Correct me if I'm wrong, but hasn't "nm" as a naming scheme been kind of misleading since 157nm immersion lithography failed?

Like, before then, nanometer was a measure of distance between transistors, and a smaller difference meant a faster calculation that also used less energy, and could be made cheaper because the transistors would require less silicon for the same calculation.

But as finfet started coming onto the scene, you could essentially raise the transistor in a third dimension, which adjusted the performance profile of that transistor, allowing you to gain "effective nanometer reduction", so things like TSMC 16nm and onward weren't really "nanometer" anymore, but an essentially abstract number that indicated roughly the performance compared to previous generations, which is also why intel 10, for instance, is roughly as dense, in terms of literal nanometers, as TSMC 6/7, but doesn't necessarily perform the same in all instances.

IMO Moore's Law, as originally described is dead (a doubling of transistors, and performance every 8 months), but the "Layman's Moore's Law", that "Computation will advance geometrically and we'll be able to acquire higher levels of performance for the same money", is still well alive.

There's plenty of interesting and technically challenging ways to improve performance, such as 3d stacking (IBM, AMD), disaggregation (AMD, Apple), heterogenous compute (Arm, Intel) and so on, without even going into the upcoming AI accelerators that will take advantage of improved multi-threading / parallel compute to shore up on our lack of raw single threaded improvements we've seen as of late, so as a tech enthusiast I'm absolutely hyped for upcoming products, but I don't quite think that it's quite right to say that Moore's Law is still alive as it was originally used.

We don't really can go much lower than that. Quantum tunnelling is a real issue at that size. However we still can go up with the layer count. However then we have the issues with cooling. So what is relatively easy with memory it's a big problem with computing.

There’s also quantum computing processors which just took a step neared to mass production with a high rate of efficiency

Moore's law has ended for the definition of "computer power per dollar": https://en.wikipedia.org/wiki/FLOPS#Cost_of_computing

I work at a large service provider. Before 2016ish, compute costs were largely irrelevant since each hardware generation would make any investment in software or hardware optimization in the previous generation moot - resource consumption and compute costs were both increasing exponentially.

This is no longer the case: now we have major investments in software and workload-specific hardware to keep costs linear while handling exponential resource consumption.

Sorry to be bearer of bad news but fossil fuels are finite and we aren't producing more. So they will run out if we keep using them. That's a weird use of the word myth at the very least.

On the other hand, Moore's law may or may not keep holding depending on technological advances

Moore’s law lol

Moore predicted exactly what’s happening so technically it’s still Moore’s law yeah

If you believe fossil fuels will never run out I got some news for you. There's nothing replacing them. It took the Earth millions of years to create what we have