This is not investment advice. The author has no position in any of the stocks mentioned. Wccftech.com has a disclosure and ethics policy.

The Taiwan Semiconductor Manufacturing Company's (TSMC) earnings call for its second quarter of 2023 came with a bombshell announcement as the firm's chairman, Dr. Mark Liu, announced that its new chip manufacturing facility in the U.S. will delay production by a year. TSMC is building a new facility in Arizona, and the firm had previously announced that it would begin producing 4-nanometer (N4) products at the site next year. However, according to Dr. Liu, N4 production will be delayed until 2025 as his firm is facing difficulty finding employees responsible for installing the machines for advanced semiconductor fabrication.

TSMC Asserts 2023 Revenue Will Beat Overall Market Slowdown As Inventory Correction Continues

The key takeaways from TSMC's earnings covered the firm's expectations for performance this year and the longer term growth rate. On the former front, the firm's CEO Dr. Wei was forthcoming in stating that his firm had underestimated worsening macroeconomic conditions. He believes that the broader downturn in the chip sector will reduce 2023's revenue by 10% - and the current hype surrounding increased demand for artificial intelligence products will prove insufficient to counteract the slowdown.

His remarks were backed up by TSMC's management report for the second quarter, which included details about revenue growth or slowdown for the various technology platforms that are powered by its products. TSMC categorizes these platforms into six categories, namely high-performance computing (HPC), smartphones, internet of things (IoT) gadgets, automotive, digital consumer electronics (DCE) and others. Only DCE and automotive saw revenue growth at 25% and 3%, respectively. However, the two segments cumulatively represented 11% of TSMC's revenue in Q2, while HPC is the largest with a 44% share.

TSMC's revenue by segment growth or deceleration rates in the second quarter of 2023. Image: TSMC

The downturn in the industry is coming right at a time when TSMC is facing high depreciation expenses for its N3 chip technology. The first couple of years of a new process technology see fabs absorb higher costs for setting up machines for production, and TSMC kicked off N3 chip production late last year.

TSMC CEO Dr. Wei was quite clear in his outlook on the macroeconomic situation, as he explained:

First, the macro is weaker than we thought. You know, three months ago, we [were] probably more optimistic, but now it's not. For example, China economy's recovery is actually also weaker than what we thought. So the end market demand did not grow as we expected. So put all together, even [when] we have very good AI's processor demand, it's still not enough to offset all those kind of macro impact. So now we expect the whole year will become minus ten percent.

He added that all market segments were being affected and that:

'[i]t's all about the macro. The macroeconomics has become weaker than we thought. In fact, high inflation and interest rate impact end up in all market segment[s] in every region in the world. Under such [a] situation, our customers are more cautious in their inventory control in the second half of this year.

However, the executive remained optimistic that the fabless semiconductor industry will exit December with healthier and lower inventories than TSMC had expected in Q1. Additionally, TSMc expects the broader semiconductor industry to decline in the mid-teens while the firm's revenue will fall slower at roughly 10% in 2023.

Income tax payments gutted TSMC's free cash flow in Q2 2023 right when the market had slowed down.. Image: TSMC

As part of his prepared remarks, TSMC chairman Dr. Mark Liu announced right off the bat that the Arizona fab's N4 production schedule is pushed a year back to 2025. A key challenge is finding the right workers to install chip manufacturing equipment at the plant, and TSMC plans to bring in workers from Taiwan to train those in the U.S. These plans have caused some worry among workers that they might see pay cuts or job losses - claims that TSMC has denied.

Cost gaps in setting up the Arizona fab are as much as 50% higher compared to other facilities, and Dr. Liu answered in response to a question:

However, we try to work with the U.S. administration. First of all, on the subsidy, cash subsidy and tax -- investment tax credit, that is to cover the gap in the first 5 years approximately. When the tool is depreciated, then the ecosystem becomes prominent. That is, what is that, material costs, chemical costs and the labor cost. And we are working with our supplier to set up some of the more efficient supply sites to be lower, but they -- and the U.S. administration has decided also to subsidize the supply -- our suppliers.

So that is still in the work. How much it can further decrease, I don't know. But I think either way, we will strengthen our pricing values and be able to keep the corporate profitability as we forecasted now.

TSMC's chairman Dr. Mark Liu in Tainan, Taiwan in November 2022 as part of a beam lifting ceremony for a 3-nanometer manufacturing extension. Image: Liu Xuesheng/UDN

The TSMC chair also urged caution about the short-term trends in demand for AI products and any conclusions about higher demand in Q3 2023 sustaining in the later quarters. Dr. Wei added that AI products have already started to use TSMC's latest N3 process technology node, and overall, these products currently represent 6% of TSMC's revenue.

Over the next five years, the TSMC CEO expects AI chips, including GPUs, CPUs, and accelerators, to grow at a compounded annual growth rate (CAGR) of 50% and represent a "low teens percent" of its revenue. TSMC's capacity to manufacture high-end chips is currently mixed; while the firm can ship front-end products, it is facing a capacity constraint at the back end, which it believes should be resolved by the end of 2024.

Commenting on AI demand and the extent to which it overlaps with the demand for data center products, Dr. Liu commented:

Of course, we have a model, basically. The short-term frenzy about the AI demand definitely cannot extrapolate for the long term. And neither can we predict the near future, meaning next year, how the sudden demand will continue or will flatten out.

However, our model is based on the data center structure. We assume a certain percentage of the data center processor are AI processors, and based on that, we calculate the AI processor demand. And this model is yet to be fitted to the practical data later on. But in general, I think the -- our trend of a big portion of data center processor will be AI processor is a sure thing.

And will it cannibalize the data center processors? In the short term, when the CapEx of the cloud service providers are fixed, yes, it will. It is. But as for the long term, when their data service -- when the cloud service is having the generative AI service revenue, I think they will increase the CapEx. That should be consistent with the long-term AI processor demand. And I mean the CapEx will increase because of the generative AI services. Anything more for you?

TSMC expects a hefty drop in its Q3 2023 revenue in U.S. dollar terms, with its guidance outlining a minimum of $16.7 billion in revenue and a maximum of $17.5 billion. Compared with 2022, the lower-end guidance would lead to a $3.5 billion annual drop in revenue from Q3 2022's $20.23 billion.