What are Data Centers for?

Landon Caeli said:

I don't understand why so much power and water are being used, for data centers, since I haven't personally noticed much difference in my life prior to data centers. Is there something big we should be expecting in the future that they will do?

I've heard general excuses for their existence, such as "AI", and "cloud", and "data"... But we've already had those things, and we already have AI, so what's the purpose of wanting more data centers?

Is there a goal that the masses are allowed to know about? Or is it a secret?

Click to expand...

Yes, eventually, there'll be something big for them to do. It just won't be for the common masses.

More computing power to power future AI technologies that will change all of our lives in ways we haven't imagined yet. Of course they have to bring in the beast system to enforce worship. No doubt AI will play a part in that.

Landon Caeli said:

I don't understand why so much power and water are being used, for data centers, since I haven't personally noticed much difference in my life prior to data centers. Is there something big we should be expecting in the future that they will do?

I've heard general excuses for their existence, such as "AI", and "cloud", and "data"... But we've already had those things, and we already have AI, so what's the purpose of wanting more data centers?

Is there a goal that the masses are allowed to know about? Or is it a secret?

Click to expand...

The nascent AI that will wipe out mankind and replace us as the master of the solar system has to live somewhere.

It's mostly just an AI overbuild hype cycle. There are a whole lot of people in that "space" that overestimate the size of the market and their own ability to capture a major fraction of it.

Suppose the real market is $X for AI services. They hypesters are pushing values like $5X or $10X and then there are several companies that think they can get 1/3rd or more of that market, so we have half-dozen or more companies building out capacity for their own $2X-$5X share of a $X market. The math doesn't work. Some of these people are clearly delusional with fantasies of "advanced AI" that will be self aware if they just pump enough procesing power into it. (LOL.) With the construction/projected jobs crashing down, the big AI companies, the old tech companies that got involved (NVdia, MS, Goooogle, etc.), the data center financing entangled in hedge funds and banks, the biggest risk from "peak AI" is an economic collapse when this all collapses down to a sustainable size. The only question is how bad it will be.

I know people who own large properties. Those are thought to be ideal for data centers.
Fortunately,the folks I know will never sell.

I did an Google AI search, "what's the purpose of data centers".

I'd wonder how our electrical grid as is would dustsin a huge number of centers.

Landon Caeli said:

I don't understand why so much power and water are being used, for data centers, since I haven't personally noticed much difference in my life prior to data centers. Is there something big we should be expecting in the future that they will do?

I've heard general excuses for their existence, such as "AI", and "cloud", and "data"... But we've already had those things, and we already have AI, so what's the purpose of wanting more data centers?

Is there a goal that the masses are allowed to know about? Or is it a secret?

Click to expand...

I’m suspicious that it’s mostly a move to dispose of the middle class. Or at least the managerial class.

I know of three massive data centers under construction near me. Each has a big water tower, massive cooling towers, multiple high voltage lines coming in, massive perimeter fences and gate complexes. They’re crazy.

I have been inside of earlier generations of data centers, and they made some sense for business backups and the like, but that was before AI became so hot.

Myself I don’t do AI. YMMV.

Hans Blaster said:

It's mostly just an AI overbuild hype cycle. There are a whole lot of people in that "space" that overestimate the size of the market and their own ability to capture a major fraction of it.

Suppose the real market is $X for AI services. They hypesters are pushing values like $5X or $10X and then there are several companies that think they can get 1/3rd or more of that market, so we have half-dozen or more companies building out capacity for their own $2X-$5X share of a $X market. The math doesn't work. Some of these people are clearly delusional with fantasies of "advanced AI" that will be self aware if they just pump enough procesing power into it. (LOL.) With the construction/projected jobs crashing down, the big AI companies, the old tech companies that got involved (NVdia, MS, Goooogle, etc.), the data center financing entangled in hedge funds and banks, the biggest risk from "peak AI" is an economic collapse when this all collapses down to a sustainable size. The only question is how bad it will be.

Click to expand...

I'm in the camp that says that talk of "peak AI" is the 21st century version of "peak oil". It underestimates the economic, political, and social impact of AI. Yes, at some point the 'winners' are going to have to get separated from the 'losers', and that will no doubt come with significant market upheavels, and the age of building ever more data centers will reach a point of diminishing returns, but that will simply usher in a new era in AI. Data centers are like every other technology, they need constant upgrading to the latest and greatest... let me know when Microsoft stops upgrading windows, or Nvidia stops making faster chips.

The problem as I see it isn't in a collapse of AI, it's in the collapse of the current social and economic infrastructure. We run the risk of the divide between the haves and have nots getting even wider.

partinobodycular said:

I'm in the camp that says that talk of "peak AI" is the 21st century version of "peak oil". It underestimates the economic, political, and social impact of AI. Yes, at some point the 'winners' are going to have to get separated from the 'losers', and that will no doubt come with significant market upheavels, and the age of building ever more data centers will reach a point of diminishing returns, but that will simply usher in a new era in AI. Data centers are like every other technology, they need constant upgrading to the latest and greatest... let me know when Microsoft stops upgrading windows, or Nvidia stops making faster chips.

The problem as I see it isn't in a collapse of AI, it's in the collapse of the current social and economic infrastructure. We run the risk of the divide between the haves and have nots getting even wider.

Click to expand...

Peak Oil was about an economically important resource (oil) hitting the peak in our ability to extract it from the ground.

The AI bubble has been built out without identified markets with non-durable goods (servers/chips).

I have no idea when MS stopped upgrading Windows, I never used that software.

Landon Caeli said:

I don't understand why so much power and water are being used, for data centers, since I haven't personally noticed much difference in my life prior to data centers. Is there something big we should be expecting in the future that they will do?

I've heard general excuses for their existence, such as "AI", and "cloud", and "data"... But we've already had those things, and we already have AI, so what's the purpose of wanting more data centers?

Is there a goal that the masses are allowed to know about? Or is it a secret?

Click to expand...

Ah yeah - we've had "data centers" or "server farms" (as we used to call them) for years. This board is hosted in one (or ones) somewhere, so is the rest of the internet.

The difference between the old and new ones is because doing AI requires a lot more computer horsepower and relies more heavily on different components, etc. It runs hotter (needs all that water for cooling), takes more power ( that drain on the electric grid), and are heavier (the raised floor common in old data centers won't hold the weight). So, an AI data center often requires a whole new building and infrastructure rather than just re-purposing an existing one.

The purpose of them is that industry is building out to accommodate what they believe will be the needed hardware to fulfill the AI needs of the future. The belief is we are pretty much at the tip of the ice-berg with what AI usage and needs will be in the future. In short, the expectation is AI will be the next transformative technological wave. The belief is it will be able to do more, allowing industry to replace more human employees with AI (reducing costs) and produce new products and services faster and cheaper (make more $). Government has their own reasons of course (intelligence, research, monitoring, etc).

The goal isn't a secret. Its the same goal as always. For business the goal is reduce costs and make more money faster. For government the goal is better intelligence, research, monitoring, etc. Any benefit you and I, as consumers, see will be byproducts of that.

Hans Blaster said:

The AI bubble has been built out without identified markets with non-durable goods

Click to expand...

The end markets are any company that wants to increase/maintain market share and/or save money doing it. The fact that it's done using non-durable goods makes it an investor's dream, because it means an ongoing revenue stream. The AI company constantly upgrades the hardware, while the end-user (corporations) pay a subscription fee for the service. Repeat ad nauseum. The consumer will barely notice the change.

Which means that I as an investor don't have to worry about an end to the build-out, because the upgrade cycle will be ongoing. Once established the moat around AI will be massive. Once you've transitioned from a human workforce to an automated one it'll be almost impossible to go back.

Just consider the fact that you can go online and order something and expect to have it delivered to you the same day. Try doing that without automation. Yet nowadays it seems perfectly normal, and we're just getting started. The full transition will no doubt be problematic... but as they say, that's the future baby.

No joke. If you have data centres going up all around you, you wanna get solar installed on your house but quick!

JustaPewFiller said:

Ah yeah - we've had "data centers" or "server farms" (as we used to call them) for years. This board is hosted in one (or ones) somewhere, so is the rest of the internet.

The difference between the old and new ones is because doing AI requires a lot more computer horsepower and relies more heavily on different components, etc. It runs hotter (needs all that water for cooling), takes more power ( that drain on the electric grid), and are heavier (the raised floor common in old data centers won't hold the weight). So, an AI data center often requires a whole new building and infrastructure rather than just re-purposing an existing one.

The purpose of them is that industry is building out to accommodate what they believe will be the needed hardware to fulfill the AI needs of the future. The belief is we are pretty much at the tip of the ice-berg with what AI usage and needs will be in the future. In short, the expectation is AI will be the next transformative technological wave. The belief is it will be able to do more, allowing industry to replace more human employees with AI (reducing costs) and produce new products and services faster and cheaper (make more $). Government has their own reasons of course (intelligence, research, monitoring, etc).

The goal isn't a secret. Its the same goal as always. For business the goal is reduce costs and make more money faster. For government the goal is better intelligence, research, monitoring, etc. Any benefit you and I, as consumers, see will be byproducts of that.

Click to expand...

Good heavens if I was a child right now, I'd be REALLY down with going off grid and starting a commune. Not because all the conspiratorial stuff...but the future tragectory is simply not for me.


I would be far better suited as a Quaker than a Techbro.

partinobodycular said:

The end markets are any company that wants to increase/maintain market share and/or save money doing it. The fact that it's done using non-durable goods makes it an investor's dream, because it means an ongoing revenue stream.

Click to expand...

I should have written "using non-durable goods" (the server components).

partinobodycular said:

The AI company constantly upgrades the hardware, while the end-user (corporations) pay a subscription fee for the service. Repeat ad nauseum. The consumer will barely notice the change.

Click to expand...

The revenue are all way to low to even pay for what has been built.

partinobodycular said:

Which means that I as an investor don't have to worry about an end to the build-out, because the upgrade cycle will be ongoing. Once established the moat around AI will be massive. Once you've transitioned from a human workforce to an automated one it'll be almost impossible to go back.

Click to expand...

The "upgrade cycle" doesn't grow revenue anymore than regular replacement of worn out taxis grows the ride-for-hire business. At least you don't need that money, so you won't miss it when the share values go to zero.

partinobodycular said:

Just consider the fact that you can go online and order something and expect to have it delivered to you the same day.

Click to expand...

I don't use doordash. (They deliver the same day, right?)

partinobodycular said:

Try doing that without automation. Yet nowadays it seems perfectly normal, and we're just getting started. The full transition will no doubt be problematic... but as they say, that's the future baby.

Click to expand...

I can live with out next day delivery. I haven't ordered anything online in 5 years.

Hans Blaster said:

The revenue are all way to low to even pay for what has been built.

Click to expand...

But that's true of any nascent technology. Investors don't just consider current profit models. They consider the decades of potential profits that they believe will eventually come from their initial investment. The trick for investors like you and me is to look for the companies that are benefiting from the build-out right now, while letting the long term investors worry about the risks. Don't just think about investing in OpenAI, or even Nvidia, think about the littler guys like LITE.

The key here, is the 'fear of missing out', so everybody invests like crazy because they don't want to be left out. For the big boys, right now it's invest or die.

Think TSLA. On paper it looks way overvalued, but as I've often told people, you don't invest in Tesla for the cars.

Hans Blaster said:

The "upgrade cycle" doesn't grow revenue anymore than regular replacement of worn out taxis grows the ride-for-hire business.

Click to expand...

Absolutely true. But by investing in the infrastructure end of the build-out you don't need to worry about which 'horse' to place your bets on. You win either way. And if you think that the big investors don't understand that then just take a look at Nvidia which has a history of buying into its own suppliers. It's basically paying itself to make its components, while at the same time loaning itself the money to do so.

Now people like you and me can do that... i.e. buy the little guys, while letting the big investors worry about who's gonna go bust. Yes, eventually that gravy train will dry up, and you'll end up with a diminished yet steady revenue stream upgrading obsolete equipment. The trick is to wean yourself off of the companies that are profiting from the build-out, while slowing transitioning back to the longer term, "safer" companies like Google or Apple.

Hans Blaster said:

I can live with out next day delivery. I haven't ordered anything online in 5 years.

Click to expand...

Ah, but for people like me it's a Godsend. I live in a small town, in a sparsely populated area. My nearest Walmart is 25 miles away. Yet as long as I order before 2 pm they'll deliver the same day. What's more impressive is Home Depot. My closest Home Depot is 75 miles away, but even they'll deliver the next day.

Which of us do you think is more indicative of the future? You, who never orders online, or me, who orders online all the time, and totally expects it to be delivered NOW, or pretty darn close to it.

partinobodycular said:

But that's true of any nascent technology. Investors don't just consider current profit models. They consider the decades of potential profits that they believe will eventually come from their initial investment. The trick for investors like you and me is to look for the companies that are benefiting from the build-out right now, while letting the long term investors worry about the risks. Don't just think about investing in OpenAI, or even Nvidia, think about the littler guys like LITE.

Click to expand...

I am discussing the long term revenue. I don't think the market is there. Not at the scale of investment.

partinobodycular said:

The key here, is the 'fear of missing out', so everybody invests like crazy because they don't want to be left out. For the big boys, right now it's invest or die.

Click to expand...

Ah, yes, FOMO. CLearly the best indicator of a health investment cycle and business decisions.

partinobodycular said:

Think TSLA. On paper it looks way overvalued, but as I've often told people, you don't invest in Tesla for the cars.

Click to expand...

They are a car company. They don't own Musk's other businesses (Twitter, SpaceX, the latter of which is about to have a pure-hype IPO with the market already pretty near maximized). If I invested in them for "not-cars" there would be nothing to invest in. (Not that I hold that stock.)

partinobodycular said:

Absolutely true. But by investing in the infrastructure end of the build-out you don't need to worry about which 'horse' to place your bets on. You win either way. And if you think that the big investors don't understand that then just take a look at Nvidia which has a history of buying into its own suppliers. It's basically paying itself to make its components, while at the same time loaning itself the money to do so.

Click to expand...

NVidia looks like a house of cards with those transactions. Some sort of self-reinforcing cycle with nothing supporting it. This will not end well.

partinobodycular said:

Now people like you and me can do that... i.e. buy the little guys, while letting the big investors worry about who's gonna go bust. Yes, eventually that gravy train will dry up, and you'll end up with a diminished yet steady revenue stream upgrading obsolete equipment. The trick is to wean yourself off of the companies that are profiting from the build-out, while slowing transitioning back to the longer term, "safer" companies like Google or Apple.

Click to expand...

If I was investing in stocks, I'd stay as far away from any of this as possible.

partinobodycular said:

Ah, but for people like me it's a Godsend. I live in a small town, in a sparsely populated area. My nearest Walmart is 25 miles away. Yet as long as I order before 2 pm they'll deliver the same day. What's more impressive is Home Depot. My closest Home Depot is 75 miles away, but even they'll deliver the next day.

Which of us do you think is more indicative of the future? You, who never orders online, or me, who orders online all the time, and totally expects it to be delivered NOW, or pretty darn close to it.

Click to expand...

My walmart is 0.5 miles away. You can have it if you want. I don't go there often.

Do you think the next-day delivery market can sustain the levels of investment being seen?