What signals? UBS bluntly states "the disaster of AI": the volatility of tech stocks has reached its highest level since the internet bubble.

A recent report from UBS's research division, Holt Unit, indicates that volatility in the global technology sector has surged to its highest level since the burst of the internet bubble, as investors reassess whether the cash flow returns from AI-driven hyperscale data centers and semiconductor companies can be sustained. The HOLT Unit (HOLT Team) is a core quantitative team within UBS focused on value-driven investment research, and it is not an independent legal entity; it belongs to the professional groups under UBS's Global Investment Bank or Wealth Management Research division. The analysis highlights that spending on AI infrastructure is putting increasing pressure on the sector. Although profit margins remain stable, the massive investments from large hyperscale data center operators are weakening their asset efficiency, leading to a decline in cash flow return on investment (CFROI) by 2028. UBS estimates that the top five hyperscale data center operatorsMicrosoft, Meta, Alphabet, Amazon, and Oraclewill face a total funding gap of $227 billion next year, unable to meet their operational and financing commitments. At the same time, historical data shows that since 1998, there have been approximately 650 spikes in large capital expenditures, with 60% of those spikes resulting in a permanent decline in CFROI. This effect is most pronounced for companies that already had a high initial return rate. On the other hand, UBS points out that the semiconductor industry faces a variety of different risks. The sector's yields have nearly tripled, reaching around 30%, with fewer than 1% of companies achieving such a feat since 1990. UBS notes that the current valuation assumes this high yield will last for five years, which contradicts typical competitive dynamics. The report also cites examples from Chinese AI developers DeepSeek and Moonshot AI to illustrate that, given China's tendency to prioritize market share over profitability, the so-called "moat" in the industry may not be insurmountable. Furthermore, due to concerns about the disruptive impact of AI, valuations of software, enterprise data, and services stocks have reset, with these stocks' total price-to-book ratios decreasing by about 40% over the past 18 months. UBS finds that historically, 80% of stocks that have experienced a similar degree of devaluation failed to return to prior valuation levels within ten years. The price-to-book ratio (P/B) reflects the ratio of a companys stock price to its net asset value per share, indicating how much ordinary shareholders are willing to pay for every 1 yuan of net assets, representing the market's evaluation of the company's asset quality. Finally, looking beyond the technology sector, UBS notes that during the recent sell-off triggered by tech stocks, value stocks and low-volatility stocks have been the strongest performing style factors, with value stocks excelling in all six major sell-offs since 2004 and low-volatility stocks performing well in four of those instances. However, the firm warns that both strategies have limitations in the current environment. Since 2023, the traditional link between value stocks and the economic cycle has weakened, while low-volatility stocks tend to perform poorly outside of sell-off periods unless their fundamentals are strong. This article is reproduced from "Caijing Lianhe," author: Huang Junzhi; GMTEight editor: Xu Wenqiang.