AI infrastructure dividends continue to be paid out! TSM.US Q2 net profit surged 77% to a new high and raised annual capital expenditure and revenue guidelines

The Zhitong Finance App learned that thanks to strong global demand for artificial intelligence (AI) chips, TSM.US on Thursday announced second-quarter profits far exceeding expectations. According to the data, TSMC's Q2 revenue was NT$1.27 trillion (US$39.45 billion), up 36% year on year, with market expectations of NT$1.264 trillion; net profit of NT$706.562 billion, up 77.4% year on year, up 23.4% month on month, with market expectations of NT$632.64 billion.

The net profit of the world's largest chip foundry hit a record high for the fifth consecutive quarter. This performance was mainly due to the explosive growth in global demand for AI infrastructure, which led to strong orders for advanced manufacturing processes and advanced packaging technology. According to TSMC, advanced process technology at 7 nm and below accounts for 77% of total wafer revenue.

TSMC raises annual capital expenditure to anchor long-term demand for AI chips

Notably, TSMC raised this year's capital expenditure and full-year revenue expectations, demonstrating the company's strong confidence in global demand for AI chips.

TSMC currently expects capital expenditure to reach 60 billion to 64 billion US dollars in 2026, which is higher than the previous forecast of 52 billion to 56 billion US dollars. The company also expects revenue growth in US dollars to be slightly above 40%, higher than the previous forecast of more than 30%.

Performance continues to increase, confirming TSMC's central position in the global AI industry chain: the vast majority of the world's advanced chips used in data centers and smartphones are manufactured on an OEM basis. The company with the highest market capitalization in Asia has become a core weather vane for measuring the global AI infrastructure boom.

TSMC confirmed on Thursday that it is likely to increase spending further in the next three years. This trend of accelerated expansion reflects that large technology companies, from Meta to Alphabet, will continue to purchase large quantities of the chips and hardware needed to build data centers.

TSMC Chief Financial Officer Huang Wende told analysts during the earnings call: “We are confident in the big trend of artificial intelligence. The investment expenses in the next three years will be even greater, far higher than the past three years.”

Part of the increased spending is likely to go to Arizona in the US, where TSMC has set aside a total budget of 265 billion US dollars for capacity expansion in Arizona. TSMC President Wei Zhejia said as early as June that even if a large amount of chip manufacturing capacity is added in the US, the company will not be able to meet the needs of American customers in the next few years.

Market differences still exist: the boom in computing power expansion conceals hidden concerns

TSMC's capital expenditure has attracted much attention and is regarded as a key leading indicator of global core chip supply and demand. It covers categories from Nvidia AI accelerators, Tesla in-vehicle processors, to server core chips carrying various AI services.

South Korean chip giant SK Hynix also believes that the shortage of memory chips will continue beyond 2030, as the spending boom in data center operators stimulates demand for traditional memory and high-bandwidth memory (HBM) used in conjunction with artificial intelligence systems.

This year alone, the amount of capital invested by tech giants such as Meta into A-computing power infrastructure is expected to exceed 725 billion US dollars. However, market differences have emerged: Currently, the overall valuation of the AI sector is at a high level, and many investors are concerned that major manufacturers are piling up to expand their computing power, which may cause an oversupply of chips.

In addition, leading data center operators continue to borrow and finance to raise funds for computing infrastructure expansion, and a large part of their AI-related investment is supported by rising debt. Although the pace of implementation of the global AI industry continues to accelerate, there is still great uncertainty about whether the huge investment can deliver impressive returns.

Since this year, the global semiconductor sector has risen sharply, catalyzed by an explosion in global demand for AI computing power. After the earnings season began, the market began to re-examine the chip industry's valuation, and whether the AI market can continue to be the focus. The collective pressure on Asian chip stocks on Thursday dragged down the market index, fully reflecting current investors' concerns about the overvalued AI industry chain.