Michael Burry's Depreciation Thesis Against NVIDIA: Accounting Practices and Market Impact Analysis

This analysis is based on the Reddit discussion [1] published on November 14, 2025, which examined Michael Burry’s short thesis against NVIDIA (NVDA) centered on depreciation accounting practices for AI chips. The core issue involves whether major cloud providers are using overly optimistic depreciation schedules that artificially inflate profitability, potentially masking underlying financial vulnerabilities that could impact NVIDIA’s future demand.

Michael Burry, renowned for his prescient “Big Short” analysis, has disclosed put options against NVIDIA worth approximately $187 million as of September 30, 2025 [2]. His thesis centers on a fundamental accounting discrepancy: major cloud providers including Meta, Oracle, Microsoft, Google, and Amazon are depreciating AI chips over 5-6 year periods when the actual economic useful life may be only 2-3 years due to rapid technological obsolescence [2][3].

The financial implications are substantial. Burry estimates this practice could understate depreciation by approximately $176 billion between 2026-2028, creating significant earnings inflation across the AI sector [2]. Specifically, he projects Oracle could overstate earnings by 26.9% and Meta by 20.8% by 2028 under current assumptions [2].

The AI infrastructure ecosystem shows significant disagreement on equipment longevity:

• Conservative Position
  : Burry argues for 2-3 year useful life based on accelerated obsolescence patterns [3]
• Moderate Adjustment
  : Amazon reduced server useful life from 6 to 5 years in February 2025, citing “increased pace of technology development” [3]
• Extended View
  : Microsoft, Google, and Oracle maintain up to 6-year depreciation cycles for AI equipment [3]

NVIDIA’s current market position reflects both the AI boom and these underlying tensions. The stock trades at $190.17 with a $4.63 trillion market capitalization [0], but has shown 2.53% decline over the past 5 days amid this debate [0].

NVIDIA’s financial performance presents a mixed picture:

• Explosive data center revenue growth from $15 billion to $115 billion in fiscal year ending January 2025 [3]
• Strong profitability with 52.41% net profit margin and 58.09% operating margin [0]
• Analyst consensus of 73.4% Buy ratings with $235 price target (23.6% upside) [0]

• High valuation multiples: P/E ratio of 53.51x and P/B ratio of 46.28x [0]
• Extreme concentration: 88.3% of revenue from data center segment [0]
• Accelerated product cycles: Annual AI chip releases versus previous 2-3 year cadence [3]

CoreWeave, an NVIDIA-backed cloud provider, reports strong secondary market demand for older chips (95% of original price for 2022 H100s) [3]. However, this may not represent broader market conditions, as CoreWeave itself uses 6-year depreciation cycles and has seen its shares fall 57% from June highs [3].

The shift from 2-3 year to annual AI chip product cycles creates unprecedented depreciation challenges. This acceleration compresses the economic useful life of equipment far faster than traditional accounting schedules accommodate, potentially creating systematic misalignment between reported and actual asset values [3].

The controversy highlights a critical divergence: even if AI chips remain physically functional, technological obsolescence can destroy economic value much faster than accounting depreciation suggests. This creates potential cash flow squeezes for hyperscalers who may need to replace equipment more frequently than their accounting schedules reflect [2][3].

1. Regulatory Scrutiny Risk
   : If accounting practices are deemed aggressive, regulatory intervention could force restatements across the AI sector
2. Customer Cash Flow Vulnerability
   : Hyperscalers facing unexpected write-downs may reduce NVIDIA orders to preserve cash liquidity
3. Demand Sustainability Question
   : The entire AI capital expenditure boom could be vulnerable if depreciation assumptions prove overly optimistic
4. Valuation Compression Risk
   : Current high multiples could face pressure if earnings growth slows due to accounting adjustments

1. Upcoming Earnings Catalyst
   : NVIDIA’s Q3 results on November 19, 2025, will provide crucial insights into AI demand sustainability [0]
2. Accounting Transparency
   : Companies adopting more conservative depreciation practices may gain investor confidence
3. Secondary Market Development
   : Robust used GPU markets could provide economic offset for rapid obsolescence

1. Hyperscaler Accounting Changes
   : Watch for depreciation policy adjustments from major customers
2. Auditor Commentary
   : Monitor for auditor qualifications or concerns in company filings
3. Secondary Market Pricing
   : Track used GPU values as indicators of actual equipment longevity
4. Technological Pace Indicators
   : Assess whether annual product cycles sustainable or represent temporary acceleration

The depreciation debate centers on whether major AI infrastructure providers are using realistic equipment life assumptions. If Burry’s thesis proves correct, the $176 billion estimated depreciation understatement could trigger significant earnings adjustments across the sector [2]. This would likely impact NVIDIA through reduced capital expenditures as customers adjust to more realistic replacement cycles.

However, the situation remains complex. The lack of industry consensus on GPU useful life, combined with strong secondary market demand for older chips, suggests the actual economic impact may vary significantly across different use cases and customer segments [3].

NVIDIA’s fundamental business strength, demonstrated by explosive revenue growth and strong profitability, provides some buffer against these concerns [0]. Nevertheless, the high valuation multiples and extreme revenue concentration create vulnerability to any slowdown in AI infrastructure spending [0].

The resolution of this accounting question will be crucial for properly valuing not just NVIDIA, but the entire AI infrastructure ecosystem. Decision-makers should monitor upcoming earnings reports, accounting policy changes, and secondary market developments for clues about the trajectory of this fundamental debate.