Morgan Stanley warns that the valuation of the semiconductor sector is overheated. Tech giants developing their own chips may weaken the industry's pricing power.

Lisa Shalett, Chief Investment Officer of Wealth Management at Morgan Stanley, stated that with increasing signs showing the pricing power of chip manufacturers being challenged, investors should remain cautious about the semiconductor sector. She believes that the market's optimistic expectations for investment in Artificial Intelligence (AI) infrastructure may have driven semiconductor stocks to rise too sharply, and the growth rate of AI capital expenditure is also showing signs of slowing down. Shalett said in an interview with the media on Friday that the architecture of AI data center technology is changing, with more and more large-scale cloud service providers beginning to design their own specialized AI chips at lower costs to reduce their dependency on traditional chip suppliers. She pointed out that the current AI supply chain is experiencing a typical phenomenon where downstream companies accelerate the development of lower-cost alternative solutions when supply chain bottlenecks result in excessive profits for some companies. Currently, some storage chip manufacturers are benefiting from higher profits due to tight supply and demand, but this pricing advantage may not be sustainable in the long term. At the time Shalett made the above remarks, South Korean storage chip giant SK hynix (SKHY.US) officially listed on the NASDAQ on Friday. The company raised $26.5 billion through its initial public offering (IPO), setting a record for the largest IPO by a foreign company in the United States. However, its stock price has fluctuated significantly recently, falling by about 26% since its peak last month. Nevertheless, Shalett believes that there is still ample market capital for AI investment themes, and valuations in the semiconductor sector are clearly overvalued. A report she released this week indicated that several indicators, including semiconductor ETFs and the Philadelphia Semiconductor Index, show that the sector is in a "significantly overbought" state. Data shows that since 2022, the price-to-earnings ratio of the Philadelphia Semiconductor Index has more than tripled. She also mentioned that Meta Platforms (META.US) recently adjusted its AI strategy, reflecting that some tech giants are reevaluating the return on investment for their multi-billion-dollar AI capital expenditures. Previously, Meta CEO Mark Zuckerberg stated that the company is considering renting out some of its AI infrastructure externally to improve asset utilization efficiency and commercial returns. Shalett believes that this indicates that tech companies internally are starting to discuss the pace of AI investment construction, the return on investment, and how to achieve commercialization sooner. She expects that the trend of slowing growth in AI infrastructure capital expenditures is still in its early stages, but it is worth continued market attention in the future.