I want to understand the ins and outs of the semiconductor industry. What resources would you recommend for beginner, intermediary and technical person?
Hard to answer this question well without a little bit of additional context. I'm very focused on the software side of things, so I am mostly familiar with resources that tackle how software interacts with various platforms and architectures.
I can recommend some resources in that vein (maybe they'll be useful if you/the reader have a software background):
The most frequently recommended book from a software perspective is "Computer Systems: A Programmer's Perspective" (http://csapp.cs.cmu.edu/3e/home.html).
These resources are mostly software-oriented, so they wouldn't give you much insight into modern chip fabrication, but if you're coming from a software background, may prove to be useful for building on what you already know.
Semiconductor manufacturing by now is heavily Asia centric. So it is natural for a YouTube channel about Semi conductor manufacturing to be Asia centric. That's where most of the world production is taking place!
I've seen the YouTube channel and it pays attention to ASML for example which is based in Europe. It doesn't ignore important players, even if they are non-Asian.
> but also dangerous because it's not comprehensive
Describing something as "dangerous" because it is not comprehensive is going too far. You can watch this channel and supplement your viewing with videos from other channels. Like any normal YouTube viewer does.
From a business/investment report, most of what's in books is going to be pretty much historical at this point. Doesn't mean not worth reading though--e.g. Andy Grove's Only the Paranoid Survive. Industry analysts (both financial and IT) can be good sources for current information--although a lot of that kind of thing is expensively gated. There's also just talking to people. After all, that's a good part of what analysts do--though they may have a greater level of access than you do.
I mean, you might say "I want to learn medicine" and people would take that many ways.
I studied EECE (electrical and computer engineering) in my BS, then Physics, and my Ph.D. is in Computer Engineering, minor in CS, but very specifically on VLSI and chip-making in the graduate school. I am saying this just to note: When you say "semiconductors industry" my immediate thought is you want to learn VLSI.
Do you want to learn the Architecture of CPUs? The bible is Hennessey and Patterson: https://www.elsevier.com/books/computer-architecture/henness... However, there are several versions (not editions) so be aware of RISC vs ARM and also one is HUGE and one is a subset of the main one (for example, this book: Computer Organization and Design, David A. Patterson, and John L. Hennessy is a subset of the "A quantitative approach" I linked to at elsiver).
I would first figure out or define if you want to learn semiconductors by themselves (the first two books I referenced, which will also teach you CMOS design and high speed digital circuits) or end to end architecture of CPUs (Hennessy and Patterson), or something else (e.g. there is also the manufacturing of chips, which Weste and Rabaey cover to some degree, but there are even more serious books on, for example Rabeays book on 3D ICs).
Much appreciated! I dont quite know yet which routes to take. I need to understand the ecosystem, manufacturing, different semicodnuctor uses, architecture, manufacturing.
Planning on spending the next two years getting up to speed so I can navigate the field from an investment standpoint
Ahh, well, Rabay and Weste have a short chapter on manufacturing. The Hennessy and Patterson book does as well, albeit shorter. Architecture of semiconductors is going to be Rabay and Weste. Architecture of CPU is going to be H&P book. The ecosystem of it all (packaging, die layout, etc, is going to be in more specific books).
Do not forget that although H&P will cover architecture, today architecture is done from the behavioral level down, e.g. they design what they want the chip to do, synthesize that in some Hardware Description Language (HDL, like Verilog or VHDL) and then they have a huge team of engineers who will optimize parts of that, as the synthesis is not totally optimized. Then on top of that, another team of engineers will optimize each component or module for speed by sizing the transistors appropriately.
I am going to admit, I know from a hand-waving level end to end manufacturing, so I could not point you at a very good resource for understanding the actual "boots on the ground" process of today's lithography. Again, in Rabaeys 3D IC book there are great sections on the processing of 3D ICs which would give you some insight into the process today.
Don't forget the historical aspects either! VLSI is hugely important, but how did we get here? Who were the Traitorous Eight and how did they get that name? The high level is available on Wikipedia but for the real story you need up dig up an actual history book.
What does the semiconductor landscape look like right now? Who is TSMC, who is ARM? Learning VLSI isn't going to teach you the ins and outs of the industry either.
Learn the Mandarin language. Taiwan is the epicenter of the cutting edge work in the industry. China is the powerhouse of production for it too. Both speak Mandarin natively and it would behoove you to understand it and their cultures.
If you had a decent Ph. D. from anywhere in the world in Semiconductor technology, I'm sure the fabs in Taiwan would still be interested in hiring you.
Even if you didn't know a single word of Mandarin. (Though in that case they might expect you to know English at Minimum)
Additionally, you should know Taiwan sources many parts of the production process from technologies in other countries.
For example ASML is the company that actually makes the EUV machines that etch the patterns onto the silicon, and that company is Dutch. In addition to that the lenses used in the EUV machines are custom made from Carl Zeiss, a German company.
Both of the aforementioned technologies are highly advanced and there is no way for TMSC to replicate each one.
Taiwan does not hold a monopoly on semiconductor technology. You should note that Korea (Samsung) also holds second place in this area and is not far behind.
China is actually pretty behind in the semiconductor race. Worse then intel as of right now.
It doesn't change the fact that none of this information on how TSMC, etc. actually produce at scale is written anywhere in english (or even in a book). These are intensely guarded trade secrets and you're going to need to start working with and around these companies to even begin to comprehend how what they do is possible.
TSMC is also opening a fab in the united states. Arizona. Look it up.
Don't dismiss the photo-lithography machines from ASML. The technology involved in the creation of those machines rivals TSMC itself. ASML gets less press but it is indeed as critical if not more critical in the supply chain than TSMC.
TSMC can be overtaken by samsung or intel... the possibility exists. No one, however is even close to being able to replicate the EUV machines created by ASML.
37 comments
[ 0.25 ms ] story [ 91.9 ms ] threadI haven't looked at the IRDS roadmaps, but they're free with some registration to the IEEE.
I can recommend some resources in that vein (maybe they'll be useful if you/the reader have a software background):
The most frequently recommended book from a software perspective is "Computer Systems: A Programmer's Perspective" (http://csapp.cs.cmu.edu/3e/home.html).
For a nice gentle intro, you can also read "The Elements of Computing Systems" (https://mitpress.mit.edu/books/elements-computing-systems) that accompanies a course called "Nand2Tetris".
George Hotz also compiled a list of topics to cover to learn the computing stack from bottom-to-top (https://github.com/geohot/fromthetransistor).
These resources are mostly software-oriented, so they wouldn't give you much insight into modern chip fabrication, but if you're coming from a software background, may prove to be useful for building on what you already know.
Hennessey and Patterson's computer architecture.
Good way to learn about TSM nonetheless
Semiconductor manufacturing by now is heavily Asia centric. So it is natural for a YouTube channel about Semi conductor manufacturing to be Asia centric. That's where most of the world production is taking place!
I've seen the YouTube channel and it pays attention to ASML for example which is based in Europe. It doesn't ignore important players, even if they are non-Asian.
> but also dangerous because it's not comprehensive
Describing something as "dangerous" because it is not comprehensive is going too far. You can watch this channel and supplement your viewing with videos from other channels. Like any normal YouTube viewer does.
Nothing is ever truly comprehensive.
oh and Jeri Ellsworth https://www.youtube.com/c/jeriellsworth
I mean, you might say "I want to learn medicine" and people would take that many ways.
I studied EECE (electrical and computer engineering) in my BS, then Physics, and my Ph.D. is in Computer Engineering, minor in CS, but very specifically on VLSI and chip-making in the graduate school. I am saying this just to note: When you say "semiconductors industry" my immediate thought is you want to learn VLSI.
So, do you want to learn VLSI? Half the VLSI population will direct you to Westes book: https://www.pearson.com/us/higher-education/program/Weste-CM... and the other half will direct you to Jan Rabaeys book: http://bwrcs.eecs.berkeley.edu/Classes/IcBook/ I read both. Rabaey is a great writer and I prefer his book quite a bit.
Do you want to learn the Architecture of CPUs? The bible is Hennessey and Patterson: https://www.elsevier.com/books/computer-architecture/henness... However, there are several versions (not editions) so be aware of RISC vs ARM and also one is HUGE and one is a subset of the main one (for example, this book: Computer Organization and Design, David A. Patterson, and John L. Hennessy is a subset of the "A quantitative approach" I linked to at elsiver).
I would first figure out or define if you want to learn semiconductors by themselves (the first two books I referenced, which will also teach you CMOS design and high speed digital circuits) or end to end architecture of CPUs (Hennessy and Patterson), or something else (e.g. there is also the manufacturing of chips, which Weste and Rabaey cover to some degree, but there are even more serious books on, for example Rabeays book on 3D ICs).
Planning on spending the next two years getting up to speed so I can navigate the field from an investment standpoint
Do not forget that although H&P will cover architecture, today architecture is done from the behavioral level down, e.g. they design what they want the chip to do, synthesize that in some Hardware Description Language (HDL, like Verilog or VHDL) and then they have a huge team of engineers who will optimize parts of that, as the synthesis is not totally optimized. Then on top of that, another team of engineers will optimize each component or module for speed by sizing the transistors appropriately.
I am going to admit, I know from a hand-waving level end to end manufacturing, so I could not point you at a very good resource for understanding the actual "boots on the ground" process of today's lithography. Again, in Rabaeys 3D IC book there are great sections on the processing of 3D ICs which would give you some insight into the process today.
Hope that helps some.
What does the semiconductor landscape look like right now? Who is TSMC, who is ARM? Learning VLSI isn't going to teach you the ins and outs of the industry either.
Even if you didn't know a single word of Mandarin. (Though in that case they might expect you to know English at Minimum)
Additionally, you should know Taiwan sources many parts of the production process from technologies in other countries.
For example ASML is the company that actually makes the EUV machines that etch the patterns onto the silicon, and that company is Dutch. In addition to that the lenses used in the EUV machines are custom made from Carl Zeiss, a German company.
Both of the aforementioned technologies are highly advanced and there is no way for TMSC to replicate each one.
Taiwan does not hold a monopoly on semiconductor technology. You should note that Korea (Samsung) also holds second place in this area and is not far behind.
China is actually pretty behind in the semiconductor race. Worse then intel as of right now.
Don't dismiss the photo-lithography machines from ASML. The technology involved in the creation of those machines rivals TSMC itself. ASML gets less press but it is indeed as critical if not more critical in the supply chain than TSMC.
TSMC can be overtaken by samsung or intel... the possibility exists. No one, however is even close to being able to replicate the EUV machines created by ASML.
"oh c'mon it ain't semiconductor engineering!"
https://www.amazon.com/Makers-Microchip-Documentary-Fairchil...
https://fs.blog/knowledge-project-podcast/the-ultimate-barga...
VLSI Circuit Design Methodology Demystified: A Conceptual Taxonomy:
https://www.wiley.com/en-us/VLSI+Circuit+Design+Methodology+...
2) Chip Design for Non-Designers: An Introduction by Juan-Antonio Carballo.
3) Demystifying Chipmaking by Ricahrd F. Yanda et al..
4) Fabless: The Transformation of the Semiconductor Industry by Daniel Nenni et al.