Printing infinite money with a glorified crud app is far better business

It runs on semiconductors.

Until they figure out how to make semiconductors only available as a service most of the market value will be captured by software.

Sir, we call that cloud computing.

There are many hardware startups in silicon valley. Tons in VLSI as well. Just google for it.

You'll find most are doing various ip blocks on someone else's process or finding novel process flows on known equipment. Obviously most are fabless because it's just so much less capital intensive. My company has invested many millions into just the equipment in just my lab and I can only measure stuff someone else made. CAD software is very expensive as well.

As for startup specifically related to novel manufacturing techniques. I guess it depends on what a startup is. Is a newly founded company with $50M in funding a startup or just a new company? The bay area has some companies doing interesting things in silicon photonics (in-package optical io is a hot topic as well as quantum computing)

And if you're willing to expand the geographic area outside Silicon Valley there is quite a lot more.

Take a look at the machines required of a modern fab. Just turning the water on, ultra pure water, can easily cost millions. The actual litho machines are beasts the size of school busses and cost hundreds of millions to spec and install. Then there are dozens of little suppliers of the special reagents, each of which comes from its own billion-dollar supply chain. There are only a few places on earth where all of this is availible at reasonable cost, and silicon valley is thousands of miles away from all of them.

> silicon valley is thousands of miles away from all of them.

Intel does 10nm production at their fab complex in Chandler, AZ. About 600 miles away from Silicon Valley.

> There are only a few places on earth where all of this is availible at reasonable cost

Whats the issue in transporting them? The more expensive, the less it should matter?

Look at the number of successful hardware vs software startups in VC funding and then look at the ratio of hardware to software billion dollar startups the last ten years.

Anything with effectively zero incremental costs will seem a better investment in the short term

Cerebras is an example of a hardware startup in Los Altos that basically is taking a wafer and using that to do acceleration based tasks for AI. They are or were in stealth mode but made an announcement at hot chips [1] [2].

[1] https://hotchips.org/assets/program/tutorials/HC2020.Cerebra...

https://www.crunchbase.com/organization/cerebras-systems

I'm not sure about "new" companies, but the market leader for lithography (Taiwanese HMI now owned by Dutch ASML) has major R&D in San Jose.

And they don’t make real money, compared to the Apples, Facebooks and Googles of the world. https://en.wikipedia.org/wiki/ASML_Holding says they had about 10 billion in revenues, 2½ billion in net income for about 18 billion in assets, for about 14% return on investment.

That looks good, but they’re in a highly cyclical market; every recession, their sales more or less completely dry up, and those figures are for a good year.

That’s why I can understand why investors aren’t lining up to invest in that kind of company.

Because building the next Facebook or Google is realistic?

A lot of other equipment goes into making semiconductors beyond lithography. Applied Materials and Lam Research are two major manufacturers of semiconductor manufacturing equipment that are located in SV. There are many smaller ones also.

I think you need crazy money and a lot of expertise to get into this game. Intel and others would love to master this and absorb this part of the market. But so far no one has with any expertise is willing to get into the ring so...