Nvidia's Older AI Chips Retain Value, Challenging Michael Burry’s Depreciation Concerns
Data from Silicon Data shows that six-year-old Nvidia A100 chips retain a residual value of nearly $5,000 each. This challenges concerns raised by investor Michael Burry about Big Tech potentially overstating profits through extended depreciation schedules for AI hardware. The sustained rental income offsets the typical decline in asset value over time. Burry previously warned that hyperscalers might understate depreciation expenses by up to $176 billion between 2026 and 2028, assuming longer useful lives of computing equipment. However, companies like CoreWeave have signed contracts using A100 GPUs through 2029, indicating ongoing economic utility for older chips in tasks such as inference and fine-tuning. The debate is significant given the growing capital investment into AI infrastructure by Wall Street firms partnering with Nvidia. Kalshi traders also suggest rental prices for these older GPU models are expected to remain stable or even increase through 2027, supporting their continued value.
Topics: AI hardware Big Tech profits depreciation accounting Michael Burry Nvidia GPUs