Can Qualcomm's acquisition of Intel change the dilemma of US chip manufacturing?

09/24 2024 517

The saying "Real men have fabs" comes from Jerry Sanders, the co-founder of AMD. Although it is sexist, the core view is still important today. Now, we should say "Real countries have fabs". A fab is a factory that manufactures semiconductors. The construction of these factories requires billions of dollars and several years, and they are extremely complex to operate.

For many years, Intel has been a leader in semiconductor manufacturing. However, starting in 2018, due to a series of mistakes, Intel's leadership began to waver, while TSMC steadily rose to become the world's top chip manufacturer.

Why do countries need fabs?

Intel's decline is not only a corporate failure but also a significant strategic and geopolitical issue for the United States. Many well-known American "chip makers" actually do not manufacture chips but design them, which are then typically manufactured by TSMC. Apple and other large technology companies also rely on TSMC to manufacture their internally designed chips.

It is extremely difficult to manufacture these complex products without making any small mistakes. Without TSMC, it would be a disaster for the United States and Europe. We would not be able to quickly obtain new iPhones, NVIDIA would have to find other places to manufacture its graphics processors, and the development of artificial intelligence might stagnate.

This is why the saying "Real countries have fabs" is particularly important today. Chips are the driving force of the modern economy. If you have to obtain these components from overseas, you become vulnerable.

Intel's decline is concerning because it is the only company in the United States that knows how to mass-produce high-performance chips. Recently, Qualcomm has approached Intel regarding a potential deal, but this does not address the US chip manufacturing problem. Qualcomm may not be interested in Intel's manufacturing business but rather its chip design business.

Intel's design and manufacturing businesses were once tightly integrated, which worked well for a long time. However, as the world shifted towards the model pioneered by TSMC – operating fabs exclusively and manufacturing chips for other companies – Intel's model began to seem outdated.

TSMC's success is partly due to the large and diverse manufacturing volume it obtains from various major customers, which has allowed it to learn how to manufacture chips more efficiently than anyone else. In contrast, Intel's manufacturing business has only one customer: itself.

Eliminating Intel's manufacturing problems will be an expensive, risky, and complex task. Intel has even started paying TSMC to manufacture some of its chips. The US company recently separated its foundry business from its chip design business, which helps external customers trust Intel to manufacture chips for them more easily.

The next challenge for Intel is to become genuinely proficient in chip manufacturing again. Until Intel's foundry business secures several major customers, it cannot truly challenge TSMC. To excel in chip manufacturing, you need a large, diverse production volume so that you can identify defects, improve processes, and feed that knowledge back into your fabs.

The US government is working to encourage other companies to use Intel's fabs. US Secretary of Commerce Gina Raimondo has been pushing shareholders of companies like NVIDIA and Apple to recognize the economic benefits of having an American fab capable of producing AI chips.

Technologically, Intel has a new process node called 18A, which could potentially make Intel more competitive with TSMC's leading nodes in the coming years. Intel's foundry business just needs more customers like this, and the 18A technology must be exceptional to achieve this.

Finally, how well will Intel's foundry business perform if it is spun off from the design side? This remains an open question, but the key is volume. Currently, Intel's foundry business needs to obtain manufacturing volume from the company's own chip designs. Without this, it will have almost no manufacturing experience to draw upon. Only when it attracts a significant amount of third-party business will the spin-off truly make sense.

Solemnly declare: the copyright of this article belongs to the original author. The reprinted article is only for the purpose of spreading more information. If the author's information is marked incorrectly, please contact us immediately to modify or delete it. Thank you.