All the DRAM was promised to him 3000 years ago.

‘Give up your home computers so we can use the parts to build the surveillance state to enslave you.’

Shit, I didn’t even realize how bad the situation was until just yesterday when I thought about picking up a replacement SSD. Prices are insane!
 
Whenever I talk to Canadian normies, they have no clue any of this stuff is happening. They usually say something like: "if the leaders of the AI companies are saying something, we should listen". 0 critical thinking on their part. The Canadian way.

Since the start of 2023 about $1T has been spent already (not pledged but spent) on AI development by the major American AI companies. Investors are using a WACC of ~28% for pure AI companies so $1T in spending would imply a quick rule-of-thumb of 280B (1T*0.28) of perpetual cash flows coming in to justify the investment.

The combined annual income attributed directly to AI software and services across these companies is currently negative:

These jews realize they've really gone and fucked things up. No one wants to be left holding the bag so they're all scrambling.
 
Last edited:
It has to be said though that middle-aged liberal cat ladies, who are usually the standard bearers for all NPC opinions, are strongly opposed to AI art and especially to data centres
 
@ChevChelios There are two different propaganda campaigns going on at the same time. One is the Palantir and Open AI techbro-sphere insisting that we need to give up our personal computers and personal freedom on the altar of advancing AI, while also lobbying for market capture by demanding the government ban their competition. The second is the Chinese money driving all of the anti-data centre rhetoric because they're trying to convince the West to retard its own AI development to give the CCP an edge, and they're using exactly the same environmentalist rhetoric that they used to convince us to de-industrialize our countries. The anti data-centre leftists are proof that the second narrative is actually winning out, at least in terms of public opinion.

That being said, just because the CCP doesn't want us building data centres, that doesn't mean we should be building data centres.
 
I am retarded when it comes to computers and such but my good faith question is what’s stopping the nerd industry from just making more memory chips? I was always told that home computers were obsolete by the time you got them out to your car, so what’s the problem?
 
The machines to make chips are basically magic. They cost hundreds of millions of dollars, take years to make, only made by 1 company (ASML), and are probably on order out for years already. The high prices from high demand is actually somewhat disincentivizes companies from increasing production.
 
It's because the facilities that make them take years and the cost is absolutely absurd up front. So you can spend two of the years making a factory to make this stuff and then if a boom ends you're left holding the bag.

Also, being as AI companies are not spending their own money. It's just capital money. They're spending other people's money. The memory companies know they can take a $10 chip and sell it to them for hundreds of dollars and easily get paid for it.

The last I read pretty much the next two years were of wafers that you're going to use to make memory is already pre-booked by AI companies.


If you're talking about building a semiconductor factory that manufactures computer memory (RAM or NAND flash storage), you're looking at approximately 3–5 years from start to finish for a new, large-scale manufacturing facility.

And that's assuming the company has the money, equipment, permits, and technology lined up.

Here's how the process breaks down.

| Stage | Estimated time |
| ---------------------------------------------- | -------------- |
| Land, permits and engineering | 6–18 months |
| Factory and cleanroom construction | 18–30 months |
| Installing specialized manufacturing equipment | 9–18 months |
| Testing and initial production | 3–6 months |
| Scaling up to full production | 6–18 months |

Some stages overlap, so these aren't simply added together. csis.org +2

### A real-world example

Micron is building a major DRAM memory manufacturing facility in New York. Its first factory is expected to begin producing wafers around 2030, following its 2026 groundbreaking. investors.micron.com +1

### Why can't they build them faster?

* They require extremely specialized equipment.

* The buildings need exceptionally clean environments.

* Manufacturing requires enormous amounts of electricity and ultra-pure water.

* Producing memory chips consistently with acceptable yields takes considerable testing.

* There are only a handful of companies with the expertise to manufacture advanced memory.

### My bottom line

If the goal is to address today's shortage of RAM and computer memory, building a factory today isn't going to solve the problem next year.

Realistically, it could take 3–5 years before a new factory makes a meaningful contribution to memory supply.

One question: Are you thinking about building more factories in the United States to address the current RAM shortage, or wondering why existing manufacturers can't simply expand production faster?
 
You must log in or register to reply here.
Back
Top