Inventory is the culprit. Much of his 2023 modeling for TSMC was based on the belief that customers could sell their stockpiles and that new orders would flow in as inventories dwindled. The post-pandemic situation has not progressed at the pace TSMC expected. As a result, the company’s days-in-stock went up, not down, at least to his highest level in 12 years.
But predictions that TSMC would be shaken and cut its capital budget by as much as $4 billion to offset its weak outlook didn’t pan out. Instead, the biggest surprise on the earnings call is that he kept his January forecast for this year’s capital spending at $36 billion from his $32 billion. In 2022 he was $36.3 billion.
There are several reasons for stubbornness.
First, approximately 70% of spending is on gear used in advanced manufacturing techniques. Supplies continue to be tight for items such as ASML Holding NV’s Extreme Ultraviolet equipment. On Wednesday, the Dutch company said its forecast to ship 60 of his EUV systems this year has not changed, and the company is facing his two-year backlog on several tools. rice field. TSMC may cut orders, but doing so could jeopardize its place in the queue, putting it at a disadvantage and helping competitors who want to get their hands on the vital machinery.
Second, TSMC chief executive CC Wei told investors the company is focusing on expansion so it can meet capacity needs as soon as customer orders start pouring in. rice field. .
Third, while demand for smartphones and PCs remains weak, artificial intelligence is a promising area due to the recent boom in generative applications such as ChatGPT.
If I had to pick one segment that would be strong during a chip downturn, it would be AI. TSMC’s high-performance computing category, which includes specialized AI components, accounts for 44% of revenue. Additionally, clients in this space, such as Nvidia Corp., Microsoft Corp., and Advanced Micro Devices Inc., require modern manufacturing nodes to maintain their competitive edge. Having products made with old technology dilutes the power and performance of the chips they design. This state-of-the-art production also comes at a premium price.
“We have recently observed an increase in AI-related demand to help with ongoing inventory digestion,” said Wei. “ChatGPT is now a structural megatrend and he already has a strong belief in HPC and AI.”
Even with lower earnings, the commitment to stay ahead of TSMC will certainly hurt profitability. Declining capacity utilization, rising power prices in Taiwan, and high construction costs for expansion in the US and Japan exacerbate the problem. The ability to contain costs, as demonstrated in the first quarter, could alleviate this challenge.
Yet, for better or worse, management chose to stay on course rather than pursue profit. Investors may not like this strategy, but customers will love it.
Bloomberg Opinion Details:
• Moore’s Law puts the chip leader ahead of the pack: Tim Culpan
• Indian tech hub trapped between SVB and traffic: Andy Mukherjee
• Intel’s foundry bet could split the market into three: Tim Calpan
This column does not necessarily reflect the opinions of the editorial board or Bloomberg LP and its owners.
Tim Kalpin is a Bloomberg Opinion columnist covering technology in Asia. Prior to that, he was a technology reporter for Bloomberg News.
More articles like this can be found at bloomberg.com/opinion.
