AI Investment Risk Analysis: Depreciation Accounting vs. Growth Capex Returns

This analysis examines the emerging debate in AI investment risk assessment, contrasting Michael Burry’s focus on depreciation accounting practices with an alternative perspective emphasizing growth capital expenditure returns. The discussion, originating from a Reddit post on November 14, 2025, challenges the conventional understanding of AI sector risks [Event timestamp: 2025-11-14 15:47:51 UTC].

Michael Burry, renowned for his prescient market warnings, has accused major AI hyperscalers of artificially inflating earnings through extended depreciation schedules. His analysis, published on November 11, 2025, alleges that companies like Meta and Oracle have stretched AI hardware useful life assumptions from 2-3 years to 5-6 years, potentially understating depreciation by $176 billion between 2026-2028 [1]. This could result in earnings inflation of 27% for Oracle and 21% for Meta by 2028 [1].

The Reddit author presents a compelling alternative view: depreciation is fundamentally an accounting timing device that doesn’t affect immediate cash flow [Event content]. The real investment risk, they argue, lies in the uncertain return on massive growth capital expenditures. This perspective shifts focus from accounting technicalities to economic substance - whether AI infrastructure investments will generate adequate returns before hardware obsolescence or monetization challenges materialize.

The AI sector has shown mixed performance following these concerns:

• NVIDIA (NVDA)
  : $190.17 (+1.77%), showing resilience with a P/E ratio of 54.18x
• Meta (META)
  : $609.46 (-0.07%), relatively stable despite being specifically called out, with a 26.98x P/E and 30.89% net profit margin
• Oracle (ORCL)
  : $222.85 (+2.43%), indicating continued investor confidence
• Microsoft (MSFT)
  : $510.18 (+1.37%), positive performance
• Alphabet (GOOGL)
  : $276.41 (-0.78%), slight decline

Institutional investors are taking divergent positions. Tiger Global slashed its Meta stake by 63% in Q3 2025 [4], while Goldman Sachs increased stakes in NVIDIA, Apple, and Microsoft [4]. Notably, SoftBank liquidated $5.8 billion in NVIDIA to double down on OpenAI and Project Stargate [2].

The debate between depreciation concerns and capex return risks represents complementary perspectives on the same fundamental issue: the sustainability of current AI investment levels. Whether through accounting timing or economic returns, both analyses question whether current AI spending patterns are justified by underlying business fundamentals.

Current AI valuations appear to reflect both accounting and economic uncertainties. The lofty valuations suggest markets are pricing in significant growth expectations, while the mixed performance indicates investor caution about realization risks. Meta’s 15.06% decline over the past month [0] demonstrates this tension.

The stark contrast in institutional positioning - with some reducing exposure while others increase it - highlights the fundamental disagreement about AI investment risk assessment. This divergence often precedes significant market movements as consensus forms.

Some analysts [2] argue that AI’s strategic importance and government support create a “too big to fail” dynamic that could sustain investments despite near-term economic concerns. This geopolitical dimension adds complexity to traditional investment analysis.

1. Regulatory Scrutiny Risk
   : Burry’s allegations could trigger SEC investigations into accounting practices, potentially leading to restatements or penalties [1].

2. Hardware Obsolescence Acceleration
   : Rapid AI advancement could render current infrastructure investments obsolete faster than projected 5-6 year lifespans, creating a “AI hardware hangover” [3].

3. Monetization Timeline Risk
   : Extended delays in AI revenue generation could strain cash flows, particularly for companies with massive capex commitments [Event content].

4. Capex Sustainability Concerns
   : The collective $349 billion annual spending rate from major tech firms raises questions about long-term ROI without clear monetization paths [3].

1. Clarification Catalyst
   : Burry’s promised “more detail” on November 25th [1] could provide additional insights and trigger market movements based on the substance of his allegations.

2. Technical Validation Opportunities
   : Independent verification of actual AI hardware lifespan and performance could resolve the depreciation debate and provide clarity for investment decisions.

3. Monetization Breakthroughs
   : Successful AI service launches or unexpected revenue acceleration could justify current investment levels and drive valuation expansion.

• Burry’s detailed analysis release on November 25th [1]
• Q4 2025 earnings reports from AI hyperscalers
• Any regulatory responses to depreciation allegations
• Institutional position changes in 13F filings

• Actual AI hardware refresh cycles and obsolescence rates
• AI service monetization progress and cash flow generation
• Free cash flow vs. capex spending ratios across major AI companies

• Realized ROI on current AI infrastructure investments
• Competitive dynamics in AI chip manufacturing and supply chain
• Government policy impact on AI development and deployment

The debate between Michael Burry’s depreciation critique and the growth capex return risk analysis highlights fundamental uncertainties in AI investment assessment. While accounting practices may represent “optics” as the Reddit author suggests, the underlying economic reality of massive capital expenditures requiring adequate returns remains the critical investment consideration.

Current market valuations reflect this tension, with mixed performance across AI stocks and divergent institutional positioning. The resolution of this debate will likely depend on several factors: actual hardware lifespan verification, AI monetization timeline achievement, and regulatory responses to accounting practices.

Investors should maintain vigilance on both accounting transparency and economic returns from AI investments, recognizing that both perspectives ultimately address the same fundamental question: whether current AI investment levels will generate sustainable long-term value. The answer will significantly determine the trajectory of AI sector valuations in the coming months and years.