Hiraman
Are we in a new demand cycle for tech hardware as the COVID-era PC- and smartphone-purchase boom hits the first leg of its replacement cycle? And will AI-enabling devices act as a tailwind? If so, how can investors benefit?
Those are the questions we sought to answer earlier this year when five of our global equity portfolio managers and analysts visited Taiwan, which is tough to top from an efficiency perspective: dozens of tech companies are clustered in a handful of science parks.
We met with 25 companies in varying tech industries across the market-cap spectrum and came away with the following key insights.
A Replacement Cycle Gains Traction
Our investable universe in Taiwan effectively consists of technology hardware (33.5%) and semiconductors (66.1%), with only a few software or IT services companies.
Within tech hardware and semis, smartphones and PCs are still the largest end-markets, despite faster growth in areas such as automotive and industrials. Smartphones and PCs were also two of the biggest topics of conversation on our trip.
Both of these end-markets saw double-digit unit shipment declines in 2022 and 2023, but their outlook is improving. A destocking process led to lower inventories, but demand is turning positive, so we could see more production in 2024. With the supply-demand imbalance starting to move back in the companies’ favor, they have more pricing power.
The smartphone industry, for example, had exceptionally high inventory levels in the first and second quarters of 2023, but there was a noticeable decrease in the fourth quarter of 2023. This led to a surge in rush orders for tech hardware and semis, and subsequently component restocking and significant inventory adjustments. These adjustments suggest an improvement in growth across the sector. Moreover, we anticipate a rebound in smartphone unit growth, perhaps a 5% increase in 2024-2025, following a 10% decline in 2022-2023.
In the PC sector, our investment targets have consistently increased their market share. Each time a new PC model is introduced, industry standards evolve, typically leading to an increase in the average selling price and subsequent growth for these companies’ products.
It is important to emphasize that, in our view, the current trend is not merely a short-term cyclical upswing or a temporary restocking phase, as might have been the case with similar industries in Taiwan a decade ago. Instead, we believe we are observing a sustainable, long-term tailwind that will potentially benefit the companies in our portfolio.
One of our largest holdings—a Taiwanese multinational semiconductor contract manufacturing and design company—has a core business focused on PCs and smartphones. But smaller companies are also benefiting from this replacement cycle.
For example, a design services firm in our portfolio uses advanced packaging capacity at the foundry (the semiconductor manufacturing and design company mentioned above) on behalf of end-clients such as Intel (INTC) and Amazon (AMZN). This design services firm is a broad AI enabler in the hardware supply chain.
Another smaller company manufactures hybrid bonding equipment that is critical for advanced packaging. It is not based in Taiwan but illustrates that the supply chain is larger than the island.
As we move into a discussion of artificial intelligence (AI), it is worth noting that we’re in a traditional replacement cycle for both smartphones and PCs. But what if those devices become AI-oriented?
When you add AI to the traditional hardware upcycle, you’ll likely see increased demand. Whether that means we’ll get massive content upgrades on our existing smartphones or we’ll all have two devices is yet to be seen; but either way, it’s the spending we care about, and people will surely spend more for the productivity gain offered by the new device or the killer app. There’s an excitement there.
Another Tailwind: AI
AI presents a second avenue of growth for many of these Taiwanese companies, especially those that are aligned with the AI server.
For example, 6% of 2023 revenue for the aforementioned Taiwanese multinational semiconductor company was AI-related. Certainly, that’s a small portion, but we believe it has the potential to double or triple over the next few years.
For this company and others like it, key AI offerings would be semiconductors that are specifically designed for the tech stack that’s enabling the AI cloud—think Google (GOOG)(GOOGL), OpenAI, Meta (META), and all the other companies that are building out their own proprietary AI businesses. These companies need more advanced semiconductors; they need hardware (servers); they need the memory that goes into these servers.
Over the long term, we believe the software model is poised to capture significant AI-related profits, because a substantial portion of each sale translates into additional revenue and margin.
Microsoft (MSFT) is perhaps the best example. It charges users $10 a month for Outlook 365. But Outlook with Copilot—the AI-enabled version—costs $30 a month. That’s $20 more per user with AI added.
Think about how many people use Microsoft, and how much more efficient they could be with AI. So, we believe there will be high adoption, and thus a significant increase in revenue and margin for these software companies.
Of course, that won’t happen until acceptance sets in, and we’re still in the early stages—our team refers to this as building the foundation. But companies are building the hardware to enable that growth now, and that’s why we’re so bullish on the hardware supply chain.
Consider AI servers. Specifically designed for AI workloads, including machine learning and deep learning tasks, these servers are optimized to handle complex computations, massive data sets, and complex algorithms, which require intensive processing power and fast data throughput. That requires specialized hardware. Among the people we spoke with on our trip, there was continued optimism around unit growth for AI servers, and everyone expects non-AI server unit growth to be close to flat this year.
Growth is also expected in advanced packaging, which refers to a set of sophisticated semiconductor packaging techniques that go beyond traditional methods to improve performance, reduce size, and enhance functionality of semiconductor chips. Unlike conventional packaging, which primarily serves to protect and provide connections for a single chip, advanced packaging integrates multiple chips into a single package, enabling higher performance and functionality while reducing space and power consumption.
There is a significant constraint in the availability of advanced packaging capacity, which is necessary for the production of AI chips. This scarcity affects the ability of companies to package their AI chips effectively.
In other words, the demand for advanced packaging services, which are essential for assembling and finalizing AI chips, outstrips the available supply, leading to challenges in meeting production needs for these high-tech components. This bottleneck can impact the timely delivery and scaling of AI technologies in the market.
In the near term, volumes will likely be limited, so the revenue opportunity is unlikely to start in 2024; it’s more of a longer-term opportunity.
Portfolio Implications
Reflecting on our investment strategy, which emphasizes understanding market cycles, the past year’s gains in Taiwan tech were primarily driven by increases in stock valuations. In 2024, we anticipate the focus to shift toward earnings. In essence, we are concentrating our investments on advanced drivers—the technologies and companies at the forefront of their fields—and companies showing signs of fundamental improvement and expected demand resurgence in their respective markets.
Editor’s Note: The summary bullets for this article were chosen by Seeking Alpha editors.
