- Data centers are ridiculously efficient. The latest generation of data centers from the big cloud providers have PUE in 1.07 territory, which would've seemed like pie-in-the-sky ludicrousness even 10 years ago. Every CPU cycle that's done in a data center instead of on someone's client machine or on-prem server is a decrease in power consumption.
- 3D printers don't use very much power, and in any case, while 3D printing is going to be an ever-bigger deal in manufacturing, the 3D printing at home fad has already crested and is fading away, like 3D televisions. There's not going to be one in every home.
- Electric cars will indeed increase power consumption, but mainly when they are plugged in for the night, where there is already a lot of wasted power from e.g. nuclear plants that can't scale down at night. The total increase in necessary generating capacity and infrastructure upgrades won't be as big as you think.
> the 3D printing at home fad has already crested and is fading away, like 3D televisions
I too am skeptical of 3d printing but I don’t think you can write it off just yet.
It may be true that we’re seeing the crest of the 3d dot matrix thing. But if you include all “arbitrarily programmable assembly robots” then I think we’re just in the prelude.
The data structure for 3d printing is the “thingiverse” and I don’t think we’ve seen the end of that yet. I can imagine for example a small kitchen robot with reservoirs for some staple foods, and an app where you browse dishes that can be cooked for you. Or write your own.
I would consider that 3d printing.
Similarly 3d TV is alive and well in mobile VR. And it will see huge growth this year with the advent of $200 high quality stand-alone headsets.
You can argue these things are “not the same” but the vision is the same, only the form factor is different.
Data centers are ridiculously efficient. The latest generation of data centers from the big cloud providers have PUE in 1.07 territory, which would've seemed like pie-in-the-sky ludicrousness even 10 years ago. Every CPU cycle that's done in a data center instead of on someone's client machine or on-prem server is a decrease in power consumption.
<<
Do you know what is the global consumption of energy of data centers in 2016 ? It is same as consumption of energy of global aviation. Think about it, the energy consumed by Data Centers is equal to all commerical planes world wide flying.
(source: realclearnergy podcast)
They may very well be efficient, that is shit load of energy and that is point.
Energy consumption of data centers as of 2016 shows they are a major component of total energy consumption and therefore worth watching. It doesn't imply how that will affect _growth_ in energy usage over time. You have to look at change over time taking into account demand, supply and efficiency together in an analysis like this http://www.datacenterknowledge.com/archives/2016/06/27/heres...
> Do you know what is the global consumption of energy of data centers in 2016 ? It is same as consumption of energy of global aviation. Think about it, the energy consumed by Data Centers is equal to all commerical planes world wide flying.
Does the global aviation cost comprise just the cost of flying planes, or does it include the cost of maintaining mainframes for flight bookings?
Also, how does the consumption of power by data centers compare to the consumption of power by all consumer PCs/workstations? That's kind of the comparison we're looking at here.
> "Data centers are ridiculously efficient. The latest generation of data centers from the big cloud providers have PUE in 1.07 territory, which would've seemed like pie-in-the-sky ludicrousness even 10 years ago. Every CPU cycle that's done in a data center instead of on someone's client machine or on-prem server is a decrease in power consumption."
Those PUE statistics are self reported and aren't very accurate. I've seen them done and they are more marketing/PR fluff than actual conditions of energy conversion efficiency. Nobody in the actual field treats them as being accurate.
- 3D printers don't use very much power, and in any case, while 3D printing is going to be an ever-bigger deal in manufacturing, the 3D printing at home fad has already crested and is fading away, like 3D televisions. There's not going to be one in every home.
- Electric cars will indeed increase power consumption, but mainly when they are plugged in for the night, where there is already a lot of wasted power from e.g. nuclear plants that can't scale down at night. The total increase in necessary generating capacity and infrastructure upgrades won't be as big as you think.