22,069 karma · joined January 10, 2013
NVIDIA, like everyone else on a bleeding edge node, has hardware defects. The chance goes up massively with large chips like modern GPUs. So you try to produce B200 cores but some compute units are faulty. You fuse them off and now the chip is a GP102 gaming GPU.
The gaming market allows NVIDIA to still sell partially defective chips. There’s no reason to stop doing that. It would only reduce revenue without reducing costs.
You’re correct that few EU companies get as large as US monopolies, but that’s kind of the goal when you want a functioning market.
The EU is just itching for any opportunity to get rid of US tech firms because they’re increasingly seen as sovereignty risks. And while the GDPR fines (that this likely refers to) appear huge on absolute terms, they are still low enough that US firms voluntarily decide to violate those laws and just pay the fines.
The US sees TikTok as a risk. For the EU, it’s Microsoft Office.
If you train a system to memorize A-B pairs and then you normally use it to find B when given A, then it's not surprising that finding A when given B also works, because you trained it in an almost symmetrical fashion on A-B pairs, which are, obviously, also B-A pairs.
"The webpage credits another author: Native binary debugging for OCaml (written by Claude!) @joelreymont, could you please explain where you obtained the code in this PR?"
That pretty much sums up the experience of coding with LLMs. They are really damn awesome at regurgitating someone else's source code. And they have memorized all of GitHub. But just like how you can get sued for using Mickey Mouse in your advertisements (yes, even if AI drew it), you can get sued for stealing someone else's source code (yes, even if AI wrote it).
If you want to predict future text, you use an LLM. If you want to predict future frames in a video, you go with Diffusion. But what both of them lack is object permanence. If a car isn't visible in the input frame, it won't be visible in the output. But in the real world, there are A LOT of things that are invisible (image) or not mentioned but only implied (text) that still strongly affect the future. Every kid knows that when you roll a marble behind your hand, it'll come out on the other side. But LLMs and Diffusion models routinely fail to predict that, as for them the object disappears when it stops being visible.
Based on what I heard from others, world models are considered the missing ingredient for useful robots and self-driving cars. If that's halfway accurate, it would make sense to pour A LOT of money into world models, because they will unlock high-value products.
What people like me wanted was an iPhone 13 mini that's a bit thicker so it can have a bit more battery capacity. And with the 120 Hz PWM nausea fixed.
The iPhone Air has worse battery life. And it has a larger screen. And it's worse to handle one-handed. Coming from the 13 mini, it's not an improvement.
https://www.ft.com/content/62df8d8d-31f2-445e-bfa2-c171ac43d...
In short: Apple can get away with ignoring what those customers want.
Ooops ?
https://www.macrumors.com/2025/06/03/iphone-16e-sales-lag-be...
Looks like the market did like the SE size.
My prediction is that the age of AI and LLM assistance will make tiny devices the norm. Like those AI pins. Like Siri inside AirPods. Like Meta's AR glasses. But it seems that Apple is losing the race here. They lost their edge when it comes to developing new user interface paradigms.
EDIT: [1] Bloomberg claims 10-15% return rate, which would be massive: https://www.bloomberg.com/news/newsletters/2024-02-18/apple-... (for comparison, Galaxus reports 2% as normal for Smartphones and <5% for Meta's Quest)
iPhone 3GS
Galaxy S3
Sony XZ1 Compact
iPhone SE 2016
iPhone SE 2020
iPhone SE 2022
Unihertz Atom
In my opinion, the fact that Apple is now selling a bag to carry your oversized phone around in, is an admission that they failed to make phones that are convenient to carry.
Fully offline LLM coding agent plug-in for JetBrains IDEs. Use with Qwen3-Coder-30B-A3B-Instruct-UD-Q4_K_XL.gguf on an RTX 3090 or up. The source code is pretty short to make sure it can be audited, if needed.
I work on A LOT of highly sensitive code, usually in manufacturing, automation, or robotics. It's basically impossible for a small company like mine to get a custom exemption from the company-wide NDAs that my clients usually require everyone to sign. These guys wouldn't touch a US cloud, so they surely would never approve US-based LLM agents.
But I also don't want to completely forfeit the convenience of using LLMs for boilerplate or simple refactorings. That's why this plugin is designed to also work in air-gapped environments. It only connects to http://127.0.0.1:8081 and the source code is few enough that you can audit this.
Also, in case you missed that: StackIt is the AWS / G Cloud competitor by LIDL: https://www.stackit.de/en/ It's the basebone for their app strategy with 100 mio+ client installs and about 500k employees.