AI is killing the cheap smartphone. The memory that powers your phone now goes to data centres instead. - https://thenextweb.com/news/ai-killing-cheap-smartphone-dram-memory-crisis important social impace of #ai
AI is killing the cheap smartphone. The memory that powers your phone now goes to data centres instead.

In 1985, the best computer a reasonably affluent American could buy was the IBM PC AT, which cost $19,400 in today’s money. Today, a Tecno Spark Go costs $30 in a Nairobi market stall and runs a processor billions of times faster. No other good in history has experienced a cost decline on that scale. […]

The Next Web
AI screws us over again: memory shortages, Crucial shutting down

Micron, which makes about a quarter of all the computer memory and flash in the world, is shutting down Crucial, its retail store. Crucial is closing in February next year — the AI hyperscalers are…

Pivot to AI
@davidgerard ahead of the curve...
@glynmoody with the cost of RAM skyrocketing (CPU next?) maybe it's time to make the programs and operating systems that run on our PCs, laptops and smartphones more efficient and requiring less RAM.
Oh, I forgot, they'll all be vibe-coded in future and full of GenAI features so they'll all get less memory and processor efficient, so killing the batteries more quickly too.
@marjolica well, at least open source ones can do that

@glynmoody Yesterday at IT club someone had a computer with a dead SSD in. We looked up the price of a new Samsung EVO 250GB drive and they are going for around €200 each now. I replaced all the drives in my computers with nice Crucial ones for a quarter of the price a year or so ago...

RAM and storage has become stupidly expensive everywhere - even Samsung's phone division is complaining they can't get RAM...

@glynmoody there was no way of having an effective AI without building 'super-computing' centres, solidifying the perceived master-slave relationship which we already felt.

The user device(the slave) will be reduced to only what is necessary to interact with the AI centres(the master) or...

What is the distribution of those machines? Are there sufficiently many?

Can computing power from slave devices be amalgamated as a backstop?

@dckim well, local LLMs don't need to be huge if they are targeted. these are huge because they try to be generic