The global artificial intelligence (AI) investment theme is showing signs of fatigue for the first time since its rally began in May 2025, with investors trimming exposure to AI-linked infrastructure and supply-chain plays even as they continue to back direct beneficiaries of the technology, according to a report by Elara Capital.
The brokerage’s latest Global Liquidity Tracker noted that diversified global emerging market (GEM) funds, which had emerged as a preferred vehicle for investors seeking exposure to the AI theme through markets such as Taiwan and South Korea, have witnessed six consecutive weeks of redemptions amounting to $10 billion. Commodity equity funds and precious metal funds have also recorded sizeable outflows.
“Collectively, these trends suggest that the broad, top-down allocation into the AI infrastructure and commodity ecosystem is losing momentum for the first time in over a year,” Elara said.
The report, however, highlighted that investor conviction in core AI beneficiaries remains intact. During the recent correction, US technology-focused funds attracted a record $9 billion of inflows, while foreign investments into US equities remained positive for the 11th consecutive week, touching a five-month high of $10 billion.
For India, the global AI-driven capital rotation continues to pose a challenge. Along with China, India has emerged as a key source of funds for investors reallocating capital towards AI-linked opportunities abroad. The report estimates that India-focused funds witnessed outflows of $770 million in the latest week, including $460 million from focused funds.
According to Elara, India’s underperformance relative to GEM funds has fallen to record lows on both one-year and three-year horizons. The brokerage attributed much of this trend to the migration of global capital towards AI-related opportunities since June 2025.
“Since June 2025, the widening underperformance has largely reflected the shift of global capital toward AI-linked opportunities,” the report said.
The brokerage noted that historical data suggests such extreme underperformance levels have often coincided with major turning points in relative performance, indicating that the current trend may be approaching an inflexion point.
Meanwhile, domestic investors in Taiwan and South Korea have used the recent correction to increase exposure, with inflows of $5.3 billion and $3.7 billion, respectively, suggesting continued confidence in the long-term AI growth story despite the near-term slowdown in flows.
