The feast of generative AI has brought a glimmer of hope to semiconductor chip companies in the midst of a cold winter. TSMC has announced that it is ready for the next wave of growth.
On June 6th, TSMC held its shareholders' meeting, during which Chairman Liu Deyin stated that &#34;there are already signs of a rebound in demand in certain market segments.&#34; TSMC may experience a slight negative growth this year, but it is prepared for the next wave of growth:
<blockquote>
2022 will be a year of growth for TSMC, and 2023 will be the year to work through inventory. However, customer inventory is gradually decreasing, and there are already signs of demand recovery.
TSMC's sales in 2023 are expected to decline by a single digit percentage, and sales in the first half of the year are expected to decline by 10% in US dollars.
</blockquote>
According to analysis by Wall Street in multiple articles, the downstream demand for the chip industry is indeed low. In 2022, as the shipments of consumer electronics such as smartphones, PCs, and tablets continue to decline, the semiconductor industry has entered a downward cycle.
Data released by Canalys shows that global smartphone shipments in 2022 fell by 12% YoY, the lowest in a decade; PC shipments fell by 16% compared to the demand peak in 2021.
Correspondingly, global semiconductor industry sales in 2022 were nearly $580 billion, a YoY increase of 7%, significantly slower than the 20.6% increase in 2021.
TSMC is also unable to escape the market downturn. According to the financial report released this year, the company's revenue and net profit both fell significantly MoM. However, during the financial report conference call, TSMC CEO Wei Zhejia stated that &#34;3nm advanced process chips will be mass-produced in the second half of the year, and the demand for chips relying on artificial intelligence will increase, bringing new opportunities for the company's development.&#34;
At this shareholders' meeting, Liu Deyin emphasized that &#34;advanced packaging technology is the key to TSMC's value growth, and AI has brought more demand for advanced chips:&#34;
<blockquote>
&#34;75% of TSMC's R&amp;D budget is used for advanced technology, and 25% is used for special technology. The booming development of generative AI has brought TSMC more orders. TSMC is preparing to increase production capacity to meet the demand brought by AI, and it is expected that the demand for AI in the terminal equipment field will increase.&#34;
</blockquote>
Overnight, TSMC's US stock fell 0.9% to about $98. TSMC's US stock has risen by 31.63% this year, far exceeding the 11% rise of the S&amp;P 500 index.
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/c169d124-c7fc-48bd-a141-7539e3cb1169.png?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/c169d124-c7fc-48bd-a141-7539e3cb1169.png"/>
<h2>NVIDIA Helps TSMC</h2>
There is no doubt that the surge in demand for TSMC's advanced process chips is closely related to the explosive demand for NVIDIA's GPUs in the market.
The relationship between the two companies can be traced back to the mid-1990s. NVIDIA, which had just been established, was worried about chip outsourcing. NVIDIA CEO Huang Renxun found TSMC's founder Zhang Zhongmou, who was constantly breaking through in the field of semiconductor outsourcing at that time. From then on, the connection between the two companies was established.
With the popularity of generative AI, the demand for NVIDIA GPU chips in the market is also soaring, and NVIDIA is constantly adding orders to TSMC. Earlier, some media reported that NVIDIA added CoWoS (wafer-level chip) packaging orders to TSMC, and added another 10,000 pieces on the basis of the 2023 scheduled quantity to meet demand.
Currently, TSMC has listed the production of NVIDIA AI GPUs as a &#34;super urgent item&#34;, and orders are full until the end of the year.
According to semiconductor equipment manufacturers, the shortage of NVIDIA AI GPUs is mainly due to the fact that NVIDIA and TSMC's previous estimates of annual capacity and orders were quite conservative, and they did not expect the outbreak of demand for HPC chips such as AI GPUs. The CoWoS wet process advanced packaging equipment, which originally had limited capacity, is difficult to meet demand.
In addition, some packaging equipment and zero components have a lead time of up to 3-6 months, and the new capacity can only be gradually put in place by the end of the year, which means that the next six months will be a shortage situation. It is rare to see TSMC making urgent orders without cutting prices. TSMC President Wei Zhejia also revealed for the first time that he recently had a telephone conference with a certain customer and requested to expand CoWoS orders.

NVDA

Previously, it was reported that Nvidia is continuously increasing its orders to TSMC. However, the unexpected surge in demand for AI GPU and other HPC chips in the market has made it difficult for the CoWoS wet process advanced packaging equipment, which already had limited production capacity, to meet the demand.

Generative AI brings good news! TSMC: Ready to embrace the next wave of growth.