Temasek's Chief Investment Officer warns: AI trading reversal is the biggest risk in the market, and 2027 may be bumpy

The Zhitong Finance App learned that Singapore's state-owned investment giant Temasek said that the reversal of artificial intelligence transactions is the biggest risk facing the market.

Temasek Chief Investment Officer Rohit Sipahimalani said, “We don't think this is an imminent thing, but will 2027 be bumpy? Yes, it's possible.”

Sipahimalani said that even if US Treasury yields rise sharply, artificial intelligence is still one of the key forces supporting US stocks to maintain near record highs. He pointed out that the profit performance of large companies related to this technology is strong.

He said that despite rising borrowing costs, the S&P 500 index is still hovering near record highs, supported by “AI and profit momentum surrounding top players.”

However, strength at the exponential level overshadows apparent weakness. Sipahimalani pointed out that about half of the constituent stocks in the Russell 3000 index fell at least 20% from their June high, highlighting the extent to which the resilience of the market depends on a small group of winners.

He said the reversal of AI transactions could be triggered by multiple factors, including security issues leading to tighter regulations, or signs that clients' investment in this technology is not yielding sufficient returns.

However, Temasek has long been optimistic about artificial intelligence and continues to increase investment in this field. Sipahimalani said that about half of the investment company's current AI exposure is allocated to publicly traded assets, and ideally this ratio will increase to around 70% to 75%.

Compared to private equity assets that are more difficult to exit quickly, this will give Temasek greater ability to adjust investments as the industry evolves.

“What we recognize is that AI is a rapidly changing environment, things can change easily, and you have to be able to pivot,” he said.

Temasek has invested in private AI model developers such as OpenAI and Anthropic, but “the size of our exposure in these areas will be different compared to others where we have more flexibility,” Sipahimarani said.