Joachim
Joachim boosted

"New servers are less energy-intensive." Really? 🤔

This is said so often that it could almost be taken as an absolute truth.

By extension, it is often said that each new generation of hardware consumes less energy than the previous one.

Manufacturers communicate pretty often about this, including HPE (1), Nvidia (2), etc..

A 🧵

#energy #it #ai #efficiency #datacenters

It's odd to me that people talk about data centers in terms of megawatts, a measure of electric power. For one example among many, a Bloomberg article from March about Microsoft cancelling data center plans began:

Microsoft Corp. has walked away from new data center projects in the US and Europe that would have amounted to a capacity of about 2 gigawatts of electricity, according to TD Cowen analysts, who attributed the pullback to an oversupply of the clusters of computers that power artificial intelligence.
"Data center projects that...amounted to a capacity of about 2 gigawatts of electricity" is a nonsensical statement. The (technical) capacities of a data center have to do with storage, compute, transmission, and latency. I understand there's probably some Fermi calculation along the lines of converting electric power to compute capacity using the TDP of NVIDIA's latest GPUs or something like that. Nevertheless, it's misleading to speak this way, not to mention lazy. It is oversimplifying in a bad way, treating data centers as utilities that supply a commodity when that is just not the case (at least not with AI, where prices for services still fluctuate fairly wildly).

#dev #tech #AI #energy #power #DataCenters

bhaugen
bhaugen boosted

Canada needs to reduce its dependence on US tech companies — and cloud infrastructure is a perfect place to start.

For the Canadian Centre for Policy Alternatives, I explain the threat posed by our dependence on US-owned data centres and why we need a public cloud that isn’t shaped by the pressure for shareholder profits.

https://www.policyalternatives.ca/news-research/canada-should-build-public-cloud-infrastructure-rather-than-relying-on-u-s-tech-giants/

#tech #cloud #datacenters #cdnpoli#cdntech #canada #markcarney

As data centers rapidly expand to meet surging AI-driven demand, their soaring electricity consumption largely due to cooling need is straining power systems and prompting tech giants to explore nuclear power solutions. https://www.japantimes.co.jp/commentary/2025/07/01/world/data-centers-southeast-asias-nuclear-dynamic/?utm_medium=Social&utm_source=mastodon #commentary #worldnews #datacenters #southeastasia #nuclearenergy #ai #artificialintelligence #bigtech

Google's emissions are up over 50%, Amazon builds huge data centers powered by 75% natural gas.

Remember all those posts telling us that "AIs climate impact isn't that bad" supported by some really funky math/perspective and/or numbers Sam Altman invented?

Here's the actual impact.

"AI" is a fossil fuel technology.

https://www.nytimes.com/2025/06/24/technology/amazon-ai-data-centers.html?unlocked_article_code=1.SU8.2JRa.e3Ju6r_pL1Im

https://www.theguardian.com/technology/2025/jun/27/google-emissions-ai-electricity-demand-derail-efforts-green

@tante @Nonya_Bidniss
The internet was construction to be resilient to failure of any node, with many pathways. How is building a (small number) of mega-data centers contributing to resilience. Do these not represent a small number of single points of failure?
#resilience#AI #datacenters #risk
#ClimateChange and #energy

We did the math on #AI’s energy footprint. Here’s the story you haven’t heard.
The #emissions from individual AI text, image, and video queries seem small—until you add up what the industry isn’t tracking and consider where it’s heading next.
The latest reports show that 4.4% of all the energy in the #US now goes toward #datacenters. The #carbon intensity of #electricity used by data centers was 48% higher than the US average.
https://www.technologyreview.com/2025/05/20/1116327/ai-energy-usage-climate-footprint-big-tech/