Investor Michael Burry has taken a bearish short position against the AI cloud-infrastructure company Nebius Group [1, 2].
The move signals a growing skepticism from some high-profile investors regarding the sustainability of AI-driven valuations. While many market participants remain bullish on artificial intelligence, Burry's bet suggests that the current pricing for computing capacity may be detached from long-term value.
Burry's concerns center on the pricing of short-term AI computing capacity, which he views as overvalued [3, 6]. He said that some short-term contracts could carry rates of $40 per unit [5], creating a risk for investors heavily exposed to the sector.
Despite Burry's outlook, Nebius reported strong growth in its second-quarter earnings released earlier this month [4, 5]. The company's revenue more than quintupled year-over-year [1]. This financial performance triggered a significant rally in the company's stock, though reports on the exact gain vary.
Some data indicates the stock surged over 34% in a single trading session following the results [1]. Other reports place the increase at 28% [4] or 30% [5]. This upward momentum has led some analysts to suggest that short-pressure from investors like Burry may actually be helping the rally by fueling a short squeeze.
Nebius continues to expand its infrastructure to meet the demand for AI processing power. However, Burry's position serves as a warning that the rapid ascent of AI-focused stocks may be reaching a tipping point. The conflict between the company's actual revenue growth and Burry's bearish thesis reflects a broader debate on Wall Street over whether AI infrastructure is a bubble or a fundamental shift in computing [6].
“Burry's bet suggests that the current pricing for computing capacity may be detached from long-term value.”
The tension between Nebius's explosive revenue growth and Michael Burry's short position illustrates the current volatility in the AI sector. While fundamental growth is evident in the earnings reports, the disagreement over pricing models for computing capacity suggests a looming correction if demand fails to justify current premium rates.
