[Gap] Samsung TSMC’s revenue gap has widened, and the AI chip market has become the key; President Rapidus： the production speed of 2 nanometers will be three times that of TSMC; Several chip projects

1. Samsung TSMC’s revenue gap is the key to expand the AI chip market;

2. President 2.Rapidus: The production speed of 2 nanometers will be three times that of TSMC, and cooperation has been discussed with 40-50 enterprises;

3. India seeks to strengthen its chip manufacturing capabilities, but many projects have suffered major setbacks;

4. Morgan Stanley: In 2050, the global output value of humanoid robots will reach 5 trillion US dollars, and the cost reduction to 15,000 US dollars is a key turning point.

1. Samsung TSMC’s revenue gap is the key to expanding the AI chip market.

Samsung Electronics recently released a financial report showing that its semiconductor division’s revenue in the first quarter of this year was 25.1 trillion won (about 51.733 billion yuan), down 17% from the previous quarter. At the same time, TSMC’s revenue in the first quarter increased by 42% year-on-year to 839.35 billion Taiwan dollars (about 38.82 trillion won). As a result, the quarterly revenue gap between the two companies has exceeded 10 trillion won.

It is reported that the decline in revenue of Samsung Electronic Equipment Solutions (DS) department is mainly affected by factors such as semiconductor export control, resulting in a decline in sales of high-bandwidth memory (HBM) for AI chips. As a latecomer to the HBM market, Samsung has not yet entered the supply chain of NVIDIA, which makes its HBM business unable to effectively boost its overall performance.

In contrast, TSMC’s outstanding performance is mainly due to the strong demand for AI and the accumulation of chip inventory due to tariff considerations. Wen Junhao, a researcher at Samsung Securities, pointed out that although TSMC’s smartphone business entered the off-season, the growth of high-performance computing (HPC) business successfully made up for this impact.

According to historical data, Samsung surpassed Intel to become the world’s top semiconductor revenue in 2021, but with the memory market weakening since the third quarter of 2022, TSMC began to overtake Samsung. Although Samsung briefly regained its leading position in the second quarter of last year, with the rapid growth of the AI semiconductor market, TSMC has taken the lead again since the third quarter, and the gap has been widening-from about 3 trillion won in the third quarter of last year to 8 trillion won in the fourth quarter, and then to more than 10 trillion won in the first quarter of this year.

The securities industry predicts that the revenue of Samsung’s DS division is expected to rebound to 28-30 trillion won in the second quarter of this year, mainly due to the bottom of the first quarter performance and the recovery of traditional chip demand before the implementation of the US tariff policy. However, TSMC has announced that its revenue forecast for the second quarter is US$ 28.4 billion to US$ 29.2 billion (a contract of 39 to 40 trillion won), which is still about 10 trillion won higher than Samsung’s expectation. Wei Zhejia, chairman of TSMC, said that this year’s annual revenue is expected to increase by 20% in US dollars, and AI-related demand will continue to be strong.

It is worth noting that although Samsung is a comprehensive semiconductor enterprise covering both memory and non-memory, TSMC only focuses on wafer foundry business, and there are differences in their business structure, they have a high degree of influence in the global semiconductor market and compete directly in the foundry field, so the status of "revenue first" is of great symbolic significance. Judging from the current trend, it is difficult to narrow the revenue gap between the two companies in the short term.

In the increasingly fierce competition in the global semiconductor market, the demand change in the AI chip market has become a key factor affecting the company’s revenue. In the future, the layout and performance of Samsung and TSMC in the AI field will further determine their position in the global semiconductor market.

2. President 2.Rapidus: The production speed of 2 nm will be three times that of TSMC, and cooperation has been discussed with 40-50 enterprises.

In an interview with the media recently, President Koike Chunyi of Japan Rapidus said that Rapidus is negotiating with many companies including GAFAM (a giant technology company) and artificial intelligence chip design start-ups in the United States to expand the foundry business. However, obtaining orders from manufacturers in China still faces difficulties.

Rapidus’s trial production line in Chitose City, Hokkaido, has been partially put into operation on April 1st, and it is expected that all processes will be started within this month. At present, the company has signed a memorandum of cooperation with two start-ups such as US AI chip design company Tenstorrent.

In the field of advanced semiconductor production, Rapidus obtained the manufacturing technology of 2 nm products from IBM in the United States, and strived to achieve mass production in 2027. Although this time is two years later than TSMC’s planned 2025, Koike Chunyi emphasized that the speed from ordering to chip production and assembly can be increased to more than 2 to 3 times that of TSMC, forming a differentiated advantage.

Koike Chunyi is full of confidence in the company’s technical prospects, saying that "technicians who know that Japan was the first in the world have mastered the 2-nanometer technology. This is a difficult challenge, but everything is going well at present. " He also pointed out that the improvement speed of the trial products is very fast, and the yield will gradually increase.

At present, TSMC occupies a leading position in the most advanced semiconductor production, and exclusively OEM AI chips and other products from NVIDIA, USA.
Koike Chunyi believes that with the tension between China and the United States, the demand of American customers for second suppliers is increasing, which provides market opportunities for Rapidus.

Looking forward to the future, Koike Chunyi said that Rapidus will also actively deploy 1.4 nm semiconductor technology after 2 nm. He stressed: "If you can’t devote yourself to the next generation of technology in about two and a half to three years, you can’t win in the world." Once the 2 nm OEM business is on the right track, Rapidus will continue to advance the preparation of the next generation of semiconductors.
At present, Japanese manufacturers can only produce 40 nm products at most, while 2 nm technology adopts a completely different semiconductor structure. As a latecomer, Rapidus hopes to return to the forefront of global semiconductor manufacturing by mastering this advanced technology.

3. India seeks to strengthen its chip manufacturing capability, but many projects have suffered major setbacks.

In order to promote the development of the industry from OSAT operation to mature manufacturing capacity, India is currently focusing on the independent manufacturing of intermediate products in the semiconductor value chain, especially chip manufacturing.

It is reported that in the past five years, the compound annual growth rate of investment in India’s semiconductor industry was 10.5%, and it will reach 37.1 million US dollars by 2024. It is expected that this upward trend will continue, and the investment will increase to 109.6 million dollars by 2030. The factors driving this growth include the domestic manufacturing growth of automotive electronics, telecommunications infrastructure, industrial automation and consumer electronics products, especially the expansion of 5G and artificial intelligence-based technologies.

The data shows that in 2024, driven by the electric vehicle revolution, India’s largest demand for semiconductors came from the automotive industry (33%). Driven by the deployment of 5G networks, India’s second largest demand for semiconductors comes from the telecommunications equipment industry (24%). Followed by data centers (14%) and industrial automation (14%), these areas have increased due to the continuous advancement of digital transformation of enterprise operations.

In addition, in 2024, India’s exports of semiconductor components reached US$ 4.21 billion, with a compound annual growth rate of 7.4% in the decade from 2015 to 2024. The main products include integrated circuit chips, capacitors, printed circuit boards and photovoltaic cells, and more than 200 countries/regions imported products from India. Among them, the rapid growth of photovoltaic panels and telecommunications equipment exports reflects India’s rising position in the global market.

In recent years, some initiatives/plans launched by the Indian government have prompted many terminal application fields to establish local OSAT facilities. The analysis points out that India has formulated relevant policy frameworks under the "Made in India" plan and the "Indian Semiconductor Mission" initiative to support its semiconductor ambitions. With the entry of global chip manufacturers, the expansion of domestic production capacity and the growth of R&D investment, India is on the verge of a semiconductor breakthrough.

However, India’s ambition in the global semiconductor manufacturing field has recently suffered a major setback, and two heavyweight companies have successively adjusted or abandoned their original semiconductor manufacturing plans.

In early May, Zoho, a software supplier, officially announced that it would abandon its semiconductor manufacturing plan. The company is famous for providing Zoho One, a business operating system with tools such as CRM, HR and marketing. Zoho announced in May last year that it planned to invest 700 million US dollars to set up a semiconductor wafer factory. Recently, Sridar Vembu, former CEO and current chief scientist of Zoho, said on social media X that the company’s board of directors thought it was not ready to invest in the field of chip manufacturing, so it decided to give up the plan.

At the same time, Adani, a large Indian industrial group, has also suspended discussions with Israeli chip maker Tower Semiconductor on the plan of a $10 billion wafer factory. The cooperation of this project was announced in September last year, and it is planned to build a wafer factory in Panvel, Maharashtra. It is expected to produce 80,000 wafers per month after full production, mainly focusing on the analog and hybrid chip markets. . According to reports, Adani decided that the project was not commercially feasible, so she chose to quit.

According to informed sources, Adani Group previously said that the project was still under evaluation, but after internal evaluation, it was found that there was uncertainty about how much demand the business could generate, especially in the Indian market, so the negotiations with Gaota Semiconductor have been put on hold. Another source said that Adani Group was dissatisfied with the financial resources that Gaota Semiconductor was willing to invest, but did not disclose specific details.

In addition, it is worth mentioning that the $19.5 billion joint venture between Vedanta, an Indian enterprise group, and Foxconn, Taiwan, China, broke down in July 2023, because New Delhi was worried about the project cost and the delay in the approval of incentives. At present, the projects being developed in India include a chip manufacturing plant and another chip testing plant invested by Tata Group with an investment of US$ 11 billion, and a chip packaging department invested by Micron with an investment of US$ 2.7 billion.

4. Morgan Stanley: In 2050, the global output value of humanoid robots will reach 5 trillion US dollars, and the cost reduction to 15,000 US dollars is a key turning point.

Recently, Morgan Stanley (referred to as "Morgan Stanley") released the latest research report, saying that it is estimated that the global output value of humanoid robots will exceed 5 trillion US dollars in 2050, and the deployment scale will be 1 billion units.

According to this report, the humanoid robot market will grow in a step-by-step way, starting from 2036 and entering the commercialization stage, breaking through 127 million units in 2040 and reaching the deployment scale of 935 million units in 2050. Morgan Stanley pointed out that the cost reduction to $15,000 is a key turning point, and it will be first applied to standardized fields such as industry and logistics, while home applications are limited by security and AI technology bottlenecks, and the penetration rate is below 8% for a long time.

In addition, the report also said that China has advantages in the field of humanoid robot hardware, while the United States occupies a high position in software, ecology and algorithms. The dispute between geopolitical risks and technical standards will probably lead to industrial division, and China and the United States will each become a camp.

However, Morgan Stanley also admitted that most of the current industries are in the "proof-of-concept stage", and only 4% of enterprises have real soft and hard integration capabilities, and even warned that the industry is suspected of "pulling out the seedlings to encourage".