AnalysisEarnings QualityDepreciation AnalysisAI Infrastructure
Part of the Earnings Quality Analysis Hub series

The Depreciation Divergence: Why Amazon Is Big Tech's Earnings Quality Play

Michael Burry warns Big Tech will understate depreciation by $176B through 2028. Amazon is taking a contrarian approach—shortening useful lives and accepting a $700M profit hit while peers extend to boost earnings. Learn how to screen for earnings quality using the capex/depreciation ratio.

10 min read

Key Finding: Amazon is the ONLY hyperscaler to shorten depreciation useful lives—accepting a $700M profit hit while Meta, Google, and Microsoft extended useful lives to boost earnings by a combined $8.8B annually. This creates a stark divergence in earnings quality.

In November 2025, Michael Burry broke a two-year silence to warn about what he called "one of the most common frauds in the modern era"—Big Tech's depreciation manipulation.

His thesis: hyperscalers will understate depreciation by $176 billion between 2026-2028 by extending asset useful lives from 4-5 years to 6-10 years. The effect? Inflated profits that mask the true cost of the AI infrastructure arms race.

But here's what Burry's thesis misses: not all hyperscalers are playing the same game.

Amazon is doing the exact opposite—shortening depreciation schedules and taking a $700M annual profit hit while competitors extend useful lives to boost earnings. This creates a powerful screen for earnings quality investors.


The Depreciation Divergence Explained

When a company buys AI infrastructure (GPUs, servers, data centers), it doesn't expense the cost immediately. Instead, it depreciates the asset over an assumed "useful life":

  • 5-year useful life: Depreciate 20% per year ($100B capex → $20B annual depreciation)
  • 10-year useful life: Depreciate 10% per year ($100B capex → $10B annual depreciation)

By extending useful lives, companies can cut depreciation expense in half—directly boosting reported earnings and margins. The accounting is legal, but the question is whether it reflects economic reality.

Here's how the four hyperscalers diverged in their depreciation policies:

CompanyPreviousNewEffective DateAnnual P&L Impact
Amazon6 years5 yearsJan 2025-$700M profit
Meta4.5 years5.5 yearsJan 2025+$2.9B profit
Alphabet4 years6 yearsJan 2023+$3.4B profit
Microsoft4 years6 years2022+$2.5B profit

The divergence is stark: Amazon is accepting lower reported profits to accelerate depreciation, while Meta, Google, and Microsoft are boosting profits by $8.8B combined through extensions.


Amazon: The Contrarian Signal

Amazon's decision to shorten useful lives from 6 to 5 years is remarkable for two reasons:

1. It's the Opposite of Peers

Every other hyperscaler extended useful lives to boost profits. Amazon went the other direction—the only company in the group to accept a profit reduction from depreciation policy changes.

This is not what you'd expect from a company trying to inflate earnings. It's the opposite.

2. It Creates a Conservative Accounting Signal

The $700M annual profit hit from shortened depreciation means Amazon's reported earnings are understated relative to peers. When you compare Amazon's operating margins to Meta's or Google's, you're not comparing apples to apples—Amazon is front-loading costs while peers defer them.

What this might signal:

  • Conservative management prioritizing earnings quality over short-term optics
  • Genuine belief that AI infrastructure will become obsolete faster than peers assume (Nvidia releases new GPU architectures every 1-2 years)
  • Preparation for a future where depreciation extensions become untenable

The Capex/Depreciation Ratio Tells the Story

The capex/depreciation ratio reveals which companies are converting infrastructure spending into depreciation expense—and which are deferring costs to the future.

CompanyQ3 2025 Ratio8Q AverageSignal
Amazon1.7x1.74xMost Conservative
Microsoft2.1x1.89xConservative
Meta3.5x2.91xRed Flag
Alphabet4.3x3.76xWorst in Group

Amazon's ratio is 2.5x lower than Google's. For every dollar Google spends on capex, it recognizes only $0.23 in depreciation. Amazon recognizes $0.59. That's a massive difference in earnings quality.


Meta: The Double Whammy

Meta presents the most concerning case in the depreciation divergence.

Extending Depreciation While Increasing Intensity

Meta didn't just extend useful lives—it did so while dramatically increasing capex intensity:

  • Capex intensity: 38% of revenue (highest in group)
  • Capex growth: +111% YoY
  • Revenue growth: +26% YoY
  • Gap: 85 percentage points

The 85pp gap between capex and revenue growth is the largest in the group. Combined with extended depreciation schedules, this creates compounding risk:

  1. Current profits inflated: The $2.9B annual boost from extended depreciation masks true infrastructure costs
  2. Future depreciation balloon: When assets are eventually replaced, catch-up depreciation hits the income statement
  3. Intensity keeps rising: The December $600B commitment through 2028 escalates capex intensity toward 50%

The Burry Math Applied to Meta

Michael Burry estimated Meta's profit inflation at 20.8%. Here's how that works:

  • Q3 2025 operating income: ~$20B
  • Estimated understated depreciation: ~$4.2B (20.8% of operating income)
  • True operating income: ~$15.8B

If Burry's thesis is correct, investors are paying for $20B in earnings when the economic reality is closer to $16B. That's a material valuation difference.


Michael Burry's $176B Thesis Breakdown

In November 2025, Burry shared his analysis of Big Tech depreciation practices:

Core claims:

  • Big Tech will understate depreciation by $176 billion between 2026-2028
  • Profit inflation estimates: Oracle 26.9%, Meta 20.8%
  • Called it "one of the most common frauds in the modern era"

The mechanism:

  • Assets purchased at current capex rates ($400B+ annually across hyperscalers)
  • Depreciated over extended useful lives (6-10 years vs. historical 4-5 years)
  • Creates multi-year gap between economic cost and recognized expense

Why it matters for investors:

  • Reported earnings are higher than economic reality
  • P/E ratios appear lower than they actually are
  • When depreciation eventually catches up (asset retirements, policy changes), earnings compress

Companies Failing the Burry Screen

Using the capex/depreciation ratio >2.5x as a red flag threshold:

Failing (>2.5x sustained):

  • Alphabet: 4.3x ratio (worst)
  • Meta: 3.5x ratio (worsening)

Passing (<2.5x):

  • Amazon: 1.7x ratio (most conservative)
  • Microsoft: 2.1x ratio (acceptable)

How to Screen for Earnings Quality

The Capex/Depreciation Screen

Use this framework to identify companies at risk of earnings manipulation:

Ratio LevelSignalAction
1.0-1.5xConservativeLow manipulation risk
1.5-2.0xNormalMonitor for changes
2.0-2.5xWarningCheck depreciation policies
>2.5x sustainedRed FlagApply earnings discount

MetricDuck Screener Setup

  1. Go to metricduck.com/screener
  2. Add filter: capex / deprec_amort
  3. Set threshold: >= 2.5
  4. Set period type: Q.MED8 (8-quarter median filters volatility)
  5. Run screen

This identifies companies with sustained elevated ratios—the most concerning pattern for earnings quality.

Applying Earnings Discounts

For companies failing the Burry screen, consider applying valuation discounts:

RatioSuggested Discount
2.5-3.0x10-15%
3.0-3.5x15-20%
>3.5x20-25%

Example (Meta at 3.5x): If forward P/E is 20x, apply 20% discount → effective P/E is 25x. This accounts for potential earnings overstatement.


Investment Implications

Quality Rotation Strategy

For investors seeking AI infrastructure exposure with earnings quality focus:

Rotate FROM:

  • Alphabet (GOOGL): 4.3x depreciation ratio, worst earnings quality signals
  • Meta (META): 3.5x ratio + 38% capex intensity + $600B commitment

Rotate TO:

  • Amazon (AMZN): 1.7x ratio, shortened depreciation, most conservative accounting
  • Microsoft (MSFT): 2.1x ratio, supply-constrained demand validates capex

This maintains AI exposure while upgrading accounting quality.

For a comprehensive sector-adjusted analysis of all Magnificent 7 stocks using appropriate valuation frameworks for each business model, see our Magnificent 7 Valuation Scorecard.

What to Watch in Q4 2025 Earnings

Bullish signals (increasing confidence):

  • Any company crossing below 2.5x threshold
  • Depreciation policy changes toward shorter useful lives
  • Management addressing accounting practices in earnings calls

Bearish signals (decreasing confidence):

  • Ratios rising above 3.5x
  • Further useful life extensions announced
  • Capex acceleration without revenue guidance support

The Bottom Line

The depreciation divergence creates a clear framework for earnings quality:

CompanyDepreciation DirectionCapex/Deprec RatioEarnings Quality
Amazon⬇️ Shortened1.7xBest
Microsoft⬆️ Extended2.1xGood
Meta⬆️ Extended3.5xConcerning
Alphabet⬆️ Extended4.3xWorst

Amazon's contrarian move—shortening depreciation while peers extend—is a powerful signal for fundamental investors. It suggests management prioritizing long-term earnings quality over short-term optics.

For investors concerned about Burry's $176B depreciation thesis, the capex/depreciation ratio provides a simple screen to identify at-risk holdings and rotate toward quality.

For the complete quarterly monitoring framework, see our main guide: How to Track AI Capex Efficiency: 3-Metric Quarterly Monitoring Framework.


Screen for Earnings Quality

Use MetricDuck's screener to identify companies with elevated capex/depreciation ratios—a key signal of potential earnings manipulation. Pre-computed across 5,700+ companies with 8-quarter trend analysis.


Data sources: SEC EDGAR filings, company earnings calls, Michael Burry thesis analysis from Fortune (November 2025). All metrics verified against original SEC filings. Analysis current as of December 13, 2025.

Disclaimer: This analysis is for educational purposes only and does not constitute investment advice. The author may hold positions in mentioned securities. Perform your own due diligence or consult a financial advisor before making investment decisions.


Explore More Earnings Quality Analysis

This article is part of our comprehensive Earnings Quality Hub, which covers cash flow verification, accounting red flags, and quality screening frameworks.

Related earnings quality research:

MetricDuck Research Team

Financial data scientists and CFA charterholders with 40+ combined years analyzing SEC filings. All metrics sourced directly from SEC Edgar and validated against production data.