Is AI a Bubble? Howard Marks Dissects the $5 Trillion Infrastructure Bet

Published:
  Lastmod: 2025-12-17
Random Notes
by Ernest Chiang

Post Title Image (Illustration: Frozen soap bubble. Image source: Photo by Jill Warvel on Unsplash.)

✳️ tl;dr

  • Howard Marks believes AI shows signs of “irrational exuberance,” but bubbles can usually only be identified in retrospect, and current valuations, while high, have not yet reached crazy levels 1
  • “One of the most interesting aspects of bubbles is their regularity, not in terms of timing, but rather the progression they follow. Something new and seemingly revolutionary appears and worms its way into people’s minds. It captures their imagination, and the excitement is overwhelming. The early participants enjoy huge gains. Those who merely look on feel incredible envy and regret and – motivated by the fear of continuing to miss out – pile in.”
  • AI exhibits bubble characteristics: revolutionary technology, FOMO-driven speculation, extremely high valuations, but “this time is different” may hold true with a 20% probability. (Everyone wants to predict the future but also hedge their bets?)
  • Circular deals raise concerns: Nvidia invests $100 billion in OpenAI, which uses that money to purchase Nvidia chips, with Goldman Sachs estimating 15% of Nvidia’s sales come from such transactions
  • Data center investment scale is staggering: JPMorgan estimates total AI infrastructure buildout costs at approximately $5 trillion, with spending approaching $500 billion next year
  • Debt financing risks escalate: Oracle, Meta, and Alphabet issue 30-year bonds to finance AI investments, with yields exceeding US Treasuries by only 100 basis points or less
  • Warren Buffett reminds us: automobiles were the most important invention of the first half of the 20th century, but only 3 out of 2,000 car companies survived, proving that technological importance doesn’t guarantee investor profits
  • AWS Hero Ernest recommends that technical decision-makers establish a multi-vendor strategy, avoid sole dependence on Nvidia, and evaluate alternatives such as AWS Trainium and Google TPU for cost control and supply chain resilience
  • AI chips have an actual useful life of only 1-3 years, yet companies use 5-6 year depreciation schedules, with Michael Burry accusing tech giants of inflating earnings 2
  • Nvidia shifted from a 2-year to an annual product cycle, with Jensen Huang joking: “Once Blackwell starts shipping, you couldn’t give Hoppers away”
  • Anthropic derives 80% of revenue from enterprise customers, with B2B models showing more promise due to higher transaction values, suggesting product strategy should prioritize enterprise markets 3
  • OpenAI expects to continue massive losses until 2028, with HSBC estimating it won’t be profitable by 2030, requiring an additional $207 billion in funding 4
  • Google TPU emerges as Nvidia’s strongest competitor, with 7th generation Ironwood offering 2x power efficiency improvement and 1.4x Nvidia’s cost-effectiveness, targeting 10% market share by 2027 5
  • SPVs (Special Purpose Vehicles) are used for data center financing, hiding off-balance-sheet debt, raising concerns similar to the Enron model
  • WEF predicts AI will displace 85 million jobs by 2030 but create 97 million new ones, though 77% of new jobs require master’s degrees, necessitating fundamental HR strategy adjustments 6
  • 77,999 jobs have already been lost to AI in 2025, averaging 491 people unemployed daily, with Microsoft reporting 30% of code written by AI while 40% of layoffs target engineers 7
  • Historical analogy: The AI bubble resembles the 1860s railroad boom and 1920s aviation bubble, both being “inflection bubbles” that accelerated technology adoption at investors’ expense 1
  • Current AI giants average a P/E ratio of about 34, lower than the dot-com bubble’s 59, but Shiller CAPE reaches 40.40, approaching dot-com bubble levels 8

✳️ Knowledge Graph

(More about Knowledge Graph…)

graph LR
    %% Concept Classes - Orange
    AI[AI Technology]:::concept
    Bubble[Market Bubble]:::concept
    Infra[Infrastructure]:::concept
    Finance[Financing Methods]:::concept
    Competition[Market Competition]:::concept
    
    %% Instances - Blue
    Nvidia[Nvidia]:::instance
    OpenAI[OpenAI]:::instance
    Microsoft[Microsoft]:::instance
    DataCenter[Data Centers]:::instance
    GPUs[AI Chips/GPUs]:::instance
    Capex[Capital Expenditure]:::instance
    Debt[Debt Financing]:::instance
    SPV[Special Purpose Vehicles]:::instance
    CircularDeals[Circular Deals]:::instance
    FOMO[FOMO]:::instance
    Valuation[High Valuations]:::instance
    SeedRounds[Billion Dollar Seed Rounds]:::instance
    
    %% Technology Flow
    AI -->|requires| Infra
    Infra -->|consists of| DataCenter
    DataCenter -->|uses| GPUs
    Nvidia -->|manufactures| GPUs
    
    %% Investment Flow
    Finance -->|includes| Capex
    Finance -->|includes| Debt
    Capex -->|builds| DataCenter
    Debt -->|structures through| SPV
    
    %% Company Relationships
    OpenAI -->|partners with| Microsoft
    Nvidia -->|invests in| OpenAI
    OpenAI -->|buys from| Nvidia
    Microsoft -->|invests in| OpenAI
    
    %% Bubble Dynamics
    AI -->|creates| Bubble
    Bubble -->|driven by| FOMO
    FOMO -->|leads to| Valuation
    Valuation -->|manifests in| SeedRounds
    CircularDeals -->|inflates| Bubble
    
    %% Market Structure
    AI -->|shapes| Competition
    Competition -->|determines| Valuation
    Nvidia -->|dominates| Competition
    
    %% Circular Transactions
    CircularDeals -->|involves| Nvidia
    CircularDeals -->|involves| OpenAI
    
    %% Financial Risk
    Debt -->|increases| Bubble
    SPV -->|obscures| Debt
    
    %% Styling
    classDef concept fill:#FF8000,stroke:#000,stroke-width:2px
    classDef instance fill:#0080FF,stroke:#000,stroke-width:2px

sequenceDiagram
    participant Tech as New Technology
    participant Early as Early Investors
    participant Media as Media/Hype
    participant Mass as Mass Investors
    participant Market as Market Price
    
    Tech->>Early: Revolutionary promise
    Early->>Market: Generate returns
    Market->>Media: Attract attention
    Media->>Mass: Create FOMO
    Mass->>Market: Pour in capital
    Market->>Market: Price inflation
    Note over Market: Irrational Exuberance
    Market->>Mass: Bubble bursts
    Mass->>Mass: Losses occur

stateDiagram-v2
    [*] --> Installation
    Installation --> Speculation
    Speculation --> BubblePeak
    BubblePeak --> Correction
    Correction --> Deployment
    Deployment --> Maturity
    Maturity --> [*]
    
    note right of Installation
        Infrastructure building
    end note
    
    note right of Speculation
        Capital inflow
    end note
    
    note right of BubblePeak
        Maximum valuation
    end note
    
    note right of Correction
        Value destruction
    end note
    
    note right of Deployment
        Real value creation
    end note
    
    note right of Maturity
        Stable returns
    end note

flowchart TD
    Start[AI Investment Need] --> Decision{Use Debt?}
    Decision -->|Yes| Debt[Issue Debt]
    Decision -->|No| Equity[Use Cash Flow]
    
    Debt --> SPV[Create SPV]
    SPV --> OffBalance[Off-Balance Sheet]
    OffBalance --> Risk1[Hidden Risk]
    
    Debt --> DataCenter[Build Data Center]
    DataCenter --> Question1{Revenue adequate?}
    Question1 -->|No| Default[Default Risk]
    Question1 -->|Yes| Repay[Debt Repayment]
    
    DataCenter --> Question2{Tech obsolete?}
    Question2 -->|Yes| Stranded[Stranded Assets]
    Question2 -->|No| Value[Asset Value]
    
    Default --> Crisis[Financial Crisis]
    Stranded --> Loss[Investment Loss]
    
    Equity --> LowerRisk[Lower Risk]
    Repay --> Success[Success]
    Value --> Success
    LowerRisk --> Success

graph TB
    subgraph DotCom[Dot-com Bubble 1998-2000]
        DC1[No existing products]
        DC2[P/E Ratio: 59]
        DC3[Few users]
        DC4[IPO frenzy]
        DC5[No revenue model]
    end
    
    subgraph AI[AI Boom 2023-2025]
        AI1[Existing products]
        AI2[P/E Ratio: 34]
        AI3[1 billion+ users]
        AI4[Private mega-rounds]
        AI5[Revenue but losses]
    end
    
    DotCom -.Similar.-> AI
    DC1 -.vs.- AI1
    DC2 -.vs.- AI2
    DC3 -.vs.- AI3
    DC4 -.vs.- AI4
    DC5 -.vs.- AI5
    
    Both[Common Factors] --> F1[FOMO]
    Both --> F2[Circular Deals]
    Both --> F3[High Valuations]
    Both --> F4[Speculative Behavior]

✳️ Further Reading

  1. https://www.oaktreecapital.com/insights/memo/is-it-a-bubble ↩︎ ↩︎

  2. https://blog.citp.princeton.edu/2025/10/15/lifespan-of-ai-chips-the-300-billion-question/ ↩︎

  3. https://fortune.com/2025/11/12/openai-cash-burn-rate-annual-losses-2028-profitable-2030-financial-documents/ ↩︎

  4. https://fortune.com/2025/11/26/is-openai-profitable-forecast-data-center-200-billion-shortfall-hsbc/ ↩︎

  5. https://wccftech.com/hot-take-the-true-ai-chip-challenge-for-nvidia-isnt-from-amd-or-intel-its-googles-tpus-heating-up-the-race/ ↩︎

  6. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5316265 ↩︎

  7. https://www.finalroundai.com/blog/ai-replacing-jobs-2025 ↩︎

  8. https://www.allaboutai.com/resources/ai-bubble/ ↩︎