Hedge Funds Increase Bets Against AI Shares as Investors Question Valuations

LONDON/NEW YORK — Hedge funds increased short positions against several artificial-intelligence-related companies during July as investors became more cautious about valuations following a sharp technology-market sell-off.

Data platform Hazeltree, whose clients include hundreds of hedge funds, said companies including CoreWeave, Super Micro Computer and Nebius Group were among shares attracting significant bearish positioning.

Short selling involves borrowing shares and selling them in the expectation that their price will fall. The investor later attempts to buy them back more cheaply and return the borrowed stock.

The positioning reflects an important shift in the AI investment boom. Investors are no longer treating exposure to artificial intelligence itself as sufficient justification for rapidly rising valuations. Increasingly, they are asking whether companies can convert expensive computing infrastructure into sustainable profits.

July was particularly difficult for semiconductor stocks. Major chip-sector indices suffered steep declines amid concern about spending levels, financing structures and whether demand growth can continue indefinitely.

However, Wednesday produced a reminder that sentiment remains highly divided. Shares in AI infrastructure companies including CoreWeave and Super Micro rose after positive corporate updates pointed to continuing demand for data-centre capacity.

The result is a market characterised by unusually strong opposing positions: some investors see AI infrastructure as one of the most important long-term investment opportunities in decades, while others believe expectations have become too expensive.

Short positions themselves do not prove that an industry is weakening. They represent investors' expectations and risk management strategies.

The central issue increasingly confronting the sector is straightforward: can AI revenue grow quickly enough to justify hundreds of billions of dollars being spent on chips, data centres, electricity and financing?