CIO continues to favor the semiconductor space and megacaps for AI exposure, and recommends investors consider structured strategies or a buy-the-dip approach for quality AI stocks. (UBS)

However, we remain positive on the tech sector as well as the outlook for artificial intelligence (AI). Against this backdrop, we believe volatility should be utilized to build long-term AI exposure.


Major AI customers have committed to strong spending into 2025. Both Alphabet and Meta highlighted the risks of underspending, rather than overspending, on AI during the second-quarter reporting season, and our analysis shows that big tech’s combined capex intensity (capex divided by sales) is still below historical peaks. Additional support from a range of customers like other cloud platforms, enterprises, and sovereign entities also underscores our positive outlook on AI semis. We expect overall AI semiconductor industry revenues to grow from USD 58bn in 2023 to USD 168bn by the end of this year and to USD 245bn by end-2025.


The upcoming third-quarter results should provide a catalyst for markets. We believe investor focus will quickly shift to tech fundamentals as the earnings season begins this week. While second-quarter results were mixed, we expect tech and AI companies to “beat and raise” for the September quarter. Earlier today, Taiwan Semiconductor Manufacturing Company (TSMC) posted a better-than-expected 40% year-over-year rise in September revenue, with full quarterly results expected next week. Overall, we forecast earnings growth of around 35% in 2024 for our AI stock selection and another 25% in 2025.


An accelerating technology upgrade cycle is underway. The initial computing demand for generative AI has been mostly driven by graphics processing units (GPUs) or custom chips that are based on transistor (a semi device used to amplify or switch electrical signals and power) sizes of 5 and 4 nanometers. However, according to major chipmakers’ product development roadmaps, the stage is set for a leap in compute power over the next five years, leading to reduced transistor sizes. Smaller sizes would allow more transistors in a chip, meaning it can do more processing in the same amount of time. This should lead to significant performance improvements and result in elevated investments in AI chips. We also see select beneficiaries in the semicap equipment and data center supply chain segments.


So, we continue to favor the semiconductor space and megacaps for AI exposure, and recommend investors consider structured strategies or a buy-the-dip approach for quality AI stocks. For investors willing and able to manage risks such as illiquidity, we see broad-based AI opportunities in private markets with a focus on large language models, software applications, and data centers


Main contributors – Solita Marcelli, Mark Haefele, Sundeep Gantori, Daisy Tseng, Jon Gordon, Christopher Swann


Original report - Tech volatility provides opportunity for long-term AI exposure, 9 October 2024.