No way.

[plwt@mstdn.social]

@rysiek Thank you for sharing an interesting article. I do not see it as the "beginning of the end" of the AI boom, but it does make me think of the wider impact.

The nature of the commitment that big tech has made to AI is such that the fallout from any bubble bursting could lead to senior names leaving various corporates.

On the flip side, and more broadly, it could also hasten digital sovereignty efforts as people look to mitigate future risks.

[cass_m@socialbc.ca]

@rysiek at least they didn't bury the lede

[mkoek@mastodon.nl]

@rysiek I point out this simple fact at work and the response is “it will get cheaper”. No supporting evidence, it’s just a belief.

[racingreen17@mastodon.social]

@rysiek mentioning "forecasts" as if it was serious economic analysis and not just marketing by the companies involved

[bms48@mastodon.social]

@pastawoody @racingreen17 @rysiek The employment of economic arguments in necropolitics helps reify its traditional reputation as: "the dismal science"

[bms48@mastodon.social]

@mkoek @rysiek ...when everything Ed Zitron is reporting on says exactly the opposite. There is a sociological and pseudo-religious aspect to the "AI" phenomenon which is an elephant in the room

[bradthornborrow@thecanadian.social]

@rysiek @jaythvv This is why I keep this one in my stock image folder…

[damien@layer8.space]

@mkoek @rysiek I'd love to know *how* (they think) it will get cheaper when it's currently sold at a loss!

[trebach@functional.cafe]

@mkoek @rysiek Given the backlash of data centers in a state with about 600 of them, that electricity rates are going up for them and other large consumers, and they may lose state tax breaks in the state budget due by July 1, there's evidence that it's going to get more expensive

[pier@infosec.exchange]

@damien @mkoek @rysiek From what I can tell everyone is basing this belief on a historical trend of computing getting cheaper over time that doesn’t really work for LLM inference, combined with some dude at Gartner who says inference is going to be 90% cheaper in 4 years. Based on what? ‍️. And, you know, Gartner totally doesn’t have a track record of being wrong like half the time

[alper@sfba.social]

@rysiek it's ok. Just wait. Remember how some people laughed when ginger was going to change the cities around the world completely? you just have to wait a little bit.

[jackmexa4@mastodon.social]

@rysiek

LOL

[beachbum@mastodon.sdf.org]

@rysiek The bubbly will burst. I’m not sad.

[chrisgerhard@mastodon.green]

@rysiek Even if you were an AI true believer, telling people their future depends on using tokens is mad. Using tokens is trivial, you can even do so in parallel so more tokens get burned. To use token usage as a metric of success is so dumb it's beyond a joke.

[bms48@mastodon.social]

@mkoek @rysiek Oh boy. more kryptonite. Thanks to @kimcrawley ref-refs; Iris van Rooij. https://arxiv.org/abs/2507.09089

[return0media@infosec.exchange]

@pastawoody That goes on my list of greatest things I've ever had the absolute pleasure of reading.

[mkoek@mastodon.nl]

@damien @rysiek just a generic belief that all technology becomes cheaper as it matures, nothing rational about it