I’m convinced that OpenAI buisness model is to create AI bubble, to sell its “reserved memory wafers” and now Macs, at the top of ram extortion pricing moment.

small footprint lethality software tool for embedded applications

the embedded part means at the robot holding a gun level. Instead of which girls school to select for bombing.

I know huggingface. I don’t know of anything they sell, though.

The only reason I’m not upset with this, is that I will game the algorithm to make my price almost free, if these MFers want any of my $, that is. Maybe a good use of 20c of AI to harrass their algorithm into giving me a nearly free seat.

I agree the US spies. It is easy on a full general compute stack, ie. PCs/phones. Routers is harder, but NSA experts could probably find suspicious chips on hardware, and if those experts were part of a government for Americans, they could find violations installed at NSA’s requests too.

Make it look like a dram module, make the other 7 dram modules look like ones that are 7/8ths the standard size, and make the set of 8 pins work with 7 actual dram modules while still performing like a cheap standard module. so people buy it. There are no microsopic CPUs or with no pin traces leading to them.

when was the last time common people and business paid attention to IT Security experts BEFORE their systems were broken into?!

You/America can be worried about spyware from China. So the US makes an inspection department, like it has for food and drugs, that certifies there is no spyware. That is a completely different approach than fabricating spyware hysteria, and then forcing everyone to believe it.

There are experts, even stupid fat american ones, who could confirm the existence of such spyware capabilities.

Motherboards/routers (these are motherboards too) require very careful inspection, specifically if bios chip is unusual, or extra chip connected to bios. There is concern trolling over routers because Huawei is best in world. There is no concern about Motherboards, because only Taiwan is the competitor to China. The point is that inspections instead of baseless hysteria over origin is the right policy.

To make a stick of RAM spyware, would cost $100k/stick or more, and the engineering to do so, could make HBM6 instead. By all means investigate the chips to see if magic chips are inside, but that is the approach we should take. Why aren’t Chinese sneakers spying on you? Far easier to hide and do more cheaply.

The hysteria of foreign spyware is shaped by established monopolies who want to keep their monopolies. There is never a proposed solution of actually verifying/inspecting for spyware.

That only applies to spaceX saps who bought after IPO fraud. The big owners of the 2 LLM companies are Google and Amazon. They can break even on their datacenters because just their stakes will probably reach $1T before saps get to hold the bag.

Failure can just mean that they need to slow down from the absurd expansion commitments. Saying there must be a bailout is saying that there needs to be 3x more datacenters by tuesday, even if fraud is behind that push. “We need to maintain the fraud for longer” means we need to make the bubble even larger so that bailing out the fraudsters when it pops seems bigger economic impact.

If it sounds dumb, it’s because it’s meant to. “Buy them if they fail” debate leads to more onerous alternative proposals. The US is on a mission to make Skynet for Israel or China wins. It’s hard for them to fail when Skynet will pay high prices for their datacenter time. But if they fail, it will be because they are too expensive, and “American people” owning them doesn’t pay American people anything if they are stuck losing money forever. And the other alternatives that stem from debate, is “why not just throw government money at bubble top rates?”

Surprising fact about renewable energy: It provides electrical energy! That means you don’t need as much other sources of electricity compared to not having any renewable energy. Now you know.

AI: help me get straight A+s in school wink wink

There is white hat potential for this “creativity”. Find more/“all” vulnerabilities and get them patched. US empire allegiance will promote other paths, though.

Huawei is far ahead of Nvidia in AI clusters despite sanctions made against it, and 5 year behind tech stack. 50% cost/performance advantage over GB300 nvl72, (including building costs) by building outdoor siteable pods with 3 week lead times. CXMT is now most valuable company in China, but because of Huawei sanctions, they sell 64gb ddr5 modules at higher price than Samsung just to Huawei. Huawei needs its own RAM fab because it is harassed by west, and Chinese suppliers price to them with “captured customer” awareness.

We/all need AI for medical breakthrough potential… though.

Open models are still most cheaply hosted on a cloud, with batching and 24/7 use. API rates from developer lab are generally fair. Self hosting does have some significant tangible benefits though: Fine tuning for domain specific to organization, and not letting LLM provider train from your prompts/answers, followed by competing with your organization in the future as a result of “distilling your IP”.

It is burying old news in headline. The consequences/pitfalls of their “strategic shift” are fairly new. The layoffs were explicitly justified for pivot to datacenters that OpenAI will “surely” be able to rent. The new extra problems in that strategic shift just makes them look worse for going all in on the bubble.

There is zero proof of distillation. Minimax 2.7 development was surrounded by moderate use of Claude. M3 is their latest generation, and pretty solid, but its performance cannot be attributed solely (or even 5%) to distillation, and that is only lab that has been accused of significant API use. These claims are all 3-4 months old by now, and Anthropic blocked China access after publishing the accusations. Repeated BS is BS from losers trying to lobby for support.