Hacker News Reader

Comment by Animats | Hacker News Reader

It's a real problem. There's a worldwide shortage of high-end fab capacity. That's not an easy business to enter. TSMC is spending US$20 billion on their newest fab. They started in 2018, and they expect to get parts out in volume in 2021 or 2022.

If you want to duplicate TSMC, you'd need a huge amount of money. TSMC is valued at around US$500 billion. Anyone not already a major fab operator would probably fail. They'd have to go through several generations of fabs to catch up. Meanwhile, TSMC is moving forward. China is trying to catch up.

There's never been an industry like this before, where it costs half a trillion dollars to be a player. Not just to be #1, but to even be in the game.

View on HN

With all of Intel's problems right now, this current shortage is a good chance to halfway catch up since they at least have their own dated fabs.

The shortage is actually a direct outcome of Intel's problems. Historically Intel had by far the highest capacity for leading edge fabrication processes - just for their own CPU production. It didn't make sense for companies like TSMC to invest in additional fabs as long as Intel had the lead. AMD taking the lead and moving ~25% of x86 CPU production to TSMC is a huge shift.

I suspect this problem is only going to get worse.

Each node shrink has come at incredible costs and there does not look to be any path towards decreasing those costs.

Eventually, there's simply going to be no ability to shrink. After all, we have a lower limit of single atoms. They have a radius of around 0.06 -> 0.5 nm. That's not terribly far from the current state-of-the-art processes of 7nm. Further, I doubt we'll ever get to that level in the first place without some really clever engineering. The issue to solve is electromagnetic radiation will induce currents in surrounding conductive material.

I really don't know where the industry ends up going when we hit that brick wall. Different materials? More exotic CPU designs?

Who knows.

7nm for eternity. The wheel has never been improved upon.

Honestly, that will probably end up being what happens. Once node shrinks are exhausted we'll see efforts poured into improving yield and using better materials. It will be less about getting smaller and more about refining everything.

Once the node shrinking stops, that's when you are likely to see R&D funds go away from manufacturing processing and towards chip and architecture design. Because that's the only place you'll see improvements is a more well designed chip.

Wouldn't this barrier to entry apply to any industry with high capital costs?

Container shipping, aircraft manufacturing, maybe tesla.

Aircraft manufacturing, maybe. Only two companies in the world can make a good large jet airliner - Boeing and Airbus. It took Airbus 21 years of government support to get to the first good product. COMAC in China is now trying.

Container shipping has hundreds of players. Anyone who can afford a container ship can play.

There are lots of little car companies. China has a huge number of car makers.

Reply on news.ycombinator.com