A warning has emerged that investors who used debt to buy AI and data center-related stocks should cut their positions even now. Related shares have surged amid a recent boom in AI infrastructure investment, but doubts are growing over the sustainability of data center investment, raising the risks of leveraged bets as well, an analysis said.

According to CNBC on July 27 (local time), U.S. CNBC host and investment expert Jim Cramer said enthusiasm for AI investing is gradually weakening and urged selling, especially for investors using margin trading. "If you bought data center-related stocks with borrowed money, you should absolutely sell tomorrow morning at 9:30," Cramer said. "You won't regret it," he added.

Over the past year, companies tied to AI infrastructure and data centers have extended sharp gains on expectations of expanding investment in generative AI. But the market is also increasingly skeptical about whether data center construction and AI infrastructure investment can continue at the current pace.

Cramer pointed to margin investors as being exposed to the biggest risk as uncertainty grows. He said the size of margin debt in the market has risen sharply over the past year, warning, "If you're using margin, get out now." "I don't think you'll be able to get out safely anymore," he said.

Margin trading is a way to borrow funds from a securities firm to increase the size of an investment. Returns grow when share prices rise, but losses also multiply in a downturn. If losses exceed a certain level, investors may face a margin call that requires additional collateral or forces the sale of holdings.

Cramer said investors should look to companies with more diversified growth engines rather than concentrating their portfolios in AI and data center-related stocks. He cited construction materials company CRH as a representative example. CRH supplies materials needed to build data centers, but most of its revenue comes from a range of infrastructure projects such as roads, bridges and commercial buildings, meaning its reliance on any single industry is relatively low, he said.

He also stressed that the approach to investing in technology stocks needs to change. "What we want is technology, but not the kind of technology that investors in big tech used to buy in the past," Cramer said. "Now we need to pay more attention to materials technology and science technology," he said. That is interpreted as meaning investors should favor companies that can benefit in part from expanding data center investment while not relying excessively on a specific AI theme.

He did not view the broader universe of AI-related technology stocks negatively. He assessed that investors holding high-quality tech shares using only their own capital can afford to endure a short-term pullback. "If you hold good tech stocks and didn't use margin, you can withstand a certain level of volatility," Cramer said.

The warning shows that as expanding investment in AI infrastructure and data centers has become a leading market growth narrative, doubts about its pace and durability are spilling over into share-price corrections. In particular, with more investment funding involving borrowing, losses can widen faster during the same correction, making position management an emerging key variable for market participants.