A recent research report from Jefferies indicates that investor focus within the artificial intelligence (AI) sector is shifting from hyperscale cloud platform companies to upstream suppliers providing core components for the AI industry. Analyst Christopher Wood, in his latest "Greed & Fear" market analysis report, stated that the multi-year investment boom in AI has entered a new stage, with memory chip manufacturers replacing chip designers and cloud platforms as the core entities wielding pricing power. Following a significant surge in contract prices for high-end memory chips late last year, leading memory manufacturers like SK Hynix and Micron (MU.US) have seen substantial growth in both their financial performance and stock prices. The report, released on January 22nd, mentions that although overall investment in non-AI sectors shows signs of weakness, capital expenditure related to AI continues to climb. Official U.S. data shows that last year, market spending on AI-related software, equipment, and data centers achieved double-digit growth, while other types of corporate investment declined. The market has begun pricing in expectations for a broad recovery in the industrial and energy sectors, but macroeconomic data has not yet fully supported this expectation. The report also points out that sector performance in global stock markets is showing significant divergence. Since late October last year, the stock performance of leading hyperscale cloud providers and AI platform companies has consistently lagged, while the stock prices of memory chip manufacturers and semiconductor fabrication companies have risen sharply. Jefferies estimates that memory chip prices increased by approximately 50% last quarter, a trend that further solidifies the dominant position of memory manufacturers in the AI supply chain. However, this dominant position comes with rising costs. Building new wafer fabs now requires investments of tens of billions of dollars, and Jefferies states that some memory chip manufacturers are pushing customers to share these costs in exchange for a stable supply guarantee. This marks a significant shift in market focus, as concerns previously centered on large chip buyers squeezing profit margins. Despite the ongoing advancement of AI infrastructure construction, Jefferies cautions that the capital expenditure cycle in the AI field has moved beyond its initial phase. Comparing the current market environment to other capital-intensive industries, the report notes that return on investment in such sectors often gradually normalizes over time, and the core risk now is whether investors will begin to question if the massive investments in AI can generate sufficient profits to justify their scale. This concern is already becoming apparent in capital markets: in recent months, despite plans for further expanded capital expenditure by leading cloud providers and internet companies by 2026, their stock prices have fallen, while the stock prices of chip and memory manufacturers have continued to rise. Jefferies calculates that capital expenditure by hyperscale cloud providers will still increase significantly this year, which also subjects them to greater pressure to deliver returns. The report simultaneously warns that surging energy demand is becoming a structural issue accompanying the development of the AI industry. Several technology companies have begun directly securing their energy supply, signaling a gradual industry transition towards a heavy-asset business model. Jefferies' overall assessment is that the investment logic for the AI sector remains intact, but the core beneficiaries within the industry have changed. In the short term, memory chip manufacturers are the biggest winners; meanwhile, investor attitudes are becoming increasingly cautious and demanding towards companies making massive investments in AI infrastructure without clear evidence of profitability.
Comments