A recent report from J.P. Morgan indicates that the downturn observed in Asian tech stocks does not signify the conclusion of the artificial intelligence (AI) investment cycle. According to the bank's note released on Wednesday, there is a growing concern among investors regarding the ongoing viability of AI expenditure, although concrete evidence of a fundamental slowdown remains absent. The bank pointed out that the 25%-30% decline in Asian tech shares, as well as in the Philadelphia Semiconductor Index, represents the third significant pullback since the AI-driven uptrend commenced in late 2022.
J.P. Morgan emphasized that aside from the fluctuations in share prices, there are no fundamental indicators suggesting substantial weakness in the upcoming 6 to 12 months. The report stresses that the core pillars of the current investment cycle remain robust, with significant advancements in frontier AI models occurring every few months and a strong demand for AI inference across both proprietary and open-source platforms. The bank also highlighted the improving profitability within the AI ecosystem as agentic AI technologies gain momentum.
Despite growing skepticism from investors about whether large-scale tech companies can sustain their hefty AI investments, J.P. Morgan does not foresee these cloud providers reducing their AI compute spending. The bank asserted, “We do not expect any of the hyperscalers to pull back on AI compute investments by 2027,” suggesting that they will likely seek financing through equity and debt markets to enable further growth of AI infrastructure. They noted that the current market may already be accounting for a downturn that is unlikely to materialize, predicting instead a scenario of broader earnings upgrades and sustained AI-related capital investments.
Several catalysts that could enhance investor sentiment were identified, including greater adoption of generative AI and agentic workflows by software firms, increased implementation across sectors such as financial services and healthcare, and ongoing progress by leading AI laboratories in recursive self-improvement.
When examining the semiconductor supply chain, J.P. Morgan stated that equipment manufacturers appear to be the strongest segment poised for growth over the next year as spending on wafer fabrication equipment accelerates. The bank anticipates significant growth in packaging and testing sectors, particularly with 2.5D packaging—a method incorporating multiple chips side by side—becoming commonplace and a new investment cycle for 3D packaging starting at TSMC. Among semiconductor components, integrated circuit substrates were highlighted as the most promising opportunity.
The memory sector presents a more complex narrative. While the supply-demand fundamentals for memory products remain solid, the bank projects that supply will fall short of demand for the next two to three years. The analysts noted that recent decisions by Nvidia and AMD to reduce the amount of memory included in future AI products have shifted the prevailing market perception that AI-driven memory demand is resistant to price changes. They remarked, “The market narrative on memory is problematic, even though the fundamentals are sound,” and while a rebound in this sector could occur in the next six months, they do not expect memory stocks to return to their peak levels from May.
Looking forward, J.P. Morgan predicts that within the next 18 to 24 months, power availability will surpass chip production as the primary limitation on AI computing infrastructure as semiconductor capacity continues to increase.
