AI's Investment Implications


#21

I see why you’re not suitable for software engineering :rofl: You’ve to be insanely excited about whatever are the latest hype… I mean in-thing… may be, latest advances?

My brother-in-law would be very happy, he works in MU and own MU RSUs :slight_smile: You might be buying what he has sold.

I’m in China, EV and AI too through AAPLs, just not sure what is the %. One stock does all.


#22

Absolutely the wrong assumption to make.

Also, I’m very suitable for software engineering. My career in SE has allowed me to get rid of my day job before I turned 40 and concentrate on investing for a living.


#23

Just opened a position in ALB for my EV portfolio. Albemarle mines lithium that’s a critical material for batteries.

I wanted to buy some Intel for my AI portfolio but it just kept on rocketing up. Intel’s PE is less than 1/3 of Nvidia’s. It has been buying AI and self driving car companies left and right. It’s throwing the kitchen sink at the problem. I still have reservation about its execution but at a big discount compared to NVDA I am willing to put a small bet. Just not today though.


#24

Looking at Intel’s price on Friday reminded me of a day in May 2016. I want to buy a stock but it went up from $35 to $40 just before I could place the order… I said “screw it, I have buy it anyway”, although my funding is the same so I could only buy much less. That was of course the last day it traded at $40 and just broke $200 Friday :relieved:
If you believe in AI, just go ahead and fill your AI portfolio. May it be Intel or NVDA… who knows Intel may break $200 within 2 years…


#25

True. Usually I don’t nitpick on price, but 7% up is hard to swallow… :cry:

Besides NVDA what else you have eyes on for AI?


#26

I honestly don’t have much cash :moneybag: for stock. Don’t you aware that all AI companies (Google, Apple, Intel, micron, NVDA, AMD…) have in common? They all need to do hardware and by definition have to be in SV. So real estate in SV is the safest bet and I am sure you are the expert on this :wink:


#27

Buy TSLA cars?
Use AVGO chips :slight_smile:


#28

Hardware could have another run similar to when Sun, Dell, Intel, Cisco, and EMC were top performers.

Do most of these companies go to Taiwan Semi to make the AI chips? I’m not sure if they are materially different to manufacture.


#29

Eddie Tam, a Hong Kong hedge fund manager I follow, has these “4 Inequalities” on AI:

  1. Hardware more important than software
  2. GPU more important than CPU
  3. Memory more important than logic
  4. Manufacture more important than design

The full interview, in Mandarin, is here:


#30

Software engineering has always been the money machine. But now it is a reversal of fortune for hardware due to AI??? :thinking:

So no more Google, FB and start buying NVDA, INTEL, … ???


#31

Hardware is yan and software is yin. We need both.

SV got its start making radar and stuff for the military. That’s hardware. Then Shockley/Fairchild/Intel started the semiconductor industry, literally putting the silicon into silicon valley. That’s hardware again. For decades it was mostly hardware. Only after the PC came around we started having a thriving software sector.

How did Apple get started? Making computers. Hardware again.


#32

GOOG & FB are not software companies. They are advertisement companies. Sell MSFT, buy AAPL.

  1. Buy AAPL, sell MSFT.
  2. Buy NVDA, sell INTC.
  3. Buy MU, sell ?
  4. Buy AAPL, sell ?

#33

What it means to be in the AI era:


#34

How to invest in AI theme?
manch decided on semi-memory MU
I had invested in robotics-smart home IRBT, watching MU, NVDA and PANW.
Anyone has alternative stocks?

Praying at one corner for Citron or Andrew to bash MU, NVDA and PANW.


#35

Based on above,
NVDA doesn’t meet the last two points since it doesn’t make memory and is fabless.
MU doesn’t meet the 2nd point since it doesn’t make GPU.

By hardware, does he mean component (MU) or full assembly (IRBT) or any hardware product that incorporate smart features?
By manufacture, does he mean owning fab plants?

Disclosure: Didn’t read the article.


#36

Eddie Tam is just one of the guys who helped shape my thinking. Don’t take his words as gospel.

Here’s a good survey of the landscape, albeit slightly outdated:

http://www.moorinsightsstrategy.com/research-paper-a-machine-learning-application-landscape/

I am adding Xlinix to my AI profolio, and monitoring Intel and AMD.

Actually @ed-lo brought up a good point. Real estate may be a good angle to play this wave. South Bay RE plus Fremont will outpace SF and surrounding suburbs.


#37

Also Austin has a lot of hardware people. Most of AMD R&D is in Austin I think. Outside of Bay Area I think Austin has the most hardware expertise. Its local economy should benefit. Whether that leads to higher property value I don’t know. You need to look at the housing supply too.


#38

Thanks for the link. Disclosure: I did view the video :smile:
I strongly agree the first 2 points. You never have apps utilizing every clock cycle in parallel until AI came in.
Point 3 is still true but to a much less extend with the quantum leap of Reinforcement Learning demonstrated by AlphaGo Zero. There is no data to input, just an empty Go board with rules will do. It only learned by playing 3.9 millions games vs over 30 millions games as Alpha Go Master did.
For point 4, I believe ASIC is the way to go, so design and manufacture are both important.


#39

I too think Tam understates the importance of chip design. It may not be as difficult as getting 5nm to work but there is still a lot of work left to be done, especially on the inference side.


#40

Can’t be putting every cents in RE right? What is your current asset allocation of RE (exclude primary) vs stock equity (include index fund & IRA/401k)?

XLNX is a good choice. Was thinking about whether to do or not to do for awhile… problem is I want to reduce the number of investment… too many is hard to follow. May be I should since I have no China stock yet even though have wanted to like at least 10 years ago… haha.