PhonlamaiPhoto
thesis
KLA Corporation (NASDAQ:KLAC) is well positioned to capitalize on the growth of the computer industry and the desire to re-embed supply chains. We believe KLA is fundamentally undervalued and view weakness as a buying opportunity.
Fears about WFE spending
Spending on WFE has increased dramatically over the past 7 years after remaining flat for over a decade. Below is a visualization of capital expenditures in the global semiconductor industry from 2000 to 2022:
Investments in the global semiconductor industry from 2000 to 2022 (Statista)
The rapid growth in WFE spending has led many investors to fear that spending will revert to mean. We believe these fears are short-sighted. In our view, there are a variety of tailwinds that will drive continued WFE spending over the long term.
Overall Growth in Computing
Generative AI tools like ChatGPT are enjoying fast-growing popularity, and this is just the beginning of what’s possible. Although the term “AI” is quickly becoming a buzzword, at the end of the day what’s happening now is just a continuation of what’s been happening for the past decade. New technologies require more computing power and thus more clusters, since complex tasks can no longer be efficiently performed on a piece of silicon.
A while ago I read an MIT paper on training AI to play Chess and Go (at least that’s why I read it originally), but there were many passages that caught my eye. Here are two sections that I believe are very relevant for investors in the semiconductor industry:
4.4 Implications for Future Performance Improvements This paper shows that progress in computer chess, computer go, weather forecasting, protein folding, and oil exploration depends on an exponential increase in computing power. We suggest that this is probably a broader phenomenon since the computational techniques used for these areas are also used in many others. In particular, we point out that advances in natural sciences and engineering often require finer approximations, more complicated evaluations, repeated analyzes under different initial conditions, greater search depth and more degrees of freedom. Support for this view comes from Hyperion, a computer research company that collected 690 examples of high-performance computing. They found that investing in high-performance computing significantly improves commercial success and fosters scientific innovation in many areas [90].
And from the summary:
In this article, we bring together direct quantitative evidence of the impact that computing power has had on five areas: two computer heroes (chess and go) and three economically important applications (weather forecasting, protein folding, and oil exploration). Processing power explains 49% to 94% of performance improvements in these areas. But while economic theory typically assumes a power-law relationship between inputs and outputs, we find that linear improvements in these outcomes require exponential increases in computational power. This helps illustrate why the exponential growth in computing power driven by Moore’s Law was so important to progress, and why improvements in performance in many areas become economically weak when Moore’s Law collapses.
The two things investors should focus on are:
1. An exponential increase in computing power is required to achieve linear improvements in results.
2. As Moore’s Law becomes increasingly difficult to achieve, the only way to achieve this exponential increase in computing power is through more semiconductors, rather than relying on efficiency gains.
This is one of the reasons we expect spending on WFE to increase over the long term.
Supply chain re-shoring
Another reason we think WFE spending will increase is governments’ desire to relocate semiconductor manufacturing capacity to mitigate geopolitical risks.
Congress passed the CHIPS and Science Act back in August. This law was intended to boost investment in semiconductor manufacturing capacity in the US. The desire to secure the supply of semiconductors is found in many governments around the world, and in order to achieve this, many more semiconductor foundries must be built. All of these foundries require equipment that KLA manufactures.
That’s really just icing on the cake, as global computing power growth is enough to fuel WFE spending for the next decade. Investors should be aware of the potential for excess capacity build-up as several countries seek to produce onshore and be prepared to sell KLA if their equipment sales surge unreasonably and this excess capacity scenario appears to materialize.
price action
KLA hasn’t done much in 2023, but we believe this is a buying opportunity.
Data from YCharts
Over the past five years, KLA has significantly outperformed the S&P 500, and we expect that outperformance to continue over the next decade.
Data from YCharts
Evaluation
KLA is trading at the low end of its historical valuation range. We see their current valuation as attractive, but if shares are trading above 30 times earnings we would think they are overvalued and investors should probably take their profits and look elsewhere.
Data from YCharts
risks
A risk to this optimistic thesis is the potential for advances in software to reduce the need for more hardware and thus reduce WFE investment requirements. This would drastically slow sales growth for semi-cap gear makers and damage the bull case for KLA.
Another risk to this optimistic thesis is the potential for new technological developments to reduce the need for KLA devices.
We view the overall risk/reward trade-off as favorable, however investors should be aware that the risks are significant and could have a massive impact on KLA’s long-term value.
key to take away
We believe there are several tailwinds that will fuel WFE spending growth over the next decade. KLA is well positioned to benefit from this and we believe it is attractively valued for long-term investors.
