The insistence/assumption that llm models will consistently get better, smaller, and cheaper is so annoying. These things fundamentally require lots of data and lots of processing power. Moore's Law is dead; devices aren't getting exponentially faster anymore. RAM and SSDs are getting more expensive (thanks to this insane bubble).
> RAM and SSDs are getting more expensive (thanks to this insane bubble).
That's not a matter of Moore's Law failing, but short-term capacity constraints being hit. It's actually what you want if Moore's Law is to keep going. It's a blessing in disguise for the industry as a whole.
Computing power still has practically flatlined. Memory density is decelerating in its improvement. My point still stands despite the temporary pricing situation.
Single-threaded compute, maybe - but that's increasingly a niche. Highly parallel workloads are still going strong in the latest device nodes, and power use for any given workload is decreasing significantly.
NASA has been directed by Congress to use the remaining Space Shuttle RS25 engines on SLS. There aren't that many RS25's left, so Artemis requires that they make the most of each launch. Getting more RS25's produced is one of those "nobody's made them in a long time and it would be terribly expensive and time-consuming to do so" type of situations.
correction: there are 16 RS25's left, but production has begun on more for the Artemis V mission. However, production is slow so they can't just yeet SLS's into space and test rapidly.
reply