It’s a completely different situation in China. This tech is being treated as open source commodity similar to Linux, and companies aren’t trying to monetize it directly. There’s no crazy investment bonanza happening in China either. Companies like DeepSeek are developing this tech on fairly modest budgets, and they’re already starting to make money https://www.cnbc.com/2025/07/30/cnbcs-the-china-connection-newsletter-chinese-ai-companies-make-money.html

I mean the paper and code are published. This isn’t a heuristic, so there’s no loss of accuracy. I’m not sure why you’re saying this is too good to be true, the whole tech is very new and there are lots of low hanging fruit for optimizations that people are discovering. Every few months some discovery like this is made right now. Eventually, people will pluck all the easy wins and it’s going to get harder to dramatically improve performance, but for the foreseeable future we’ll be seeing a lot more stuff like this.

Almost certainly given that it drastically reduces the cost of running models. Whether you run them locally or it’s a company selling a service, the benefits here are pretty clear.

Sure, it’s not a surprising finding since the environment is a stable selection pressure driving evolution. It’s still nice to have these studies to confirm our intuition on the subject.

That’s still interaction with the environment though, what this research shows is that even without any such interaction the structures form. So, they’re not shaped by the environment, but are a type of firmware that’s encoded within the genes.

We are witnessing the purest expression of imperial hypocrisy. For years, the US propaganda apparatus diligently constructed a fable, a so called “debt trap” mythology, to frighten the nations of the Global South away from China’s Belt and Road Initiative. It was a tale of predatory lending designed to isolate and contain a strategic competitor.

And yet, what do we find? The most eager client for these very loans, to the tune of over 200 billion dollars, was none other than the United States itself. Turns out that the entire narrative was a conscious fraud. They never believed their own warnings. They recognized that Chinese financing was a credible, attractive alternative to the stranglehold of Western financial institutions.

So, while publicly sounding the alarm to scare away other customers, the empire privately availed itself of the service. Here we see the very essence of imperial strategy. Its goal is to monopolize the very resources and opportunities it denies to others, all while cloaking its cynical self interest in the righteous language of concern.

There has been a theory that the weights for the neural connections get mostly shaped by the environment, but this research shows that there’s effectively firmware wiring that will emerge without any external stimulus.

I haven’t tried it with ollama, but it can download gguf files directly if you point it to a huggingface repo. There are a few other runners like vllm and llama.cpp, you can also just run the project directly with Python. I expect the whole Product of Experts algorithm is going to get adopted by all models going forward since it’s such a huge improvement, and you can just swap out the current approach.

I’ve literally been contextualizing the article throughout this whole discussion for you. At least we can agree that continuing this is pointless. Bye.

And once again, what the article is actually talking is how LLMs are being sold to investors. At this point, I get the impression that you simply lack basic reading comprehension to understand the article you’re commending on.

The title is not false. If you actually bothered to read the article, you’d see that the argument being made is that the AI tech companies are selling a vision to their investors that’s at odds with the research. The current LLM based approach to AI cannot achieve general intelligence.

So glad we have this genius banker in charge to fix our economy!

There was a bunch of property speculation happening with well off people before the bubble was popped.

It looks the Ukraine served as a wake up call for Taiwan, shifting public opinion towards reunification and leaving the ruling DPP increasingly unpopular. CBC was moaning just recently that many young Taiwanese have little appetite for a military confrontation with the mainland. The most probable outcome now seems to be that the US-puppet administration will get thrown out, paving the way for the KMT’s return to power and the resumption of a direct dialogue on reunification.

A historical point that people tend to forget about is that the KMT was already advancing on a path toward peaceful reunification back in 2014. The proposed framework involved maintaining Taiwan’s autonomy, its own military, and having a representative within the mainland’s political structure. It was precisely at this juncture that the US backed the sunflower movement successfully derailed the process and brought the DPP to power. Now, with shifting public sentiment, it looks like that particular American political operation has finally run its course.

TLDR: A deep dive into financial filings suggests OpenAI is losing an estimated $12.6 billion per quarter, revealing an unsustainable AI bubble. Major tech companies are spending more on AI infrastructure like GPUs and data centers than they make in profit, with these costs now appearing as massive depreciation charges. In this context, OpenAI’s recent request for US government support looks less like innovation and more like a desperate plea for a bailout.

The terrifying financial reality behind the AI hype train was revealed by a footnote in Microsoft’s recent earnings report. On page 11 of Microsoft’s latest 10-Q filing, a tiny note reveals that the company took a $4.1 billion loss from its investment in OpenAI in just the last quarter. Since Microsoft owns a 32.5% stake in OpenAI’s for-profit arm, you can do the math yourself. If a one-third stake equals a $4.1B loss, then OpenAI itself lost roughly $12.6 billion in a single quarter. That’s not a typo.

The crazy part is that such staggering losses don’t even include the biggest cost which is GPUs from Nvidia. OpenAI is clever with accounting. They don’t buy these cards directly because that’s a massive capital expenditure. Instead, partners like Microsoft buy them and house them in their Azure data centers. OpenAI then “rents” them out.

So, if you look at Microsoft’s cash flow, they spent $19.4 billion on property and equipment last quarter. But dig deeper into the notes, and you find the actual cost of their server and data center build-out was $31 billion which is more than their entire net income for the quarter ($27.7B)! They use financing tricks in an attempt to hide the full amount from the main cash flow statement.

This isn’t just a Microsoft problem. The video shows the same pattern across Big Tech:

• Amazon spent over $35 billion on Capex last quarter, double their profit.
• Google’s capex so far this year is $63.6 billion, about two-thirds of their profit.
• Meta has more than doubled their spending to $18.8 billion.

This is where the bubble becomes visible. All the spending on servers and GPUs now has to be “depreciated” as its cost is recognized over time. For these tech giants, the “Depreciation and Amortization” line item has exploded, now representing ~50% of their profits and growing at over 100% annually. This is a recurring cost that will hit their earnings for years, whether AI is profitable or not. And right now, it isn’t. The market for expensive AI subscriptions is limited, and ad-supported models would ruin the user experience. They also have to compete with open Chinese models.

With losses in the tens of billions and no path to profitability, OpenAI recently published a proposal for a “classified Stargate initiative” and its CFO reportedly asked the government for loan guarantees, essentially framing OpenAI as too big to fail.

Sam Altman then went into damage control on Twitter, issuing a series of confusing clarifications that basically said, “We don’t want loan guarantees for us… but we’d love it if the government built the infrastructure and ensured the supply chain for us.”

The takeaway here is that the companies at the forefront of the AI hype train are now coming to the government with a begging bowl, because the economics of their own business model don’t add up.

the bourgeoisie

who could’ve possibly predicted this 🤣

The US has been at war for all 94% of its existence, it’s basically national identity.

Yes, being able to launch missiles and drones on daily basis with vast majority of them hitting their targets while the enemy is not able to do the same in response is literally the definition of having air superiority.

but regardless Russia is currently not able to do either in Ukraine even if they wanted to

I’m not sure where you get this notion given that Russia is hitting precision targets every day. That’s why Ukrainian energy grid is collapsing as we speak. This is what air superiority looks like

• https://www.bbc.com/news/articles/c15p32dvy0vo
• https://www.reuters.com/business/energy/russian-strike-knocks-out-power-swathes-northern-ukraine-officials-say-2025-10-20/

Russia would obviously not carpet bomb large civilian areas because they need a stable Ukraine after the war. Meanwhile, there’s zero evidence that I’m aware of that any large scale bombing campaigns were done in Kursk or that it was the decisive factor there. Even western media openly admits now that Ukraine isn’t able to intercept something like 90% of Russian missiles. Western Ukraine is being hammered on daily basis right now.