Instead of focusing only on the latest NVIDIA H100 chips, analysts should watch the rental rates for older A100s. Their steady and rising prices indicate that demand for AI inference is so strong that even previous-generation hardware is being fully utilized as a 'workhorse' for a growing number of less complex tasks.

CoreWeave dismisses speculative analyst reports on GPU depreciation. Their metric for an asset's true value is the willingness of sophisticated buyers (hyperscalers, AI labs) to sign multi-year contracts for it. This real-world commitment is a more reliable indicator of long-term economic utility than any external model.

AI software is improving so rapidly that older hardware, like a three-year-old NVIDIA inference chip, is now more profitable than it was when new. This phenomenon, where software advancements outpace hardware depreciation, is unprecedented and makes existing infrastructure increasingly valuable.

Despite the rapid pace of hardware innovation, the value of older NVIDIA GPUs like the H100 is holding strong. Cloud provider CoreWeave reports these chips are retaining 90-95% of their pricing power over a 5-6 year lifespan because compute demand far outstrips supply.

Despite the rollout of Blackwell and the announcement of Vera Rubin, cloud provider Nebius reports that pricing for older Hopper GPUs is not dropping, and in some cases is rising. This shows a persistent market for "good enough" compute for specific workloads.

Andreessen highlights a unique economic phenomenon: the pace of AI software improvement outstrips hardware depreciation. This means a three-year-old NVIDIA inference chip can generate more revenue today than when it was new, a complete reversal of typical tech hardware value cycles.

Contrary to the belief that AI chips quickly become obsolete, CoreWeave's CEO argues their value holds, citing average five-year client contracts as proof. Older chips like the A100 have even appreciated in price as new use cases emerge, making rapid depreciation a myth.

Contrary to the belief that AI hardware becomes obsolete quickly, older GPUs like A100s will have a long depreciable life. As companies optimize costs, they'll use model routing to send simple queries to older, cheaper hardware, extending its utility for six to eight years.

Countering the narrative of rapid burnout, CoreWeave cites historical data showing a nearly 10-year service life for older NVIDIA GPUs (K80) in major clouds. Older chips remain valuable for less intensive tasks, creating a tiered system where new chips handle frontier models and older ones serve established workloads.