"It's Nonsense Being Pushed by Short Sellers." - CoreWeave CEO on the GPU Depreciation Debate

In this episode, Andrew Huberman discusses the controversy surrounding GPU depreciation claims and their impact on AI infrastructure companies like CoreWeave. The conversation centers on allegations from short sellers that GPU values are declining rapidly, creating uncertainty in the market and affecting investor confidence in AI infrastructure plays. The CoreWeave CEO addresses these claims directly, providing counterarguments grounded in market data and operational metrics. The fundamental argument presented is that GPU depreciation narratives are being weaponized by investors with financial incentives to drive stock prices down, while the actual market dynamics tell a different story. The episode explores how the AI boom has created enormous demand for GPUs across enterprise clients, data centers, and cloud computing platforms. This demand, according to the discussion, far outpaces the available supply, creating a healthy market dynamic that contradicts depreciation narratives. The conversation delves into the semiconductor supply chain improvements that have occurred since the chip shortage crisis, examining how manufacturing capacity has expanded and distribution networks have stabilized. These structural improvements provide better visibility into future GPU availability and pricing stability. The CEO presents specific data on GPU utilization rates across CoreWeave's customer base, showing sustained high demand and pricing stability. These metrics suggest that enterprise customers view GPU infrastructure as a critical capital expenditure rather than a speculative asset. The episode also addresses the broader context of how misinformation and coordinated narratives can distort market perception. The discussion examines the incentive structures that encourage short sellers to disseminate negative narratives, regardless of underlying truth. This connects to broader themes about information quality, market efficiency, and the role of science-based analysis in countering misleading claims. Huberman explores how the AI infrastructure market is still in early stages of maturation, with many use cases not yet fully developed. This suggests that current demand metrics may actually underestimate future growth potential. The conversation includes specific examples of enterprise adoption patterns, customer acquisition costs, and retention rates that demonstrate market strength. The episode concludes by emphasizing the importance of rigorous, data-driven analysis when evaluating emerging technology sectors. Rather than accepting narratives from either bullish or bearish camps uncritically, Huberman advocates for examining first principles, customer behavior, and fundamental economic metrics.