Isn’t the domestic semiconductor company ruined by the media?

• corporate confidence

Many people are interested in the competition between TSMC and S Electronics regarding the foundry microprocess.

Recently, the density data of transistors according to micro-processes have been released. Although it is not confirmed and estimated

Foundry, system semiconductor, and fabless In common, the three elements of semiconductor are power consumption, operating speed, and transistor density.

Three elements of semiconductor.

1. Density: How many transistors go up in the same area?
2. Speed: How many gigabytes does the transistor also work?
3. Power: How many watts of power is the key to the operation.

This standard is different for each foundry company, so it is difficult to objectively compare it, but the data was recently released.

TSMC and S Electronics’ semiconductors are presentation data on how many transistors are accumulated in 1mm by 1mm.

There are 2.9 million TSMCs and 1.7 million Korean companies. This estimate was not released, but it was recently updated.

In other words, the density of domestic foundry 3-nanometer transistors appears to be less than that of TSMC 5-nanometer. TSMC has recently mass-produced 3-nanometer products and succeeded in developing 2-nanometer. Taiwanese companies have a density of 4.9 million units.

Such a thing is not published in the domestic media. They just think that domestic semiconductor technology is the best in the world, and a Taiwanese company reports that it will soon be overtaken by S Electronics and fail. How good would it be if the company develops with brain media?

In fact, the opposite is true. Korea’s semiconductor foundry technology is not even a game in Taiwan. That’s why Apple, Nvidia, AMD, Qualcomm, Intel, Google, Tesla, and Meta all order from Taiwanese foundry. When 4.9 units are put in the same silicon die area, it can only fit 1.7 units in Korea. In this situation, does AMD, which values density for speed power consumption performance, prefer domestic foundry? Even if it is entrusted, there seems to be no hope other than low-cost entry-level type.

To put it simply, what this means is that if you use a T company foundry, you get 49 CPU cores, but if you use a domestic foundry, you get 17 CPU cores. The same design on the same chip…

Isn’t it important to spend hundreds of trillions on foundry construction in Yongin or Texas now, and the remaining buildings are a priority as the competitiveness of basic foundry miniaturization is falling behind?

How can we increase the density of such transistors’ micro-processes? After all, it’s researchers and people, and it’s the efforts of suppliers and customers to improve the process. Isn’t it not important to build a factory with rat tail subsidies in Yongin or Texas, but investment in internal software and people first?

Since executives are not good at management, they keep focusing on concrete that seems to be clearly numerical. Isn’t the issue of long-term investment in people and R&D expenses more important than concrete?

Love and thought

Fineprocess #nm #TSMC #FOUNDRY #AI #semiconductor #transistor #density #comparison #media fact distortion #I also want to avoid bad stories. Let’s do well