13. New Variable of Quantum Processor Manufacturing

13.1 Quantum Processor Manufacturing

If we look at the long term (ten to twenty years), our analysis may have a major technological variable, that is, Moore's Law has come to an end, and chip nanotechnology that exceeds this law will become increasingly difficult, while quantum technology may replace existing classical processors. In addition, whether the production technology of classical processors (CPU, GPU, NPU) can partially assist in the production of quantum processors seems to be still under study. Whether TSMC's current advanced process is still valuable in the long run, Trump's attempt to force TSMC to move to the United States may also be in vain. What is your assessment?

ChatGPT said:

Your analysis touches on long-term trends in the semiconductor industry, especially the end of Moore's Law, the rise of quantum computing, and the future value of TSMC's advanced process technology. We can evaluate this issue from the following aspects:

13.1.1 Has Moore's Law really come to an end?

Moore's Law (i.e. the number of transistors on a chip doubles every 18-24 months) has reached its physical limit, as shown in the following:

Advanced process difficulty has increased significantly: The industry has now entered the 2nm and 1.4nm nodes, but the technical difficulty and cost of further reducing the size of transistors are increasing exponentially.

Quantum tunneling effect: When transistors are reduced to a width of a few atoms, electrons may not be effectively controlled, causing quantum effects to interfere with chip operations.

Advanced packaging technology becomes a new direction: In order to continue to improve computing performance, advanced packaging (Chiplet, 3D IC, CoWos) has replaced the strategy of simply shrinking process nodes, which also makes the leading position of process technology less decisive.

Conclusion

The end of Moore's Law means that "process scaling is no longer the only decisive factor, and future semiconductor innovation will shift to architecture design, packaging technology and new materials." This raises questions about the long-term value of TSMC's 2nm and below process technologies, as these technologies may not be able to maintain their past market advantages.

13.1.2 Will quantum computing replace existing chip technology?

Quantum computing has the potential to surpass the computing power of classical processors, but it still faces several key challenges:

Lack of technological maturity: Currently, companies such as IBM, Google, and Intel are still in the experimental stage, and quantum computers have not yet been able to replace traditional chips on a large scale.

Limited application scenarios: Quantum computing is suitable for solving specific problems (such as password cracking, drug simulation, and AI training), but it still cannot replace CPU/GPU/NPU for general computing (such as mobile phones and servers).

Different hardware technology: Current quantum processor technology (such as superconducting qubits, ion traps, and photon computing) is very different from traditional semiconductor processes, and it is not clear whether existing TSMC manufacturing technology can be directly applied to quantum chip production.

Conclusion

In the short term (5–10 years), quantum computing will not completely replace traditional processors. In the long term (within 10–20 years), if there is a breakthrough in quantum technology, it could have a significant impact on the traditional chip market. If TSMC wants to maintain its technological leadership in the future, it may need to invest more resources in developing quantum semiconductor-related technologies rather than focusing solely on traditional chip manufacturing.

13.1.3 Is TSMC’s 2nm Process Still Valuable in the Long Run?

If Moore's Law slows down and quantum computing is not yet mature, TSMC's advanced processes will still maintain a certain value in the next 10-15 years. The main reasons include:

AI, HPC (high-performance computing), and self-driving cars still require high-end chips: Currently, AI chips (such as NVIDIA GPU, Google TPU, Tesla Dojo) still rely on the most advanced processes. These fields will still be highly dependent on TSMC's process technology in the next 10 years.

The demand for 5G/6G and military chips is still increasing: The United States, Europe, and Japan all want to ensure independent supply of high-end chips, so they still need TSMC's technology to manufacture advanced communication chips and military chips.

Old processes still have a market: Even though process miniaturization after 2nm may become difficult, 3nm, 5nm, and 7nm still have long-term market demand, especially in the fields of automotive, Internet of Things (IoT), industrial AI, etc.

Conclusion

In the short term (10–15 years), TSMC's advanced process is still valuable, but in the long term (20 years later), if quantum computing technology breaks through, TSMC's advanced process may be weakened.

13.1.4 Is it possible that Trump’s forced move of TSMC to the United States will be in vain in the long run?

There are several variables as to whether TSMC's plan to build a factory in Arizona, USA, will be successful in the long run:

Possible conditions for success

✅ The United States establishes a complete semiconductor supply chain: If the United States can establish a complete supply chain of materials, equipment, and talent in its own country, TSMC's US factory may be successful.

✅ Long-term subsidies from the US government: If the United States can provide long-term subsidies (such as funding from the CHIPS Act), the Arizona wafer factory will have a chance to continue operating.

✅ The United States trains enough semiconductor talent: If the United States can successfully train engineers and technicians, then the U.S. wafer manufacturing may be able to gain a foothold.

Possible risk of failure

❌ US labor costs are too high, affecting competitiveness: If the production costs of US factories remain 50-100% higher than those of Taiwan for a long time, then these wafer fabs may find it difficult to operate in the long run.

❌ China absorbs TSMC talent and builds an independent supply chain: If China eventually succeeds in absorbing TSMC's technology and engineers, the technological advantage of TSMC's US factories may be weakened.

❌ Quantum computing breakthroughs reduce the value of traditional processes: If quantum chips replace traditional chips in 10-20 years, US semiconductor investment may become "outdated technology", making TSMC's US factories have no strategic value.

13.1.5 Conclusion

In the short term, TSMC's US plant will help the US reduce its dependence on Taiwan, but in the long run, if technological development changes (such as the rise of quantum computing) or the US is unable to reduce costs, then this investment may become a waste.

Final assessment: Whether TSMC's relocation to the US is successful depends on technological change and supply chain development

In the short term (5-10 years), TSMC's 2nm process is still valuable, and setting up factories in the US may still play a strategic role.

In the medium term (within 10–15 years), if demand for AI, self-driving cars, and 5G/6G continues and traditional chip manufacturing remains important, TSMC’s factories in the United States may still have value.

In the long term (within 15-20 years), if quantum computing technology achieves a breakthrough and Moore's Law no longer dominates the market, TSMC's advanced processes may lose their key position and the value of U.S. semiconductor investments may be seriously challenged.

Therefore, the success or failure of TSMC's relocation to the United States ultimately depends on whether semiconductor technology remains competitive and whether the United States can overcome the high cost problem.

13.2  Impact of classical manufacturing technology to assist Quantum Processor Manufacturing

The above analysis lacks the analysis of "whether the production technology of classical processors (CPU, GPU, NPU) can partially assist in the production of quantum processors?" If so, then Trump's plan to transplant TSMC is still valuable in the long run?

ChatGPT said:

You raised a key question: Can the production technology of traditional processors (CPU, GPU, NPU) partially assist in the production of quantum processors? If so, is Trump's plan for TSMC to set up a factory in the United States still valuable in the long run? The following is an analysis of this issue:

13.2.1 The relevance of traditional semiconductor manufacturing technology to quantum processor production

Similarities:

Precision manufacturing requirements: Both traditional semiconductor manufacturing and quantum processor production require precision processing at the nanoscale. Existing semiconductor manufacturing facilities have extensive experience in precision manufacturing, which may assist in some aspects of the production of quantum processors.

Differences:

Materials and structures: Traditional processors mainly use silicon-based materials, while quantum processors may use different physical implementation methods such as superconducting materials, ion traps or photons, which are significantly different from traditional semiconductors.

Manufacturing process: The manufacture of quantum processors may require ultra-low temperature environments or special vacuum conditions, which is different from traditional semiconductor manufacturing processes.

Conclusion:

Although the experience of traditional semiconductor manufacturing technology in precision manufacturing may be helpful to the production of quantum processors, due to the significant differences in materials, structures and processes between the two, traditional technology can only assist the production of quantum processors to a limited extent.

13.2.2 Long-term value assessment of TSMC's fab in the United States

Considerations:

Technology transition risk: If quantum computing makes breakthrough progress in the next 20 years, the demand for traditional semiconductor technology may decline, which will affect the long-term value of TSMC's traditional fab in the United States.

Technology Diversification Strategy: If TSMC and the U.S. government invest in emerging technologies such as quantum computing in the future and incorporate them into U.S. manufacturing facilities, the long-term strategic value of these facilities will be enhanced.

Conclusion:

The long-term value of TSMC's US plant depends on whether it can adapt to future technological development trends. If these facilities can be transformed or expanded to support the manufacturing needs of emerging technologies such as quantum computing, they will continue to have strategic value even if demand for traditional semiconductor technologies declines.

Overall Assessment

Traditional semiconductor manufacturing technology may be helpful in the production of quantum processors in some aspects, but due to the differences in materials and processes between the two, this auxiliary role is limited. Therefore, the long-term value of TSMC's factories in the United States will depend on its ability to adapt to and support future emerging technology needs.