Michael Burry's NVIDIA Short Thesis: Depreciation Accounting Risks in AI Infrastructure

This analysis is based on multiple reports regarding Michael Burry’s bearish position against NVIDIA (NVDA) and concerns about AI infrastructure accounting practices, with the original discussion originating from a Reddit post questioning depreciation’s impact on profitability statements [1][2][3][4].

Michael Burry, renowned for predicting the 2008 financial crisis, has disclosed significant put option positions against NVIDIA valued at approximately $187 million as of September 30, 2025 [1][2]. His thesis centers on allegations that major tech companies, particularly NVIDIA’s hyperscaler customers, are artificially inflating earnings by extending the useful life of AI computing equipment beyond realistic economic lifespans. The core concern is that AI chips and servers actually become obsolete in 2-3 years due to rapid technological advancement, but companies are using 6-year depreciation cycles, thereby understating depreciation expenses and overstating near-term profitability [1][2].

NVIDIA’s stock has experienced significant volatility following these concerns. As of November 14, 2025, NVDA is trading at $186.86, down 3.58% on the day, with a 52-week range of $86.62-$212.19 [0]. The decline aligns with broader Technology sector weakness, which is down 1.57% today, suggesting that Burry’s specific allegations may be exacerbating existing market concerns about tech valuations [0].

Despite this bearish thesis, analyst consensus remains overwhelmingly bullish on NVIDIA:

• Overall Consensus
  : BUY with 73.4% of analysts rating it a Buy [0]
• Price Target
  : $235.00 consensus target, representing 25.8% upside potential [0]
• Target Range
  : $200.00-$350.00 [0]

Recent analyst actions have been consistently positive, with major firms including Citigroup, Loop Capital, Rosenblatt, and Goldman Sachs all maintaining Buy ratings in November 2025 [0].

NVIDIA’s current valuation metrics reveal some concerns that align with Burry’s bubble thesis:

• P/E Ratio
  : 52.58x (elevated relative to historical averages) [0]
• P/B Ratio
  : 45.47x (suggests premium valuation) [0]
• EV/OCF
  : 59.04x (high multiple on cash flow) [0]

However, profitability metrics remain strong with a Net Profit Margin of 52.41% and Operating Margin of 58.09% [0]. The relatively low ROE of 1.05% given these high margins warrants attention [0].

NVIDIA’s revenue breakdown reveals significant vulnerability through concentration in the Data Center segment, which accounts for 88.3% of FY2025 revenue ($115.19B) [0]. This makes NVIDIA particularly susceptible to any slowdown in AI infrastructure spending by hyperscalers, which are precisely the companies Burry accuses of accounting irregularities.

Geographically, China represents 13.1% of NVIDIA’s revenue ($17.11B) [0], creating additional vulnerability given ongoing export restrictions and recent reports of Chinese companies accessing banned NVIDIA chips through Indonesian cloud providers [1].

Burry’s analysis estimates that aggressive depreciation assumptions could lead to $176 billion in overstated profits across major tech companies from 2026-2028 [1][2]. His specific projections suggest Oracle could overstate earnings by 26.9% and Meta by 20.8% by 2028 [1][2]. This accounting practice, if accurate, creates a fundamental disconnect between reported profitability and actual cash generation capabilities.

Burry’s track record adds significant weight to his concerns - he successfully predicted the 2008 housing crisis and has been warning about market bubbles for several years [2]. However, his timing has historically been early, and markets have often continued rising despite his warnings. Notably, other prominent short-sellers like Jim Chanos have expressed similar concerns about AI infrastructure spending [2], while major banks including Morgan Stanley and Bank of America have warned about future depreciation expense impacts on tech profitability [1].

Several crucial information gaps remain:

1. Depreciation Policy Analysis
   : Detailed comparison of NVIDIA’s own depreciation policies versus customer practices
2. Hyperscaler Spending Patterns
   : Real-time data on whether major customers are accelerating replacement cycles
3. AI Chip Obsolescence Rate
   : Independent verification of actual equipment lifespans versus claimed depreciation cycles
4. Burry’s Detailed Methodology
   : He promised “more detail coming November 25th” [2], which could provide crucial insights

1. Accounting Risk
   : If Burry’s allegations about overstated earnings prove accurate, it could trigger a massive revaluation of AI stocks and significantly reduce customer spending power [1][2]. The potential $176 billion in overstated profits across the tech sector represents a systemic risk.

2. Concentration Risk
   : NVIDIA’s 88.3% reliance on Data Center revenue makes it extremely vulnerable to any slowdown in AI infrastructure investment [0]. A correction in spending patterns could have outsized impacts on revenue.

3. Valuation Risk
   : Current P/E ratio of 52.58x suggests high expectations that may be difficult to maintain if growth slows or accounting issues surface [0].

4. Regulatory Risk
   : Ongoing export restrictions and compliance issues, particularly regarding Chinese market access, add another layer of uncertainty [1].

1. November 25 Disclosure
   : Burry’s promised detailed analysis could provide crucial methodology and calculations [2]
2. Q3 Earnings (November 19)
   : Customer commentary on AI infrastructure spending cycles will be critical
3. Hyperscaler Accounting Changes
   : Any announcements of revised depreciation policies by major customers
4. AI Chip Replacement Data
   : Independent verification of actual equipment lifespans in the field
5. Institutional Positioning
   : Monitor 13F filings for major institutional changes following these concerns

The debate over AI infrastructure accounting practices highlights a fundamental tension between rapid technological advancement and traditional accounting methods. While NVIDIA continues to show strong operational performance with 52.41% net margins [0], the sustainability of current spending patterns depends heavily on the accuracy of depreciation assumptions used by its major customers.

The market appears to be pricing in continued growth, with analysts maintaining bullish positions and price targets suggesting 25.8% upside potential [0]. However, Burry’s thesis introduces a significant counter-narrative that questions the fundamental accounting assumptions supporting AI infrastructure investment.

The convergence of concerns from multiple respected sources (Burry, Chanos, major banks) suggests this issue deserves serious consideration, particularly given NVIDIA’s high valuation multiples and revenue concentration [0][1][2]. The upcoming earnings release on November 19 and Burry’s detailed analysis on November 25 will be critical events for assessing the validity of these concerns.

[0] Ginlix Analytical Database - NVIDIA market data, financial metrics, and analyst consensus
[1] CNBC - “Big Short investor Michael Burry accuses AI hyperscalers of artificially boosting earnings” (November 11, 2025)
[2] Business Insider - “Michael Burry’s AI Warning Sounds Similar to Short-Seller Jim Chanos” (November 2025)
[3] Futunn News - “The Big Short switches to combat mode! Burry: Tech giants…” (November 2025)
[4] Saxo Bank - “The Big Short: Is Michael Burry right about the AI trade?” (November 12, 2025)