Prominent investor Michael Burry, known for predicting the 2008 housing crisis, has taken substantial short positions against Nvidia through put options, warning that the AI boom mirrors the dot-com bubble with Nvidia playing the role of Cisco as the central hardware supplier.

Burry’s thesis centers on several key risks. He argues hyperscalers are overbuilding infrastructure, potentially leading to unused capacity similar to the fiber optic glut in the early 2000s. Extended depreciation schedules for Nvidia chips inflate short-term earnings but could trigger massive writedowns later. He estimates industry-wide profit inflation of billions through 2028 from these practices. Circular financing—deals where Nvidia invests in customers who then buy its chips—props up demand artificially.

Nvidia pushed back strongly, circulating a memo to analysts rebutting Burry’s claims point by point. Executives emphasize genuine demand, long chip lifespans enabled by CUDA software, and minimal circular investments relative to revenue. CEO Jensen Huang dismissed bubble talk, highlighting sold-out Blackwell chips and visibility into trillions in future AI infrastructure spending.

Despite Nvidia’s record results—third-quarter revenue hitting $57 billion, up 62% year-over-year, with data center sales dominating—shares faced pressure in late 2025. Volatility stemmed from broader debates over whether massive capex by Microsoft, Amazon, Meta, and Google will yield sufficient returns. Analysts project hyperscaler spending exceeding $350 billion in 2025, rising further, but some warn of sustainability if AI monetization lags.

Other investors echoed caution. Peter Thiel and Masayoshi Son exited Nvidia positions entirely. Short interest rose as competition from custom chips and alternatives grew. Yet bulls point to Nvidia’s dominance, accelerating earnings forecasts—analysts see over 60% EPS growth in 2026—and projections of $3-4 trillion annual industry AI spend by 2030.

Market concentration remains extreme, with Nvidia powering most advanced models. Historical bubbles burst when supply outstripped demand, but proponents argue AI’s transformative potential justifies investment, unlike past hype cycles.

Conservative investors monitor returns on deployed capital closely. If hyperscalers demonstrate revenue growth from AI services matching capex, concerns fade. Otherwise, corrections could deepen.

Nvidia’s position offers enormous upside if demand holds, but risks substantial downside in an overbuild scenario. Prudent allocation favors diversification beyond single-stock bets in heated sectors.

As 2026 approaches, earnings from customers will clarify whether this cycle sustains or falters.