undefined | Better HN

0 pointsjdietrich2y ago0 comments

Sure, but the semiconductor market is mature bordering on stagnant. There's a high-risk/high-reward competition for the leading edge that is currently dominated by TSMC, but there's just not much money to be made from the older nodes that the automotive, aviation and military segments care about. Building out capacity to secure your supply chains is potentially a very sensible strategic decision, but I'll eat my hat if it ever turns a real profit.

0 comments

mejutoco2y ago

After, if we assume from Apple sales, everybody got a mobile phone, AI could be what makes it non-stagnant again. Just companies ordering GPUs and servers, plus more and more similar features being shipped to consumer hardware could make a difference.

I do not have any data though, just thinking out loud.

FrustratedMonky2y ago

Yes. Think this is it. Even if some sectors appear stagnant now, everyone is expecting some huge growth spurt with AI. Lots of new data centers, GPUs, etc...

and of course, for national defense, to have homegrown supply chains.

junipertea2y ago

I would assume all those companies want cutting edge (higher performance at lower energy cost), not chips that are 5-10 years behind. Those can barely run current android.

throwaway20372y ago

> the semiconductor market is mature bordering on stagnant

Is this true? And by what measure? I see nothing but incredible growth driven by cloud, mobile, and AI/ML.

pas2y ago

barriers to entry are getting higher by the minute, there are no new entrants, most of them are state-sponsored, and it's true for the whole supply chain.

sure, there's technological progress and demand, but the market is in a near pathological state. (very short lifecycles, huge investments required even for fabs to keep up, etc. sure, it's unlikely that it will collapse, as demand is very strong, but market participants are very exposed to external factors. there is a huge discontinuity due to the dilemma of staying put and trying to beat the competition on price, or going all-in and building a new fab or at least upgrading one to be able to offer a newer/better process node. which leads to even more consolidation.)

FirmwareBurner2y ago

But that's only a small part of semiconductors localized in a couple of hubs. The vast majority of semi work isn't working on that.

paulmd2y ago

Do you mean the market is stagnant or the tech is stagnant, because those are different things imo.

If we have finally reached the final boss of silicon lithography, then the market will figure out whatever nodes make sense (best efficiency, best performance, best value) and presumably we will need ever increasing amounts of them… especially if future solutions just stack tons of dies together etc. it’s fundamentally a pivot from growing through the tech to growing through packaging and putting more silicon into a product… at higher power and cost. But as that market model matures you will very much continue to need more and more capacity to replace the growth that used to come from Moores law.

j / k navigate · click thread line to collapse