Renowned investor Michael Burry, who gained fame for his successful bet against the mortgage market ahead of the 2008 financial crisis, has once again drawn the attention of the expert community. This time, his sharp commentary was sparked by discussions surrounding the lifespan and economic value of Nvidia's graphics chips. The occasion was a recent presentation by the tech giant, which showcased data indicating that its accelerators retain economic value well beyond standard five-year accelerated depreciation schedules.
Background of the Presentation and Nvidia's Arguments
During the presentation, Nvidia featured a slide titled 'NVIDIA AI Infrastructure Retains Value Beyond Accelerated Depreciation Schedules,' comparing the residual value of advanced A100, H100, and cutting-edge B200 chips against standard write-off schedules. Company management and its supporters sought to prove that investments in AI infrastructure possess lasting durability, and that GPUs do not become obsolete as rapidly as skeptics claim. Corporate CEO Jensen Huang even described the computing infrastructure as a durable and highly rentable asset.
Historical Parallels: The 'Computer Bubble' from Adam Smith's Book
In response to these claims, Michael Burry published an analytical piece titled 'Don't Believe Your Lyin' Eyes, GPU Depreciation & Useful Lives.' In it, the investor referenced Adam Smith's famous 1968 book, 'The Money Game.' Burry drew a direct historical parallel to the computer leasing market of sixty years ago, when market participants assured each other of infinite demand for computing hardware and dismissed depreciation questions as 'the older generation's outdated view of a new market.' According to the investor, market psychology and human behavior remain unchanged despite rapid technological progress.
Contradictory Data
Opinions within the expert community regarding the durability of AI equipment are divided. On one hand, Nvidia relies on real market data showing preserved residual values of graphics accelerators and rising rental rates for three-year-old H100 chips. On the other hand, critics, including prominent investor Jim Chanos, raise a logical question: if GPU rentals are so profitable, why doesn't Nvidia run the rental business itself? The core debate today is not just about accounting depreciation schedules, but whether existing infrastructure can generate stable revenue amid the rapid emergence of new hardware generations.
Market Implications and Analyst Assessments
Comparing the current artificial intelligence boom to historical speculative bubbles is causing many institutional investors to rethink their portfolio strategies. Analysts note that while deep learning and generative AI technologies are indeed transforming the global economy, exaggerated expectations regarding infrastructure payback periods could lead to significant corrections. Michael Burry's assessment highlights the risks of overvaluing assets whose actual economic value may fall short of projections due to the pace of technological upgrades.
