In-Depth Analysis of Michael Burry's Investment Logic for Shorting AI Tech Stocks

Based on the latest market data and research materials, I will systematically analyze Michael Burry’s investment logic for shorting AI tech stocks and the impact of cloud computing capital expenditures on tech stock valuations.

---

According to the latest disclosures, Michael Burry (the prototype for The Big Short) has established short positions against multiple AI-related tech companies[1]. On January 9, 2026, he confirmed in a Substack post that he holds put options on Oracle, while admitting to directly shorting the company’s stock over the past six months[2]. As early as November 2025, Burry had disclosed put option positions on Nvidia and Palantir[3].

This is not Burry’s first time expressing a bearish stance on tech stocks. He previously shorted semiconductor stocks via ETFs, then closed the positions as chip stocks continued to rise. Now he is striking again, targeting core enterprises in the AI industry chain.

---

One of Burry’s core arguments is that the valuations of current AI-related stocks have seriously decoupled from fundamentals. Take the three targets he is shorting as examples:

Stock	Current P/E (TTM)	5-Year Average P/E	Valuation Deviation
Oracle (ORCL)	36.62x	~22x	+66%
Palantir (PLTR)	383.02x	~120x	+219%
Nvidia (NVDA)	45.33x	~35x	+29%

[0]

Palantir’s 383x P/E is particularly striking, meaning that at the current earnings level, it would take investors 383 years to recoup their investment. Even considering its high growth expectations, Wall Street predicts its compound annual earnings growth rate will be approximately 25% over the next five years[4], which still cannot support such extreme valuation. Burry pointed out that this valuation level requires the company to quadruple its earnings by 2030 to “live up to its price”, an assumption that is overly optimistic.

Burry’s second core argument is that the capital expenditure frenzy in the cloud computing sector has evolved into a systemic risk. According to the latest research from JPMorgan[5]:

• The current share of tech capital expenditure in GDP has
  surpassed the peak of the Dot-com era
• AI-related capital expenditures of the four tech giants (Amazon, Alphabet, Microsoft, Meta) are expected to
  exceed $350 billion
  in 2025
• Cloud computing infrastructure expenditure alone is expected to reach
  $611 billion
  in 2026, a 67% increase from previous estimates

1. Debt-Fueled Expansion
   : Oracle and Meta have funded data center construction through the debt market. JPMorgan points out that Oracle is financing its cloud computing infrastructure expansion via debt[5]

2. Sustained Negative Free Cash Flow
   : Oracle’s latest financial data shows its free cash flow is negative[0], indicating that the company’s capital expenditure pace outstrips its cash generation capacity

3. Current Ratio Imbalance
   : Oracle’s current ratio is only 0.91, below the 1.0 warning line[0], implying hidden risks in short-term solvency

Burry has questioned the “internal circulation” issue in the AI industry on multiple occasions. He points out:

• Nvidia invests in OpenAI
  → OpenAI uses the funds to purchase Nvidia chips
• Oracle invests billions in purchasing Nvidia hardware
  → OpenAI signs a
  $30 billion cloud computing agreement
  with Oracle
• Startups such as CoreWeave, Figure, XAI
  are invested in by AI companies like Nvidia, then become their customers

Burry analogizes this model to “selling shovels to oneself to dig for gold”, arguing that this circular dependency creates an artificial demand bubble rather than growth based on real commercial value.

Burry has sharply criticized the “useful life assumption” for AI hardware:

• Accelerated GPU iteration speed (from Hopper to Blackwell to Rubin architectures)
• Yet enterprises still use
  extended asset depreciation periods
• This contradiction between “accelerated hardware obsolescence vs. extended accounting treatment” is called by Burry “one of the most common fraud tactics in modern financial reporting”

---

Current tech stock valuations exhibit the following characteristics:

Indicator	Current Level	Historical Comparison
S&P 500 Forward P/E	~24x	Dot-com peak of ~45x
Tech Stock PEG Ratio	1x-3x	4x-8x during the Dot-com era
Free Cash Flow Yield	Below historical average	Under sustained pressure

: While technical valuation multiples have not yet reached the extreme levels of the Dot-com era, [5]. The capital expenditure intensity (Capex/Sales) of tech companies is currently around 30%, the historical normal level.

Taking Oracle as an example, analyze the impact of massive cloud computing capital expenditures on financial health[0]:

• ROE as high as 67.22%, indicating strong shareholder returns
• Net profit margin of 25.28%, with robust profitability
• 21% year-over-year revenue growth in 2025, with sufficient growth momentum

• Current ratio of 0.91 (below the 1.0 warning line)
• Debt risk rating:
  High Risk
• Conservative accounting policies (high depreciation/capital expenditure ratio)
• Sustained negative free cash flow

Oracle’s price-to-book (P/B) ratio is as high as 18.54x, far exceeding the software industry average. This reflects both the market’s growth premium for its cloud computing business and investors’ certain tolerance for debt expansion.

JPMorgan’s research shows[5]:

“In the first half of 2025, infrastructure expenditure accounted for 92% of GDP growth. Without data center and semiconductor expenditure, U.S. economic growth would have been only 0.1%.”

This data reveals two key issues:

1. Overreliance of Economic Growth on AI Infrastructure
   : If capital expenditure slows, it may trigger a chain reaction across the entire economy
2. Vulnerability of Valuations to the “Growth Narrative”
   : Current valuations heavily rely on the assumption of sustained high-speed growth in AI capital expenditures

---

Dimension	Burry’s View	Mainstream Institution Consensus
AI Valuation	Severe Bubble	Despite adjustment risks, fundamentals are robust
Nvidia Rating	Short	51 out of 54 analysts give a Buy rating[6]
Palantir Rating	Short	Consensus Rating: Hold (15/24 analysts)[7]
Oracle Rating	Short	Consensus Rating: Buy (60% of analysts)[8]

Notably, despite Burry’s bearish stance, mainstream institutions remain optimistic about core AI assets. Institutions such as UBS and Deutsche Bank have named Oracle their top pick for 2026, believing its AI infrastructure layout is undervalued[9].

The feedback from Nvidia’s latest CES press conference is also positive. Analysts expect its 2026 revenue to grow by approximately 65%, followed by another 50% growth in 2027[10]. Institutional investors have continued to net-buy Nvidia across all four quarters of 2025, with a buy/sell ratio of approximately 10:1[10].

---

• Extreme valuations do exist (especially for Palantir)
• Capital expenditure intensity has reached a historical high, with systemic risks accumulating
• Circular dependency patterns do exist; while not illegal, they amplify volatility
• Current ratio and free cash flow indicators show financial health pressure on some enterprises

• Unlike the Dot-com era, current AI enterprises have strong profitability (Nvidia’s net profit margin is 53%, Oracle’s is 25%)
• Enterprises like Nvidia have real cash flow and profit margin support
• Analysts’ earnings forecasts are still being revised upward, with no “expectation cliff” emerging yet
• Accelerated technological iteration does not necessarily mean value destruction; it may also bring about economies of scale

Scenario	Probability	Trigger Condition	Impact on Burry’s Short Positions
AI Bubble Burst	15%	Plummeting capital expenditure, collapse of earnings expectations	Substantial Profit
Valuation Mean Reversion	35%	P/E falls back to historical average	Moderate Profit
High-Range Volatility	40%	Tug-of-war between growth and valuation	Break-Even
Continued Rise	10%	Accelerated AI application implementation, capital expenditure exceeding expectations	Loss

---

Michael Burry’s core investment logic for shorting AI tech stocks is built on four pillars:

. From a data perspective, these arguments are not groundless—Palantir’s 383x P/E, Oracle’s negative free cash flow and current ratio below 1.0, and the record-high capital expenditure intensity across the tech industry all provide a factual basis for Burry’s concerns.

However, unlike shorting real estate in 2008, Burry is now facing an industry with real and strong profitability. While AI infrastructure construction comes with debt growth and financial risks, it also generates real computing capacity and potential commercial value.

For investors, Burry’s warning should be viewed more as a

rather than an explicit short-selling recommendation. The current market pricing of AI includes both reasonable expectations for future growth and excessive exuberance. Maintaining diversified allocations, focusing on valuation safety margins, and being wary of enterprises with high capital expenditure dependency may be prudent strategies to navigate this period of uncertainty.

---

[1] Bloomberg - “Investor Michael Burry Reveals Options Bet Against Oracle Shares” (2026-01-10)
https://www.bloomberg.com/news/articles/2026-01-10/investor-michael-burry-reveals-options-bet-against-oracle-shares

[2] Yahoo Finance - “Investor Michael Burry Reveals Options Bet Against Oracle” (2026-01-10)
https://finance.yahoo.com/news/investor-michael-burry-reveals-options-003350357.html

[3] LiveMint - “‘The Big Short’ fame investor Michael Burry remains bearish on artificial intelligence” (2026-01-10)
https://www.livemint.com/market/stock-market-news/michael-burry-big-short-fame-investor-bearish-ai-put-option-bets-against-major-us-tech-stock-oracle-larry-ellison-why-11768013217840.html

[4] Bravo’s Research - “Why Michael Burry is Shorting Nvidia and Palantir” (2025-11-25)
https://bravosresearch.com/the-macro-report/why-michael-burry-is-shorting-nvidia-and-palantir/

[5] JPMorgan - “Smothering Heights, Eye on The Market | Outlook 2026” (2026-01-01)
https://am.jpmorgan.com/content/dam/jpm-am-aem/global/en/insights/eye-on-the-market/smothering-heights-amv.pdf

[6] MarketBeat - “NVIDIA’s Next Leg Higher May Have Started at CES” (2026-01-10)
https://www.marketbeat.com/originals/nvidias-next-leg-higher-may-have-started-at-ces/

[7] GuruFocus - “Oracle Corporation (ORCL) Company Overview” (2026-01-10)
https://www.gurufocus.com/

[8] Insider Monkey - “Jim Cramer Reveals When You Can Buy Oracle (ORCL)” (2026-01-09)
https://www.insidermonkey.com/blog/jim-cramer-reveals-when-you-can-buy-oracle-orcl-1672502/

[9] MarketWatch - “Oracle and Amazon are AI ‘loser’ stocks” (2026-01-09)
https://www.marketwatch.com/story/oracle-and-amazon-are-ai-loser-stocks-but-heres-why-thats-primed-to-change-68d87116

[10] CNBC - “Are we in an AI bubble? What 40 tech leaders and analysts are saying” (2026-01-10)
https://www.cnbc.com/2026/01/10/are-we-in-an-ai-bubble-tech-leaders-analysts.html

[0] Jinling AI Financial Database (Real-time market data, financial analysis, technical indicators)