Short-seller Michael Burry publicly shorted Nebius at $211.77, calling AI infrastructure investments 'like shooting fish in a barrel' for losses.

Michael Burry, the contrarian investor famous for his bet against the housing market before the 2008 financial crisis, has turned his skepticism toward the artificial intelligence infrastructure boom. In an August 6 post on his Substack, Cassandra Unchained, Burry disclosed new short positions in Nebius Group (NBIS) and Oracle (ORCL), describing the trades as "a bit like shooting fish in a barrel."

Burry shorted Nebius at $211.77 per share and Oracle at $144.63 per share, without disclosing position sizes. The timing is particularly pointed for Nebius, scheduled to report second-quarter earnings on August 12, creating a near-term catalyst that could quickly validate or challenge his bearish thesis.

Burry opted for a direct equity short on Nebius rather than put options, noting that implied volatility on Nebius puts had surged above 100%, making them prohibitively expensive. His Oracle short marks a swift return to bearish positioning; he had closed a previous Oracle put position just days earlier on August 5, booking what he called a "substantial" profit, and had signaled he might re-enter if volatility subsided.

At the core of Burry's argument lies concern about GPU depreciation. AI infrastructure companies are spending billions on Nvidia-powered server clusters, then spreading those costs over several years through depreciation schedules. Burry contends that AI hardware becomes obsolete far faster than companies assume because newer, more powerful chips release annually. If GPUs must be replaced sooner than expected, companies could face accelerated write-downs that reduce future profits even if revenue grows. He estimates that Oracle's earnings could be overstated by roughly 26% to 27% by 2028 if depreciation assumptions prove too optimistic. Burry also highlights that numerous companies maintain long-term lease commitments significantly exceeding current revenue levels, creating off-balance-sheet risks the market may be underappreciating.

His skepticism targets a broader pattern: companies borrowing heavily to build AI infrastructure on unproven long-term demand assumptions. He maintains short positions in Nvidia (NVDA), Micron Technology (MU), and Palantir Technologies (PLTR). His thesis does not represent a blanket bet against artificial intelligence itself.

Nebius has emerged as one of the market's biggest AI winners, posting 684% revenue growth and reaching $1.9 billion in annualized recurring revenue. However, its long-term debt doubled to $8.4 billion, and capital expenditure guidance climbed to $25 billion. The company recently disclosed approximately $27 billion in AI compute contracts with Meta Platforms (META), underscoring strong demand for GPU capacity. Shares remain up 137% year to date and roughly 950% over the past three years, though they've pulled back to around $198.65.

Not everyone shares Burry's pessimism. Many analysts argue that AI infrastructure remains supply-constrained, with demand from hyperscalers such as Microsoft (MSFT), Meta, and OpenAI continuing to outpace available GPU capacity. They believe long-term AI contracts should help companies earn attractive returns despite rapid technological advances. Analyst consensus rates Oracle a Strong Buy with 80% upside potential over the next 12 months, while Nebius carries a Moderate Buy rating with nearly 31% upside.

Burry's conviction extends to Palantir, where he has maintained his short position despite the company's blowout second-quarter results. Palantir reported revenue of $1.94 billion, a 93% year-over-year increase that easily beat Wall Street's $1.80 billion estimate. Adjusted earnings of 41 cents per share topped the 35-cent consensus. US commercial revenue surged 149% to $764 million, while US government revenue climbed 90% to $809 million. The company raised its full-year revenue guidance to $8.150 billion to $8.158 billion. The stock jumped roughly 30% on the news, but Burry held firm. Analysts at Jefferies have echoed caution; analyst Brent Thill warned that current valuation leaves minimal cushion for any deceleration in growth or operational missteps, arguing the risk-reward profile appears unattractive at current levels. Burry's short thesis does not require Palantir's revenue to decline—it simply needs growth rates or investor sentiment to moderate enough for the valuation multiple to contract.

While betting against select AI names, Burry has been adding to several long positions following recent earnings. He increased his Flutter Entertainment (FLUT) stake to a full position size. He added to Fiserv (FISV) as shares dipped to $52, compared to his average entry price of $48, characterizing quarterly results as less meaningful given the company's early-stage turnaround under new leadership. On Zoetis (ZTS), Burry acknowledged softening economic conditions weighing on North American pet spending but maintained that the animal health company remains fundamentally strong. He retained his Mercado Libre (MELI) position with shares at $1,782, above his average cost of $1,611, indicating he would add more if prices declined toward $1,500. He also addressed concerns over rapid growth in Mercado Libre's credit segment, noting the company's deliberate strategy of prioritizing long-term scale over immediate profits.

Burry exited his bullish Microsoft (MSFT) position during the same period, demonstrating willingness to pivot when market dynamics shift. Since closing Scion Asset Management last year, investors no longer receive quarterly SEC portfolio disclosures. The only trades Burry voluntarily reveals come through his Substack, making these disclosures the primary window into the thinking of one of Wall Street's most closely watched contrarians.