This is so ridiculous I had to share it. I used to be a happy Allbirds customer and I did enjoy their shoes in the past, but then they started struggling, stopped offering sales and IIRC started...
This is so ridiculous I had to share it.
I used to be a happy Allbirds customer and I did enjoy their shoes in the past, but then they started struggling, stopped offering sales and IIRC started cutting corners on quality.
Now they're an AI company???
EDIT: Also, I hope the irony is not lost on them going from being an environment-first manufacturer to joining one of the most environmentally harmful industries in the world. But hey, investors are happy, so does it even matter?
For what it's worth, training GPT-4 -- not using but training, the most carbon-intensive part -- produced about half as much emissions as one day of Nike's shoe production. I don't want to be...
EDIT: Also, I hope the irony is not lost on them going from being an environment-first manufacturer to joining one of the most environmentally harmful industries in the world.
I don't want to be taken as a brain-dead AI booster. I'm not telling you not to criticize AI. My intent here is to provide more facts/context to what is a super common claim.
Isn't training only the most energy-intensive part right at release, before a model actually gets used? I would think that over time the inference cost catches up since the inference cost keeps...
Isn't training only the most energy-intensive part right at release, before a model actually gets used? I would think that over time the inference cost catches up since the inference cost keeps adding up over time.
That's kind of like saying the vehicle manufacturing is the most energy-intensive part of making a car: surely true for a lot of vehicles, but if something gets heavily used, I'm sure the operation costs catch up.
Yes! All true. I cited training for two reasons. One because of the convenient shoe comparison, which was relevant. Two because the resource cost of actually using the models is so marginally low...
Yes! All true.
I cited training for two reasons. One because of the convenient shoe comparison, which was relevant. Two because the resource cost of actually using the models is so marginally low that it's almost not worth discussing (fractions of a percentage point of one's daily carbon/water consumption), so people tend to focus on training costs, which are (mistakenly) believed to be extremely high.
Training is a small part of it; you still have the AI industry's chokehold on electronics manufacturing (and the environmental impact of that), the damaging levels of noise pollution the data...
Training is a small part of it; you still have the AI industry's chokehold on electronics manufacturing (and the environmental impact of that), the damaging levels of noise pollution the data centers make, and the toll on power plants.
Water is probably the least effective argument. I'm not sure why it comes up as much as it does. For one, it's not like they're performing water electroysis, the water is still there. Most...
Water is probably the least effective argument. I'm not sure why it comes up as much as it does. For one, it's not like they're performing water electroysis, the water is still there. Most datacenters are closed loop - not for environmental reasons, just because it's cheaper. Even in the ones that aren't, y'know, water cycle and all.
All the data centers popping up in drought zones actually does make it super annoying. Phoenix area having multiple built up, likely because of cheaper land and resources. The fact it also is dry...
All the data centers popping up in drought zones actually does make it super annoying. Phoenix area having multiple built up, likely because of cheaper land and resources.
The fact it also is dry likely plays into location being chosen, but of course this will continue taxing locals even worse with multiple price increases on a limited resource.
Fact is they're not able to even be compliant with the state constitution that requires the government to have 100 years of water available because of climate change and these combining into an unachievable situation.
Edit: There's also the issue of overbuilding housing due to the government over-zoning when not having secured water yet, so there's multiple compounding issues at play.
Is there any indication that it’s having a real impact on water demand in any region of the US? A quick google on phoenix datacenter water usage gets the following articles...
Is there any indication that it’s having a real impact on water demand in any region of the US? A quick google on phoenix datacenter water usage gets the following articles
Cooling systems like these can consume a huge amount of water, but no one knows how much they are consuming. Independent estimates suggest that an average data center can use anywhere from 50,000 to 5 million gallons of water per day. An analysis from the sustainability advocacy organization Ceres estimated that the data centers active in Phoenix last summer used around 385 million gallons of water per year. Ceres projected that the metropolitan’s data center water consumption could grow tenfold to around 3.8 billion gallons per year.
But even that worst-case-scenario would make data center usage equivalent to just around 1 percent of total residential water consumption in the Phoenix area — and less than half a percent of the region’s total 2024 water usage
Not nothing but not exactly catastrophic demand increases
TLDR is that the data centers are designed to use zero water but that there was an indirect water cost that the article is bringing attention to in the form of increased water usage in power plants. Neat, but zero figures.
There is a misconception about data centers using so much water. Some use a ton of water, some don't. But it's hard for us to draw conclusions because of the (companies') lack of transparency,” said Amber Walsh, an analyst with Bluefield Research, a water market research company. Her research suggested that in 2025 data centers across the state took some 10,200 acre-feet from local water systems.
A different analysis by Ceres, a nonprofit informing investors of sustainability issues, estimated Phoenix area data centers' annual water needs alone require 10,078 acre-feet, including indirect water needs for power. That number could balloon in six years to nearly 56,000 acre-feet a year if all planned data centers go online, and add up to a 32% increase in annual water stress. An acre-foot of water is enough to supply three to four houses a year in the Southwest.
What makes understanding AI so difficult is that 1) we don't realize how many resources we're already consuming and 2) these numbers don't mesh well with the human mind. Like if you just read...
What makes understanding AI so difficult is that 1) we don't realize how many resources we're already consuming and 2) these numbers don't mesh well with the human mind. Like if you just read "data centers in Phoenix consume about 385 million gallons of water per year," your immediate first thought is "holy shit, that's a lot of water." But then when you understand the context - that that's less than one tenth of one percent of water consumption in Phoenix, your mind boggles in the other direction.
I don't think it's particularly different from any other topic. Seems like these days (maybe it was always like this?) everyone needs a boogieman to hate on, no matter what reality is. For some...
I don't think it's particularly different from any other topic. Seems like these days (maybe it was always like this?) everyone needs a boogieman to hate on, no matter what reality is. For some people it's immigrants or trans people, for some it's AI (for me it's the country of Sweden lol). I guess the scary part is how everybody is eager to point out how those people have got it all wrong, man they must be really stupid and gullible. But then do the exact same thing with another topic.
Agreed. Sure, you can run a "closed loop" system, but unless you do water processing and filtration on site, you still need large amounts of water on a fixed cadence. Similar with electricity:...
Agreed. Sure, you can run a "closed loop" system, but unless you do water processing and filtration on site, you still need large amounts of water on a fixed cadence. Similar with electricity: sure, you can generate it onsite! If you choose to do so with renewables, it might be hard to complain. But the simplest and cheapest generators seem to be... natural gas or diesel.
In a good world, where these things are regulated, individuals might trust that The Powers That Be will limit externalities. But that is not this world.
And it goes without saying that when these datacenters consume upwards of a gigawatt of electricity, the externalities can be similarly massive. IIRC the entire state of Vermont typically uses a little over half a gigawatt of electricity, and peak usage is still less than 1.5GW.
Phoenix needs to be more water efficient, like Las Vegas. Living in a desert is fine with the right adaptations. Nobody is pushing hard enough for those changes (especially around agricultural...
There's also the issue of overbuilding housing due to the government over-zoning when not having secured water yet
Phoenix needs to be more water efficient, like Las Vegas. Living in a desert is fine with the right adaptations. Nobody is pushing hard enough for those changes (especially around agricultural water usage).
Well, they're changing their corporate charter to remove mentions of environmental sustainability, so it kinda sounds like that was their conclusion. At the very least, they decided that it was a...
Well, they're changing their corporate charter to remove mentions of environmental sustainability, so it kinda sounds like that was their conclusion. At the very least, they decided that it was a liability in their new market.
Generally yeah, but Allbirds had a reputation for trying to make shoes as close as possible to net-zero emissions. If they went for the same angle of "we wanna try to do AI as close to net-zero as...
Generally yeah, but Allbirds had a reputation for trying to make shoes as close as possible to net-zero emissions.
If they went for the same angle of "we wanna try to do AI as close to net-zero as possible" I could maybe understand, but otherwise this is a really jarring shift.
That Microsoft is paying to reactivate a nuclear reactor and prebuy all the energy tells me that it's not a small percentage more than typical cloud operations.
That Microsoft is paying to reactivate a nuclear reactor and prebuy all the energy tells me that it's not a small percentage more than typical cloud operations.
No, I've read various comparisons but it seems difficult to figure out, which is why I'm suspicious of confident assertions. For customers, I've seen a comparison that using AI and watching...
No, I've read various comparisons but it seems difficult to figure out, which is why I'm suspicious of confident assertions.
For customers, I've seen a comparison that using AI and watching Netflix are in the same ballpark, but I have low confidence in that.
Throw a rock at any of the techier tech conferences and you'll hit at least one of us who was there in the <$10 days playing with GPU hashing algorithms or working on the early VHDL code because...
Throw a rock at any of the techier tech conferences and you'll hit at least one of us who was there in the <$10 days playing with GPU hashing algorithms or working on the early VHDL code because it was interesting, and unknowingly held what would be a meaningful chunk of a billion dollars when you look back at it now.
Weirdly for someone with as much anxiety as I have in day to day life, I've never actually had serious "what ifs" or regrets about not holding on to that back in the day. It was so literally unknowable even with hindsight that it's just abstract. I think the one thing it really did do is put the final nail in the already pretty well closed coffin of my belief in our assignments of value being anything other than an absurdist joke, and honestly I do think holding onto that knowledge actually has served me well.
There were two strong hunches I had in that general era. One was that Bitcoin would easily go for $10k a pop once the finance circlejerk bros heard about it. 'Get in now and have a piece of a...
There were two strong hunches I had in that general era.
One was that Bitcoin would easily go for $10k a pop once the finance circlejerk bros heard about it. 'Get in now and have a piece of a limited-issue currency.'
And the other is that I should have defaulted on all of my debt and sunk every penny I had into AMD at $2 a share. I figured much like Microsoft propping up Apple, that worst case Intel would throw AMD a softball in order to keep them alive and stave off monopoly problems. Ryzen showed promise of being a borderline revolutionary architecture.
I listened to my rational self and I'm in a pretty stable position now, but I'd be rolling in some serious money had I listened to my manic hunches.
It's artificially inflated by Trump's policies as well as increased international money laundering. It's important to remember than Bernie Madoff's scheme lasted decades.
It's artificially inflated by Trump's policies as well as increased international money laundering. It's important to remember than Bernie Madoff's scheme lasted decades.
Genuinely thought this was an April Fool's bit, just 2 weeks late. I've enjoyed Allbirds shoes for the last few years since they're comfortable to me and I like the minimalist aesthetic. I wonder...
Genuinely thought this was an April Fool's bit, just 2 weeks late. I've enjoyed Allbirds shoes for the last few years since they're comfortable to me and I like the minimalist aesthetic. I wonder where I can go on from here. I was thinking of just joining the mainstream and getting Adidas Samba shoe or something but I'm not sure.
Adidas are pretty good fwiw and ubiquitous. I buy pairs on sale, usually under 50$ and they last me 2–4 years typically. They’re comfortable and well made. I buy a new pair way before mine are...
Adidas are pretty good fwiw and ubiquitous. I buy pairs on sale, usually under 50$ and they last me 2–4 years typically. They’re comfortable and well made. I buy a new pair way before mine are shot so that I can grab a really good deal and never be caught short.
Yeah I've had good luck with Adidas in the past, which is why the Samba came to mind. I used to have their Cloudfoam shoes which felt similar to certain Allbirds models. I actually just got a new...
Yeah I've had good luck with Adidas in the past, which is why the Samba came to mind. I used to have their Cloudfoam shoes which felt similar to certain Allbirds models. I actually just got a new pair of Allbirds so I've got time before I need to replace it.
I feel like this is more about how you define "ownership". Like, imagine I owned some vending machines. I sold those vending machines to someone else and bought some coin laundry machines instead....
I feel like this is more about how you define "ownership". Like, imagine I owned some vending machines. I sold those vending machines to someone else and bought some coin laundry machines instead.
Did a vending machine just turn into a laundry machine!? Probably not how most people would see it.
The vending machine is still there. And the shoe company is still there. It's just owned by American Exchange Group, instead of the previous parent company, which used the proceeds to try and get into the AI datacenter space.
On the other hand, you have pivoted from a vending business to a laundry business by doing that. Allbirds the company - as in, the legal entity and investment vehicle that people have bought...
On the other hand, you have pivoted from a vending business to a laundry business by doing that. Allbirds the company - as in, the legal entity and investment vehicle that people have bought shares in for any number of reasons over the last few years, rather than the people and machinery who make shoes - really has changed from one thing to another while bringing those investors along for the ride.
I'm sure those investors are happy enough about it, given the numbers involved, but there is a continuous entity making a big change here in a way that it wouldn't be if the founders had wound down Allbirds and then used the money to start NewBird AI.
I also don't really understand the spike, I don't see what qualifies these guys to get into the GPU datacenter space any more than any random startup, but I guess it's effectively a datacenter startup with a built in SPAC, and perhaps that has some additional perceived value to investors compared to any other equally untested startup.
Nor was it a surprise - I'm sure some of them just sold their equities and dipped once the transaction was completed. That is an option available to them. The ones that stick around, evidently want to do some AI stuff!
The legal entity and the investors have little to do with the production of the shoe. The employees and technology and factory and design and supply chains are now under American Exchange Group.
there is a continuous entity making a big change here in a way that it wouldn't be if the founders had wound down
In some sense, but, like, who cares? It's investor inside baseball. For all intents and purposes AllBirds was sold to American Exchange Group, and then the people who owned the company regrouped and are doing something new. Unless you were an investor in AllBirds (in which case, I'd certainly hope you'd read the letters informing you of the vote!), there's really no reason to care.
I'm more or less with you - I think "investor inside baseball" is actually a great way of putting it - and that's why I thought it was worth shining a bit of a light on. For people familiar with...
I'm more or less with you - I think "investor inside baseball" is actually a great way of putting it - and that's why I thought it was worth shining a bit of a light on. For people familiar with this kind of stuff it makes sense (like I said, spinning their own new startup into what's effectively a SPAC is how I read it), but for people outside baseball I think it can simultaneously be technically correct but still misleading to suggest nothing's really changed.
They've fairly deftly used their publicly traded status, investor recognition, and honestly probably a decent bit of investor ignorance and/or inertia to boost the profile of their new venture well above where it would've been as just another datacenter startup from people who have no real reason to be exciting as datacenter startup founders. They've also saved themselves seven, maybe even eight figures in real concrete overheads from not having to do another IPO.
It's nothing too wild for someone immersed in this world, and it's certainly not "shoe factory is now full of GPUs lol" which is obviously how the press are spinning it, but it's the kind of thing that sits far enough outside the baseline simplified understanding of how markets and companies operate and close enough to how they actually operate that I think it's good to show the working, so to speak, rather than just jumping to the fact that the result of the big scary equation all just cancels out.
This is so stupid, it honestly feels like a free money hack for the wealthy. Buy a struggling but well known business Sell off it's IP Announce a pivot to AI Congrats, you're even more wealthy
This is so stupid, it honestly feels like a free money hack for the wealthy.
Steps 1 and 2 are redundant. You paid $X to buy a company. You sell it for $X. Now you have the same amount of money lol. You just bought the world’s most expensive corporate registration. You can...
Steps 1 and 2 are redundant. You paid $X to buy a company. You sell it for $X. Now you have the same amount of money lol. You just bought the world’s most expensive corporate registration.
You can just announce a new AI company with the money you would have spent in step 1.
Those are not really comparable. Starting your own company means you need to all the work required to get it on the exchange and to get ticker recognition. Hence step 1.
Those are not really comparable. Starting your own company means you need to all the work required to get it on the exchange and to get ticker recognition. Hence step 1.
It also means you're burdened with GAAP accounting and SEC reporting as a new company when most startups would not have to do that. Companies IPO or direct list later in their life just as much...
It also means you're burdened with GAAP accounting and SEC reporting as a new company when most startups would not have to do that. Companies IPO or direct list later in their life just as much because it's advantageous not to be a public company in the early days as it is because of complexity.
Being a publicly traded company is not that expensive or complex of a process. Just see SPACs, which are founded are purely floating sacks of publicly traded money.
It is typically a great deal of effort for an existing large company because suddenly you have strict reporting regulations you have to fulfill. IPOs are expensive, but not required, as you can just do a direct listing for essentially no cost.
This is sad, as I recently bought a pair of allbirds and have enjoyed them. I was looking for a minimal pair of black runners with a white sole, and mainstream brands just did not have good...
This is sad, as I recently bought a pair of allbirds and have enjoyed them. I was looking for a minimal pair of black runners with a white sole, and mainstream brands just did not have good offerings. I also enjoyed how allbirds did not plaster their logos all over the shoes, like most shoe brands do.
When buying the shoes, I did realize that I have been tending to buy from smaller companies and wondered "will this company still be around when I need to replace it?" I am in the same boat with my phone, which I purchased a neo-dumbphone about a year ago, and I have doubts the company will still be around in a few years when I need to replace it.
For what it is worth, I like Kiziks. They have some fairly minimal shoes. Their big deal is that they are so slip ons with hard backs (though they have faux laces). Nice for me being with kids all...
For what it is worth, I like Kiziks. They have some fairly minimal shoes. Their big deal is that they are so slip ons with hard backs (though they have faux laces). Nice for me being with kids all day— in and out the door constantly, holding an infant etc. My previous shoes were allbirds.
They sold the shoe brand off to another company. In a year or two when I need to replace mine, I might consider them, but I would want to wait and see how this new company handles the brand
They sold the shoe brand off to another company. In a year or two when I need to replace mine, I might consider them, but I would want to wait and see how this new company handles the brand
I've seen people imply that this website is pro AI a couple of times now, and it's always fascinated me because I would describe tildes as an overwhelmingly anti-AI space. I have no idea which one...
I've seen people imply that this website is pro AI a couple of times now, and it's always fascinated me because I would describe tildes as an overwhelmingly anti-AI space. I have no idea which one of us is factually correct. But I think both of us have reasonable grounds for our subjective opinions/vibes, and that's... I don't want to say dangerous, exactly, because this is just a chatroom, so let's just say "not good." I don't think it's good that both anti-AI people and pro-AI people have reason to feel like they're part of a minority that's being crushed. (I hope you'll forgive the inference, but the way you phrased your post defensively suggests that's how you feel.)
It's an emotional topic, for sure. And it's a political topic, which is worse. But the dialogue does not seem to be in a good place.
I feel like this is quite common on all social media platforms. Redditors often claim that Reddit is simultaneously a conservative hellhole and a leftist hellhole. I think for a lot of people the...
I feel like this is quite common on all social media platforms. Redditors often claim that Reddit is simultaneously a conservative hellhole and a leftist hellhole.
I think for a lot of people the standard for “site supports thing I don’t like” is whether or not 100% of people don’t like it or not.
I think certain views on this site pop up much quicker without competing top-level responses. You can definitely say this site isn't just left-leaning, but nearly entirely on the left, considering...
I think certain views on this site pop up much quicker without competing top-level responses. You can definitely say this site isn't just left-leaning, but nearly entirely on the left, considering the only conservative/right-wing/republican-aligned member I know of appears to have been permanently banned.
Much how Nebula skews lefty, it's in part because fact checking/bullshit calling is a fairly important part of the process. Reality liberal bias and all that. The fact that bigoted views are...
Much how Nebula skews lefty, it's in part because fact checking/bullshit calling is a fairly important part of the process. Reality liberal bias and all that.
The fact that bigoted views are basically 0 tolerance kinda seals the deal.
You'll find the economic conservatives and some landlords, but they're nice enough and knowledgable people even if I very much dislike their stances on things.
Yes, this is a good description of @gowestyoungman, so I'm curious what the story was behind his ban.
You'll find the economic conservatives and some landlords, but they're nice enough and knowledgable people even if I very much dislike their stances on things.
Yes, this is a good description of @gowestyoungman, so I'm curious what the story was behind his ban.
Not saying this happened at all (I also have no clue), but I can imagine something akin to this microstory could have transpired: An uncle comes to Thanksgiving. He has 3 drinks. Then there was a...
Not saying this happened at all (I also have no clue), but I can imagine something akin to this microstory could have transpired:
An uncle comes to Thanksgiving. He has 3 drinks. Then there was a brief mention of the stock market and he suddenly starts ranting about how the Jews ruin everything.
My impression of Tildes is that it's a pretty mixed bag when it comes to opinions on AI, with opinions from across the spectrum showing up frequently enough when it comes up. But I think...
My impression of Tildes is that it's a pretty mixed bag when it comes to opinions on AI, with opinions from across the spectrum showing up frequently enough when it comes up. But I think negativity bias makes the opinions that one disagrees with stand out more, especially when they've been upvoted a bunch (which happens to both types of comments from what I can see, since it's a very emotional, polarizing issue and people are thus more likely to go out of their way to upvote comments they agree with about it).
But that all might just be me being biased by having seen opinions on either side of mine that I thought were too extreme in comments on Tildes before, enlightened centrist that I am.
I would guess that the majority of tildes thinks AI is useful in some way. The people here fervently against AI seem like a minority. I would guess they are a large minority (30/45%), but a...
I would guess that the majority of tildes thinks AI is useful in some way. The people here fervently against AI seem like a minority. I would guess they are a large minority (30/45%), but a minority nonetheless. However, assuming the people who like and use AI don't want the bubble to pop isn't a safe assumption. Tulip prices being high doesn't benefit the home gardener that just wants some tulips. In fact, the bubble popping is fantastic for those people. I have a Claude and chatgpt subscription and use them almost daily, and the bubble popping would be fantastic in my opinion (as long as it doesn't bring down the whole economy, which is certainly possible).
I agree! Very interesting. I imagine it's relative based on where else we each hang out online, and how we each ended up here on tildes. From my perspective this community was born out of waves of...
I agree! Very interesting. I imagine it's relative based on where else we each hang out online, and how we each ended up here on tildes. From my perspective this community was born out of waves of anti-reddit sentiment as the corporation abused its power to exploit human social life in service of extracting wealth. That destructive nature is the same in AI corporations, so I would expect tildes to have an antibody response to AI boosting that I just don't see much of.
I'm also sensitive to the topic, as most AI detractors are, and any misalignment of these values in a community I thought largely shared them feels particularly sharp.
The only reason I can make any sense of it is that the stock price had already dropped to almost 0 and it was assumed by most that they would cease to exist relatively soon.
The only reason I can make any sense of it is that the stock price had already dropped to almost 0 and it was assumed by most that they would cease to exist relatively soon.
(/noise) This is a good opportunity to announce that I am also pivoting into AI compute, and will be investing in a new graphics card for my computer. Pls gib moneys naow ༼ つ ◕_◕ ༽つ
(/noise)
This is a good opportunity to announce that I am also pivoting into AI compute, and will be investing in a new graphics card for my computer.
This is so ridiculous I had to share it.
I used to be a happy Allbirds customer and I did enjoy their shoes in the past, but then they started struggling, stopped offering sales and IIRC started cutting corners on quality.
Now they're an AI company???
EDIT: Also, I hope the irony is not lost on them going from being an environment-first manufacturer to joining one of the most environmentally harmful industries in the world. But hey, investors are happy, so does it even matter?
For what it's worth, training GPT-4 -- not using but training, the most carbon-intensive part -- produced about half as much emissions as one day of Nike's shoe production.
I don't want to be taken as a brain-dead AI booster. I'm not telling you not to criticize AI. My intent here is to provide more facts/context to what is a super common claim.
Isn't training only the most energy-intensive part right at release, before a model actually gets used? I would think that over time the inference cost catches up since the inference cost keeps adding up over time.
That's kind of like saying the vehicle manufacturing is the most energy-intensive part of making a car: surely true for a lot of vehicles, but if something gets heavily used, I'm sure the operation costs catch up.
Yes! All true.
I cited training for two reasons. One because of the convenient shoe comparison, which was relevant. Two because the resource cost of actually using the models is so marginally low that it's almost not worth discussing (fractions of a percentage point of one's daily carbon/water consumption), so people tend to focus on training costs, which are (mistakenly) believed to be extremely high.
Training is a small part of it; you still have the AI industry's chokehold on electronics manufacturing (and the environmental impact of that), the damaging levels of noise pollution the data centers make, and the toll on power plants.
and freshwater!
Water is probably the least effective argument. I'm not sure why it comes up as much as it does. For one, it's not like they're performing water electroysis, the water is still there. Most datacenters are closed loop - not for environmental reasons, just because it's cheaper. Even in the ones that aren't, y'know, water cycle and all.
All the data centers popping up in drought zones actually does make it super annoying. Phoenix area having multiple built up, likely because of cheaper land and resources.
The fact it also is dry likely plays into location being chosen, but of course this will continue taxing locals even worse with multiple price increases on a limited resource.
Fact is they're not able to even be compliant with the state constitution that requires the government to have 100 years of water available because of climate change and these combining into an unachievable situation.
Edit: There's also the issue of overbuilding housing due to the government over-zoning when not having secured water yet, so there's multiple compounding issues at play.
Is there any indication that it’s having a real impact on water demand in any region of the US? A quick google on phoenix datacenter water usage gets the following articles
https://grist.org/technology/arizona-water-data-centers-semiconducters/
Not nothing but not exactly catastrophic demand increases
There’s this Reuters article
https://www.reuters.com/sustainability/climate-energy/desert-storm-can-data-centres-slake-their-insatiable-thirst-water--ecmii-2025-12-17/
TLDR is that the data centers are designed to use zero water but that there was an indirect water cost that the article is bringing attention to in the form of increased water usage in power plants. Neat, but zero figures.
Then there’s this local news article
https://www.azcentral.com/story/money/business/tech/2026/02/04/how-much-water-power-do-arizona-data-centers-use/87270206007/?gnt-cfr=1&gca-cat=p&gca-uir=false&gca-epti=undefined&gca-ft=0&gca-ds=sophi
Their conclusion was essentially: idk
And that’s skipping all the articles that basically are about how XYZ datacenter is designed to use no water like https://datacenter.news/story/edged-us-opens-waterless-ai-data-centre-in-mesa-arizona
What makes understanding AI so difficult is that 1) we don't realize how many resources we're already consuming and 2) these numbers don't mesh well with the human mind. Like if you just read "data centers in Phoenix consume about 385 million gallons of water per year," your immediate first thought is "holy shit, that's a lot of water." But then when you understand the context - that that's less than one tenth of one percent of water consumption in Phoenix, your mind boggles in the other direction.
I don't think it's particularly different from any other topic. Seems like these days (maybe it was always like this?) everyone needs a boogieman to hate on, no matter what reality is. For some people it's immigrants or trans people, for some it's AI (for me it's the country of Sweden lol). I guess the scary part is how everybody is eager to point out how those people have got it all wrong, man they must be really stupid and gullible. But then do the exact same thing with another topic.
Agreed. Sure, you can run a "closed loop" system, but unless you do water processing and filtration on site, you still need large amounts of water on a fixed cadence. Similar with electricity: sure, you can generate it onsite! If you choose to do so with renewables, it might be hard to complain. But the simplest and cheapest generators seem to be... natural gas or diesel.
In a good world, where these things are regulated, individuals might trust that The Powers That Be will limit externalities. But that is not this world.
And it goes without saying that when these datacenters consume upwards of a gigawatt of electricity, the externalities can be similarly massive. IIRC the entire state of Vermont typically uses a little over half a gigawatt of electricity, and peak usage is still less than 1.5GW.
Phoenix needs to be more water efficient, like Las Vegas. Living in a desert is fine with the right adaptations. Nobody is pushing hard enough for those changes (especially around agricultural water usage).
Why wouldn’t manufacturing shoes be more environmentally harmful? It seems like a wild assumption to make without investigating.
Well, they're changing their corporate charter to remove mentions of environmental sustainability, so it kinda sounds like that was their conclusion. At the very least, they decided that it was a liability in their new market.
Generally yeah, but Allbirds had a reputation for trying to make shoes as close as possible to net-zero emissions.
If they went for the same angle of "we wanna try to do AI as close to net-zero as possible" I could maybe understand, but otherwise this is a really jarring shift.
Do you have any idea how much energy AI is consuming right now?
That Microsoft is paying to reactivate a nuclear reactor and prebuy all the energy tells me that it's not a small percentage more than typical cloud operations.
No, I've read various comparisons but it seems difficult to figure out, which is why I'm suspicious of confident assertions.
For customers, I've seen a comparison that using AI and watching Netflix are in the same ballpark, but I have low confidence in that.
It's the same plot of Long Island Iced Tea pivoting to blockchain in 2017.
The bubble is inflating and about to pop.
Augh ... the price of Bitcoin was ~6500 USD in 2017. It's currently ~74000 USD 😬 when will it end ...
Throw a rock at any of the techier tech conferences and you'll hit at least one of us who was there in the <$10 days playing with GPU hashing algorithms or working on the early VHDL code because it was interesting, and unknowingly held what would be a meaningful chunk of a billion dollars when you look back at it now.
Weirdly for someone with as much anxiety as I have in day to day life, I've never actually had serious "what ifs" or regrets about not holding on to that back in the day. It was so literally unknowable even with hindsight that it's just abstract. I think the one thing it really did do is put the final nail in the already pretty well closed coffin of my belief in our assignments of value being anything other than an absurdist joke, and honestly I do think holding onto that knowledge actually has served me well.
There were two strong hunches I had in that general era.
One was that Bitcoin would easily go for $10k a pop once the finance circlejerk bros heard about it. 'Get in now and have a piece of a limited-issue currency.'
And the other is that I should have defaulted on all of my debt and sunk every penny I had into AMD at $2 a share. I figured much like Microsoft propping up Apple, that worst case Intel would throw AMD a softball in order to keep them alive and stave off monopoly problems. Ryzen showed promise of being a borderline revolutionary architecture.
I listened to my rational self and I'm in a pretty stable position now, but I'd be rolling in some serious money had I listened to my manic hunches.
Probably once somebody powerful enough gets had by yet another cypto scam.
It's artificially inflated by Trump's policies as well as increased international money laundering. It's important to remember than Bernie Madoff's scheme lasted decades.
Genuinely thought this was an April Fool's bit, just 2 weeks late. I've enjoyed Allbirds shoes for the last few years since they're comfortable to me and I like the minimalist aesthetic. I wonder where I can go on from here. I was thinking of just joining the mainstream and getting Adidas Samba shoe or something but I'm not sure.
Adidas are pretty good fwiw and ubiquitous. I buy pairs on sale, usually under 50$ and they last me 2–4 years typically. They’re comfortable and well made. I buy a new pair way before mine are shot so that I can grab a really good deal and never be caught short.
Yeah I've had good luck with Adidas in the past, which is why the Samba came to mind. I used to have their Cloudfoam shoes which felt similar to certain Allbirds models. I actually just got a new pair of Allbirds so I've got time before I need to replace it.
I feel like this is more about how you define "ownership". Like, imagine I owned some vending machines. I sold those vending machines to someone else and bought some coin laundry machines instead.
Did a vending machine just turn into a laundry machine!? Probably not how most people would see it.
The vending machine is still there. And the shoe company is still there. It's just owned by American Exchange Group, instead of the previous parent company, which used the proceeds to try and get into the AI datacenter space.
On the other hand, you have pivoted from a vending business to a laundry business by doing that. Allbirds the company - as in, the legal entity and investment vehicle that people have bought shares in for any number of reasons over the last few years, rather than the people and machinery who make shoes - really has changed from one thing to another while bringing those investors along for the ride.
I'm sure those investors are happy enough about it, given the numbers involved, but there is a continuous entity making a big change here in a way that it wouldn't be if the founders had wound down Allbirds and then used the money to start NewBird AI.
I also don't really understand the spike, I don't see what qualifies these guys to get into the GPU datacenter space any more than any random startup, but I guess it's effectively a datacenter startup with a built in SPAC, and perhaps that has some additional perceived value to investors compared to any other equally untested startup.
There's no real difference. The investors are the owners of the company, just more abstract. They did approve the sale, after all: https://www.tipranks.com/news/company-announcements/allbirds-secures-shareholder-support-for-strategic-asset-sale.
Nor was it a surprise - I'm sure some of them just sold their equities and dipped once the transaction was completed. That is an option available to them. The ones that stick around, evidently want to do some AI stuff!
The legal entity and the investors have little to do with the production of the shoe. The employees and technology and factory and design and supply chains are now under American Exchange Group.
In some sense, but, like, who cares? It's investor inside baseball. For all intents and purposes AllBirds was sold to American Exchange Group, and then the people who owned the company regrouped and are doing something new. Unless you were an investor in AllBirds (in which case, I'd certainly hope you'd read the letters informing you of the vote!), there's really no reason to care.
I'm more or less with you - I think "investor inside baseball" is actually a great way of putting it - and that's why I thought it was worth shining a bit of a light on. For people familiar with this kind of stuff it makes sense (like I said, spinning their own new startup into what's effectively a SPAC is how I read it), but for people outside baseball I think it can simultaneously be technically correct but still misleading to suggest nothing's really changed.
They've fairly deftly used their publicly traded status, investor recognition, and honestly probably a decent bit of investor ignorance and/or inertia to boost the profile of their new venture well above where it would've been as just another datacenter startup from people who have no real reason to be exciting as datacenter startup founders. They've also saved themselves seven, maybe even eight figures in real concrete overheads from not having to do another IPO.
It's nothing too wild for someone immersed in this world, and it's certainly not "shoe factory is now full of GPUs lol" which is obviously how the press are spinning it, but it's the kind of thing that sits far enough outside the baseline simplified understanding of how markets and companies operate and close enough to how they actually operate that I think it's good to show the working, so to speak, rather than just jumping to the fact that the result of the big scary equation all just cancels out.
This is so stupid, it honestly feels like a free money hack for the wealthy.
Steps 1 and 2 are redundant. You paid $X to buy a company. You sell it for $X. Now you have the same amount of money lol. You just bought the world’s most expensive corporate registration.
You can just announce a new AI company with the money you would have spent in step 1.
Those are not really comparable. Starting your own company means you need to all the work required to get it on the exchange and to get ticker recognition. Hence step 1.
It also means you're burdened with GAAP accounting and SEC reporting as a new company when most startups would not have to do that. Companies IPO or direct list later in their life just as much because it's advantageous not to be a public company in the early days as it is because of complexity.
Being a publicly traded company is not that expensive or complex of a process. Just see SPACs, which are founded are purely floating sacks of publicly traded money.
It is typically a great deal of effort for an existing large company because suddenly you have strict reporting regulations you have to fulfill. IPOs are expensive, but not required, as you can just do a direct listing for essentially no cost.
This is sad, as I recently bought a pair of allbirds and have enjoyed them. I was looking for a minimal pair of black runners with a white sole, and mainstream brands just did not have good offerings. I also enjoyed how allbirds did not plaster their logos all over the shoes, like most shoe brands do.
When buying the shoes, I did realize that I have been tending to buy from smaller companies and wondered "will this company still be around when I need to replace it?" I am in the same boat with my phone, which I purchased a neo-dumbphone about a year ago, and I have doubts the company will still be around in a few years when I need to replace it.
For what it is worth, I like Kiziks. They have some fairly minimal shoes. Their big deal is that they are so slip ons with hard backs (though they have faux laces). Nice for me being with kids all day— in and out the door constantly, holding an infant etc. My previous shoes were allbirds.
I will bookmark this as an option when I am looking for shoes in a year or two
Do they run wide enough? My biggest issue is that tight toe boxes are all too common for aesthetic reasons.
Why wouldn’t you be able to continue buying allbirds shoes?
They sold the shoe brand off to another company. In a year or two when I need to replace mine, I might consider them, but I would want to wait and see how this new company handles the brand
Hopefully a sign of the AI bubble popping soon, though I imagine that won't be a popular opinion on tildes.
I've seen people imply that this website is pro AI a couple of times now, and it's always fascinated me because I would describe tildes as an overwhelmingly anti-AI space. I have no idea which one of us is factually correct. But I think both of us have reasonable grounds for our subjective opinions/vibes, and that's... I don't want to say dangerous, exactly, because this is just a chatroom, so let's just say "not good." I don't think it's good that both anti-AI people and pro-AI people have reason to feel like they're part of a minority that's being crushed. (I hope you'll forgive the inference, but the way you phrased your post defensively suggests that's how you feel.)
It's an emotional topic, for sure. And it's a political topic, which is worse. But the dialogue does not seem to be in a good place.
I think it tends to be fairly “pro” for coding with more recent models and fairly “anti” for almost everything else.
So it’s both and neither.
I feel like this is quite common on all social media platforms. Redditors often claim that Reddit is simultaneously a conservative hellhole and a leftist hellhole.
I think for a lot of people the standard for “site supports thing I don’t like” is whether or not 100% of people don’t like it or not.
I think certain views on this site pop up much quicker without competing top-level responses. You can definitely say this site isn't just left-leaning, but nearly entirely on the left, considering the only conservative/right-wing/republican-aligned member I know of appears to have been permanently banned.
Much how Nebula skews lefty, it's in part because fact checking/bullshit calling is a fairly important part of the process. Reality liberal bias and all that.
The fact that bigoted views are basically 0 tolerance kinda seals the deal.
You'll find the economic conservatives and some landlords, but they're nice enough and knowledgable people even if I very much dislike their stances on things.
Yes, this is a good description of @gowestyoungman, so I'm curious what the story was behind his ban.
Not saying this happened at all (I also have no clue), but I can imagine something akin to this microstory could have transpired:
An uncle comes to Thanksgiving. He has 3 drinks. Then there was a brief mention of the stock market and he suddenly starts ranting about how the Jews ruin everything.
Now he's forever your antisemetic uncle.
Goomba fallacy.
My impression of Tildes is that it's a pretty mixed bag when it comes to opinions on AI, with opinions from across the spectrum showing up frequently enough when it comes up. But I think negativity bias makes the opinions that one disagrees with stand out more, especially when they've been upvoted a bunch (which happens to both types of comments from what I can see, since it's a very emotional, polarizing issue and people are thus more likely to go out of their way to upvote comments they agree with about it).
But that all might just be me being biased by having seen opinions on either side of mine that I thought were too extreme in comments on Tildes before, enlightened centrist that I am.
I would guess that the majority of tildes thinks AI is useful in some way. The people here fervently against AI seem like a minority. I would guess they are a large minority (30/45%), but a minority nonetheless. However, assuming the people who like and use AI don't want the bubble to pop isn't a safe assumption. Tulip prices being high doesn't benefit the home gardener that just wants some tulips. In fact, the bubble popping is fantastic for those people. I have a Claude and chatgpt subscription and use them almost daily, and the bubble popping would be fantastic in my opinion (as long as it doesn't bring down the whole economy, which is certainly possible).
I agree! Very interesting. I imagine it's relative based on where else we each hang out online, and how we each ended up here on tildes. From my perspective this community was born out of waves of anti-reddit sentiment as the corporation abused its power to exploit human social life in service of extracting wealth. That destructive nature is the same in AI corporations, so I would expect tildes to have an antibody response to AI boosting that I just don't see much of.
I'm also sensitive to the topic, as most AI detractors are, and any misalignment of these values in a community I thought largely shared them feels particularly sharp.
I am over the moon for this to be over, at least XD
If the news was just about the pivot, then okay, interesting. But the 700% surge makes me suspect Idiocracy is a documentary.
The only reason I can make any sense of it is that the stock price had already dropped to almost 0 and it was assumed by most that they would cease to exist relatively soon.
Are you sure that this isn’t a late witted April fools joke?
They better change the name to Aibirds to reflect their new direction.
But ai means love in Japanese, so that would be a kawaii brand with a name implying Lovebirds.
(/noise)
This is a good opportunity to announce that I am also pivoting into AI compute, and will be investing in a new graphics card for my computer.
Pls gib moneys naow ༼ つ ◕_◕ ༽つ