frontpage.
newsnewestaskshowjobs

Open Source @Github

fp.

Open in hackernews

AI Bubble Crashing Its Way into AI Abundance?

2•hmichaelson24•1h ago
As we move into a world of maximum engagement over substance and hype over reality, it becomes more important than ever to sift through the noise and find the truth. This is especially true in today's AI industry. If you read Twitter, it sometimes seems like everyone believes AI is going to take over the world and everybody's jobs next year, or that the whole thing is just a giant bubble.

At this point, most people have probably seen some version of the basic AI bubble theory: AI spending is a massive bubble, frontier labs are spending aggressively ahead of IPOs, and eventually the market will crash and everyone will return to reality.

I'm exploring a different AI bubble scenario centered on the massive GPU and data center buildout of the past few years. I think the closest analogy may actually be the 2022 crypto crash, rather than the dot-com bubble or 2008.

The basic idea is a deflationary spiral that starts as a financing and liquidity problem near the top of the stack and propagates downward through the physical compute market. If GPU-backed infrastructure becomes distressed, forced selling could push GPU prices and rental rates down sharply, which could then create more distress elsewhere in the stack.

The more interesting question to me is what happens after that. What do the second- and third-order effects look like if GPUs temporarily become absurdly cheap?

Has anyone considered this possibility or know of good writing on the topic? I'd especially love to hear from people who have thought about GPU financing, data center economics, or the secondary GPU market.

Comments

rvz•57m ago
Abundance for who and what of?

Mostly likely an abundance of security breaches, exploits and other terrible things.

Does that 'abundance' benefit humanity? We will get the negatives first before the positives unfortunately.

role-model is a protocol for assigning the right model for the right job

https://github.com/try-works/role-model
1•Bluestein•29s ago•0 comments

Developing provably correct Rust code with Verus

https://www.amazon.science/blog/developing-provably-correct-rust-code-with-verus
1•matt_d•1m ago•0 comments

Catholic tech leaders urge ethical boundaries for AI

https://www.ewtnnews.com/world/us/catholic-tech-leaders-urge-ethical-boundaries-for-ai
1•journey2s•1m ago•0 comments

Pihole and IPv6

https://unorde.red/pihole/
1•speckx•2m ago•0 comments

World of Warcraft: Forever

http://worldofwarcraft.blizzard.com/en-us/forever
1•minimaxir•2m ago•0 comments

Xiaomi Book Pro 14 2026, Intel X7 358H and B390 – Linux Driver Fixes

https://grigio.org/xiaomi-book-pro-14-2026-intel-x7-358h-b390-linux-driver-fixes/
1•grigio•3m ago•0 comments

Revolut handed over personal details to fake government request

https://www.cryptotimes.io/2026/09/12/revolut-handed-over-bitcoin-histories-passports-on-spoofed-...
1•defied•3m ago•0 comments

An open letter to Dario: if you mean it, open the weights

https://jacob.gold/posts/open-letter-to-dario-amodei-about-open-weights/
2•routelastresort•4m ago•0 comments

The truth and the lies about Islam in Europe

https://www.economist.com/leaders/2026/09/10/the-truth-and-the-lies-about-islam-in-europe
1•Gander5739•4m ago•1 comments

Recommendations for Teaching with AI (Fall 2026)

https://wiki.alcidesfonseca.com/blog/fall-2026-recommendations-for-ai-in-higher-education/
1•alcidesfonseca•5m ago•0 comments

Hardware x Model x Harness

https://rheisen.me/blog/taking-over-the-world/
1•rheisen_•7m ago•0 comments

Microsoft Slaps Hours Limits on Xbox Game Pass Cloud Gaming

https://www.techdirt.com/2026/09/11/microsoft-slaps-hours-limits-on-xbox-game-pass-cloud-gaming/
1•heisenbit•9m ago•0 comments

Prompt Golf – Write the shortest prompt to get a target output

https://promptgolf.app
1•qifzer•9m ago•0 comments

Statement On V/H/S SCP Foundation Film

https://scp-wiki.wikidot.com/statement-on-vhs-film
1•shscs911•11m ago•0 comments

Apple Neural Engine: A Complete Guide

https://ane-guide.readthedocs.io/en/latest/
1•t0ffif33•12m ago•0 comments

Show HN: Lab – Local-first, serverless browser client for LLMs

https://labstudio.tech
1•soidkeo•12m ago•0 comments

Anthropic is no longer a frontier lab

https://twitter.com/trydotworks/status/2098618997230985375
4•try-working•14m ago•0 comments

Google allegedly stole open-source code for mobile AI tool

https://www.neowin.net/news/google-steals-open-source-code-for-mobile-ai-tool/
4•OutOfHere•14m ago•0 comments

The OpenSSL Library supports DTLS 1.3

https://openssl-library.org/post/2026-09-09-dtls-1.3/
1•gslin•16m ago•0 comments

Is truth future proof? [pdf]

https://khoury.northeastern.edu/~cmartens/papers/plateau26-itfp.pdf
3•encyclopedism•17m ago•0 comments

The gpg.fail aftermath: On responsible disclosure, GPG, & the state of security [video]

https://media.ccc.de/v/2026-728-the-gpg-fail-aftermath-on-responsible-disclosure-gpg-and-the-stat...
1•_djo_•18m ago•0 comments

Linux Zoom client proactively reading everything written to X11 clipboard

https://hachyderm.io/@simontatham/117201594980991062
6•encyclopedism•20m ago•1 comments

Show HN: Termique, an SSH client where the server can't read your credentials

https://termique.app/
1•masitings•20m ago•0 comments

Park ranger found trees missing from the living world for millions of years

https://spacedaily.com/t-park-ranger-australian-canyon-wollemi-pine-discovery/
1•bryanrasmussen•21m ago•1 comments

The prestige and status of research fields within mathematics(2020)[pdf]

https://arxiv.org/abs/2008.13244
2•num42•23m ago•0 comments

4,400-Year-Old Tomb of Egyptian Judge Found at Saqqara with Colors on Walls

https://arkeonews.net/4400-year-old-tomb-of-an-egyptian-judge-found-at-saqqara-with-colors-still-...
3•arunbahl•27m ago•0 comments

Exxon's Communication Strategy to Neutralize Public Opposition (2022)

https://ojs.stanford.edu/ojs/index.php/intersect/article/view/2402
3•joebig•30m ago•0 comments

NHTSA Opens Investigation into Tesla Cybercab Self-Certification

https://www.nhtsa.gov/press-releases/investigation-tesla-cybercab-self-certification
3•Betelbuddy•30m ago•0 comments

Ten Mathematics Problems I Will Never Solve(1997)[pdf]

http://www.giancarlorota.org/problems/problems.pdf
2•num42•31m ago•0 comments

Musk and Altman back Anthropic CEO on slowing AI down

https://www.yahoo.com/news/us/article/dario-is-right-musk-and-altman-back-anthropic-ceo-on-slowin...
3•andsoitis•34m ago•5 comments