The rise of $NVDA (-7,25 %) and other tech stocks, more and more investors have been drawn into the semiconductor industry in recent months and years. However, due to the enormous size and complexity of the industry, it is not easy to keep track of who is contributing to the global market now and how, and which stocks could outperform the market in the future.
1) What is the semiconductor industry doing?
The semiconductor industry basically manufactures microchips as an end product, which consist of millions and millions of tiny switches. These products have become an integral part of our lives and can be found everywhere. Examples include the data centers of $GOOGL (-2,74 %) smartphones from $AAPL (-6,8 %) , drives and inverters from $TSLA (-9,97 %) or $SIE (-7,45 %) or in the systems of $UKW (-3,14 %) . But also toothbrushes, ATM cards, passports, watches, lamps, and and and...
2) But the best microchips are made by $NVDA (-7,25 %) ?
Yes, in the area of graphics processors... If you look at the Semicondutor Top100, many well-known names appear: $AVGO (-5,25 %) and $QCOM (-7,67 %) for communication, $AMD (-8,01 %) and $INTC (-10,9 %) for processors for PCs, $TXN (-7,04 %) and $STM for discrete components, $IFX (-8,33 %) and $NXPI (-6,1 %) systems for the automotive industry and security solutions and many more...
3) Aren't the manufacturers taking market share away from each other?
No, because each company specializes in its own field and is the world market leader in one area. You can think of it like the automobile industry: Lamborghini makes sports cars, Landrover makes jeeps... Could Landrover make sports cars? Sure. Would they be good? If enough money and time is invested, certainly. Would that make sense? Probably less. And that is precisely where the value of semiconductor manufacturers lies. As the products themselves are highly complex, the know-how has to be built up over decades. The development is associated with extremely high costs and project durations from idea to mass production are rarely less than 3-4 years. That is why it would not make sense for e.g. $NVDA (-7,25 %) would not make sense to enter the power sector, as established players with a head start of many years have already established themselves there.
4) How are microchips built now?
Many steps are required to manufacture microchips. The engineers at the semiconductor manufacturers develop the circuits for the final products. As the structure sizes are in the nano-meter range, they have to rely on simulation environments such as $CDNS (-6,34 %) or $SNPS (-6 %) simulation environments. Once the engineers have developed and checked everything (a non-existent first-time-right costs many millions!), the data is sent to the fabs. Depending on the manufacturer, the in-house ($005930 , $STM , $IFX (-8,33 %) ) or, as in the case of e.g. $NVDA (-7,25 %) external. In other words, Nvidia itself only develops the "schematics" for projects, the final product comes, for example, from $TSM (-6,15 %) or $GFS (-7,62 %) . These fabs take orders from many different manufacturers and produce the products millions and billions of times in close coordination with the manufacturers...
5) How do the fabs produce the microchips?
As complex as the microchips are, the production itself is just as complex. In order to be able to create structures of just a few nanometers, clean rooms are required that are many times cleaner than operating theatres. These cleanrooms are now home to machines from $ASML (-2,09 %) , $AMAT (-5,46 %) or $7751 (-5,6 %) . These machines now manufacture 24h on the basis of various semiconductor materials such as from $ATS (-6,56 %) or $WOLF (-2,48 %) the finished products. These are sent back to the manufacturers, checked (in a first phase manually by the engineers with products from e.g: $KEYS (-5,64 %) , later fully automatically with e.g. $TEK ) and then sold. Here again there are intermediaries such as $SMCI,
$SIE (-7,45 %) or $AAPL (-6,8 %) who buy finished microchips and process them for larger products. Interesting here: Apple, for example, develops many things itself, but buys know-how and products from other companies because the entire development would be far too complex. In other words, semiconductors from many different manufacturers can be found in Apple products or other products - so there is also a great deal of interdependence, depending on the core competence.
6) What will become NVIDIA's No. 1 pursuer?
Hard to say. As described, the entire semiconductor industry is very closely linked. Based on global trends, different companies will benefit. Be it the companies in the first row (manufacturers) as well as those in the second and third (fabs, suppliers, software, end customers,...) The interdependence results in enormous opportunities for many companies but also many risks - for example, if Taiwan is attacked $NVDA (-7,25 %) with production of $TSM (-6,15 %) (not just Nvidia but virtually the entire industry) and the market shifts completely.
Addons:
- No guarantee of completeness or correctness, the entire industry is very extensive.
- If you are interested in this sector: A very good explanation of the entire process can be found at https://www.youtube.com/watch?v=_Kj58yQ67KI . Here it also becomes clear once again how large the sales figures (and thus also the risks of the companies) must be for the R&D and production costs to pay off. Remember the project durations here - who knows what will be important in 5 years?
