Hacker News Reader

Comment by echohack5 | Hacker News Reader

Isn't the current economic scenario unsolvable from the supply side in a 20 year time span? Like is ASML going to magically start mass manufacturing EUL machines, the universities magically gonna spit out 10K more PhD grade level people to work new foundries, and some country magically going to make all the infrastructure investments to build entire cities of compute manufacturing?

Maybe I am wrong but the supply side of this equation seems completely hosed from a fundamental perspective.

View on HN

That sort of thing did happen in the past though. Complicated engineering operations scaled up in months and years. Immediate post war era saw massive buildout of university engineering departments, more PhDs minted than ever before.

The fact we have so much latent demand and aren't getting these benefits right now is depressing. Where is our silico-industrial buildout? Why is our federal government cutting unversity research funding, and not expanding it to train more PhDs to run in demand fabs? We are being led by short thinking greedy assholes is why.

The reality is no one wants to be left holding the bag if the industry collapses. It costs $8-15B to build a fab for DRAM. If we rush to increase capacity, and the demand drops N-fold, everyone who rushed in will lose their shirt.

Also, fabs are easily the most complex manufacturing on the planet. You don't want to rush stuff like that.

That's it? That's cheap for stuff these days. Especially when your market cap is lets see 2 trillion for tsmc. 5 trillion for NVIDIA. 1 trillion for micron. 1 trillion for Elon Musk. Intel might have to get a loan, "only" worth 350 billion.

I was speaking only of the cost to build. There's the cost to run/maintain it (salaries, etc).

Keep in mind Intel is losing almost $2B per quarter on their foundry business. It's not easy to make the economics work out.

If you think the promise is there, then are you willing to invest all your money in a company chasing this? I'm not.

If I'm worth $10M, it doesn't mean I'll invest $200K in something I'm fairly confident will lose money. If they're not investing in building more capacity, it's because they're being appropriately cautious.

I’m actually not sure where the gap is, in the US, on foundries. It might be more like people with a bachelors and some experience.

We could also maybe use more semiconductor physics PhDs?

EUV, or EUVL, no?

Reply on news.ycombinator.com