Big tech is playing its part in reaching net zero targets, but its vast new datacentres are run at huge cost to the environment, says economics professor Mariana Mazzucato
When you picture the tech industry, you probably think of things that don’t exist in physical space, such as the apps and internet browser on your phone. But the infrastructure required to store all this information – the physical datacentres housed in business parks and city outskirts – consume massive amounts of energy. Despite its name, the infrastructure used by the “cloud” accounts for more global greenhouse emissions than commercial flights. In 2018, for instance, the 5bn YouTube hits for the viral song Despacito used the same amount of energy it would take to heat 40,000 US homes annually.
This is a hugely environmentally destructive side to the tech industry. While it has played a big role in reaching net zero, giving us smart meters and efficient solar, it’s critical that we turn the spotlight on its environmental footprint. Large language models such as ChatGPT are some of the most energy-guzzling technologies of all. Research suggests, for instance, that about 700,000 litres of water could have been used to cool the machines that trained ChatGPT-3 at Microsoft’s data facilities. It is hardly news that the tech bubble’s self-glorification has obscured the uglier sides of this industry, from its proclivity for tax avoidance to its invasion of privacy and exploitation of our attention span. The industry’s environmental impact is a key issue, yet the companies that produce such models have stayed remarkably quiet about the amount of energy they consume – probably because they don’t want to spark our concern.
I never understand this line of thought. The amounts of energy we use is never ever going to go down. It just isn't. This shouldn't be an argument against tech using power. It should be an argument for clean energy solutions.
I sort of agree with you, but also think we need to incentivize efficiency (or disincentivize inefficiency). As mentioned in the article, there is also the issue of the immense quantities of water used by data centers, a byproduct of power use and inefficiency. If we could at least capture and store that heat energy to do something useful, it would be a huge improvement.
This makes sense to me, we could continue to develop and use important technologies while at the same time setting things up so their externalities are part of their cost and companies have financial reasons to work to reduce the impact.
Not really. Every energy solution produces some kind of waste which can't be recycled. Saving energy is always good. It also saves budget and space. I'd say your opinion is a contribution to very unsustainable future.
I never understand this line of thought. The amounts of energy we use is never ever going to go down. It just isn’t.
If we don't develop practical nuclear fusion before our fossil inheritance effectively runs out we sure will. It will also go down following ecological collapse caused by using all that energy. Infinite energy doesn't make up for a collapsed ecosystem.
Just to add: Even if we can replace the energy from diminishing fossils with nuclear etc, there are still a huge forest/mountain of essential technology and products that are reliant on fossils, and they won't be replaced by anything. I can recommend Nate Hagens on YT for more on the 'energy blindness' issue, and what it means for our civilization to lose the last.
The amounts of energy we use is never ever going to go down. It just isn't.
Not voluntarily maybe. But that’s not the only way. The only outcomes of any “realistic” course correction to the current state of the world and human behavior include widespread death and societal collapse once uncontrolled climate change takes hold for real, and at that point, it’ll go down quite a bit.
I think the real question is what is the value we are getting out of the resources used? Do we need AI forced into every platform? Personally, I don't think so. But just to be sure I asked Chat GPT and here is its answer:
"Rather than integrating Al into every possible application, a more measured approach might be beneficial. Assessing the actual need and impact of Al in specific use cases can help avoid unnecessary energy consumption. Al should be implemented where it provides significant benefits and
improvements, rather than as a default addition to every platform."
So even the AI itself knows it is used frivolously.
A large part of creation is trying things and seeing what sticks. Nobody is claiming that every way LLMs are being tried out today will always be here. We are just doing what we think of and seeing what is useful. The useful things will stick around and evolve, other things won't. Go back to videos from the early 90s when computers were starting, people talked so much shit on them. Now, we all have the future generations of them in our pockets.
I staunchly disagree with this. If a government actually cared about power consumption, they would subsidize the development of better X86 to arm translation layers so we can move to more power efficient processors.
There's actually a lot we could do in order to reduce our energy consumption.
Another step I would take is completely outlawing the sale of advertisements to data brokers. There are so many resources that go to pumping out advertisements from servers, and it really just is not beneficial to anybody except for the companies.
I know this would not be popular among a lot of developers, but for certain applications, I would also curb planned obsolescence by creating a minimum viable processor and graphics requirement for certain applications. A lot of the times, the kinds of applications we use just don't need as much power as they're guzzling We need to end the feature creep.
I mean you can disagree with it but it doesn't make it any less true. Banning data brokers wouldnt do jack shit, governments are so technically illiterate that they dont even understand what you just said, nevermind know what ARM is, and youre also wrong about power usage in applications.
Opportunity cost is a pretty well understood concept.
Like, inagine you have 100 gallons of water. You could use all of them to water a single water intensive plant that will feed one person, or you could use them to water a whole farm that will feed a community, and also let people drink and bathe and stuff.
The resource is limited.
Sure, we could try to get more of the resource and make it less expensive, but we should also not squander what we have.
Same thing whenever you see articles about Bitcoin's energy use. Or the energy usage of any tech service/product:
For some reason blames the product or service people are using, not politicians for failing for decades to invest in renewable energy.
No contextual information (how much does the remittance industry use? How much energy does SWIFT, IBAN, or printing paper money use)? How efficiently do these systems actually use energy?
No mention of the many useful things it does with that energy or why it uses energy in the way it does. (Send money across the globe in under a second for under a penny in fees to anybody with a cellphone and halfway reliable internet) (low fees available on Bitcoin lightning)
No mention that most of that energy comes from renewables or how being a "buyer of last resort" for energy actually helps build out renewable grids since grid operators can guarantee whatever energy capacity they provision will be bought. Doesn't even look at energy mix and demand curves.
Just ragebait tailored to their readers who already have strong negative opinions about this asset class but not about bonds or stocks or other asset classes for some reason. Even though Bitcoin has kept all its promises for 15 years in a row, never been hacked, never experienced an hour of downtime, or bank holiday, and never had its value printed away by an ever increasing supply (supply is capped at 21 million coins).
And more importantly, take investor's money. Most tech company are constantly losing money and depend a continuous flow of investor money.
Reddit lost 90 million dollars last year, that's why they did the IPO, so they could sell the problem to some sucker. Snapchat lost 1.3 billion last year, and would run quickly go bankrupt if people stopped investing.
As usual, the best way to get rich quick is by selling a get-rich-quick scheme.
It has its usages. Unfortunately, the market pressures are going to require companies to use them in a way to make investors happy. The average Joe is probably going to have a bad time.