in his 1965 Electronics Magazine editorial:
"The complexity for minimum component costs has increased at a rate of roughly a factor of two per year. Certainly over the short term this rate can be expected to continue, if not to increase. Over the longer term, the rate of increase is a bit more uncertain, although there is no reason to believe it will not remain nearly constant for at least 10 years." - Gordon Moore
In fact, in that same editorial, the prediction which made him famous — 65,000 components within a single semiconductor — only forecasted chip density ten years out.
We are now almost 50 years past that point, and though the increase in computing power has continued to explode more or less in line with Gordon Moore’s 59 year old hypothesis, there is a wall approaching which will require a technological revolution in its own right to overcome.
That wall is thermal buildup, and is as old and constant a problem as it is seemingly insurmountable.
Heat will result from any chemical or physical interaction. It’s a law of the universe which cannot be broken.
What can be done, however, is the introduction of new materials capable of dissipating heat on new levels.
All of these problems were as theoretical as their solutions up until a few years ago, when the AI revolution started to roll.
Today, we’re about to hit the vertical portion of the hockey stick in terms of AI development and implementation.
All of that will not be possible without a solution to the heat problem.
Thermodynamics: The End of Moore's Law?
The problem with expecting AI to simply scale up, at the same rate it has been, is simply ignorant to the laws of thermodynamics.
The computing power necessary to turn our chaotic world into a mechanism of trillions of thinking moving parts will make what we have today look like the early industrial revolution by comparison.
Processors will not only need to become more numerous, but smaller, as more and more of them will need to be packed into smaller spaces.
That’s where Moore’s law comes in.
Most are familiar with the general premise: That chip density will double every year or so, but what most people might not know is that Gordon Moore, who authored this hypothesis in 1965, didn’t envision it progressing far beyond ten years.
Here is the actual quote published in his 1965 Electronics Magazine editorial:
"The complexity for minimum component costs has increased at a rate of roughly a factor of two per year. Certainly over the short term this rate can be expected to continue, if not to increase. Over the longer term, the rate of increase is a bit more uncertain, although there is no reason to believe it will not remain nearly constant for at least 10 years." - Gordon Moore
In fact, in that same editorial, the prediction which made him famous — 65,000 components within a single semiconductor — only forecasted chip density ten years out.
We are now almost 50 years past that point, and though the increase in computing power has continued to explode more or less in line with Gordon Moore’s 59 year old hypothesis, there is a wall approaching which will require a technological revolution in its own right to overcome.
That wall is thermal buildup, and is as old and constant a problem as it is seemingly insurmountable.
Heat will result from any chemical or physical interaction. It’s a law of the universe which cannot be broken.
What can be done, however, is the introduction of new materials capable of dissipating heat on new levels.
All of these problems were as theoretical as their solutions up until a few years ago, when the AI revolution started to roll.
Today, we’re about to hit the vertical portion of the hockey stick in terms of AI development and implementation.
All of that will not be possible without a solution to the heat problem.
Dan Steffens
Energy Prospectus Group
Energy Prospectus Group