Get Started

The AI Valuation Paradox: Analysing the $405 Billion Gamble

  • Insights
  • The AI Valuation Paradox: Analysing the $405 Billion Gamble

The artificial intelligence sector stands at an inflexion point reminiscent of historical market manias, yet with distinct characteristics that defy simple categorisation. Following Nvidia’s November 19, 2025 earnings report, which beat expectations with $57 billion in quarterly revenue (up 62% year-over-year) and $31.9 billion in the market, exhaled briefly[1].

But beneath this surface relief lies a complex web of accounting controversies, structural dependencies, and valuation disconnects that legendary investor Michael Burry has characterised as “one of the most common frauds in the modern era”[2].

Big Tech’s AI infrastructure spending has surged to an estimated $405 billion for 2025, up 62% year-over-year and 44.6% above initial forecasts[3]. Yet revenue growth at pure-play AI companies tells a starkly different story.

OpenAI, valued at $300 billion in its March 2025 funding round, generated approximately $10 billion in annualised revenue as of June 2025, implying a revenue multiple of 30x[4][5]. The company simultaneously burns through $8.5 billion annually while targeting $125 billion in revenue by 2029, a 12.5x growth requirement that raises fundamental questions about sustainable economics[6].

This analysis examines the AI sector through multiple lenses: Burry’s depreciation fraud allegations, the Magnificent Seven’s earnings deceleration, amplification of cross-investment risk, monetary policy implications under a changing Fed regime, and scenario modelling for the next 6–12 months.

The evidence suggests not a simple bubble destined to pop, but rather a restructuring phase where capital allocation, profitability timelines, and competitive dynamics will separate transformative businesses from speculative excesses.

Burry's Accounting Critique: The Depreciation Delusion

The Mechanics of the Alleged Fraud 

Michael Burry emerged from a two-year silence in November 2025 with a precise accusation: hyperscalers are extending GPU depreciation schedules beyond economic reality to inflate reported earnings. His analysis estimates Big Tech will understate depreciation by $176 billion between 2026 and 2028, artificially inflating profits by 26.9% at Oracle and 20.8% at Meta [7]. 

The mechanism is straightforward yet consequential. Until 2024, servers and network gear were depreciated over 3-4 years. Many companies have now extended these schedules to 5-6 years for AI infrastructure, arguing that GPUs retain value longer due to continuous software optimisation and robust secondary markets [8]. This accounting shift directly boosts earnings before interest, taxes, depreciation, and amortisation (EBITDA) by deferring expense recognition. 

Example calculation for a $10 billion GPU purchase: 
  1. Traditional 4-year schedule: $2.5 billion annual depreciation
  2. Extended 6-year schedule: $1.67 billion yearly depreciation
  3. Annual earnings boost: $830 million per $10 billion capex

With Meta planning $65-72 billion in 2025 capex (81% year-over-year growth) and Microsoft targeting $76-376 billion over 2025-2027, the cumulative effect on reported profitability is material[9][10]. 

The Counter-Arguments 

Defenders of extended depreciation schedules cite genuine technological and market factors. NVIDIA H100 GPUs purchased 18-24 months ago still command 60-70% of their original prices in secondary markets, suggesting actual utility retention [11]. Software frameworks like CUDA optimisation and model quantisation techniques extend hardware relevance beyond historical semiconductor lifecycles. 

Auditors point to GAAP flexibility: companies may use depreciation schedules that reflect genuine “expected useful life” rather than arbitrary timelines. The critical question is not whether extended schedules are permissible, but whether they accurately reflect economic reality or primarily serve to manage earnings optics. 

Market Implications 

If Burry’s thesis proves correct and companies are forced to accelerate depreciation recognition (either by regulators, auditors, or deteriorating secondary markets), the earnings impact would cascade through valuations. A 20% profit overstatement at current multiples implies 15-20% downside to fair value, all else equal. More significantly, credibility loss could trigger multiple compressions independent of earnings adjustments. 

Burry’s positioning reinforces his conviction: regulatory filings show approximately $1.1 billion in notional put exposure against tech names, though 13F reports show underlying shares rather than cash outlay[12]. His decision to deregister Scion Asset Management and step away from managing outside money amplifies the message that he views the current game as “fundamentally rigged”[13]. 

Nvidia's Earnings: The Backbone Remains Strong

Q3 FY2026 Performance Breakdown

Nvidia’s November 19 report provided the clearest signal yet that AI infrastructure demand remains robust, at least at the chip level. Key metrics:

Metric Q3 FY26 Q3 FY25 Y/Y Growth
Total Revenue $57.0B $35.1B +62%
Data Centre Revenue $51.2B $30.8B +66%
Net Income $31.9B $19.3B +65%
Gross Margin 75.0% 74.6% +40 bps

Table 1: Nvidia Q3 FY2026 vs Q3 FY2025 performance[14]

The company’s Q4 guidance of $65 billion further exceeds Wall Street’s $54.9 billion expectations, indicating that hyperscaler procurement shows no signs of slowing [1]. CEO Jensen Huang’s statement that “Blackwell sales are off the charts, and cloud GPUs are sold out” reinforces the supply-constrained nature of current demand.

Market Reaction and Concentration Risk

Nvidia’s stock jumped 3.4% in after-hours trading, pulling other Magnificent Seven names higher: Meta, Microsoft, Amazon, and Alphabet all rose in sympathy [1]. This correlation underscores a critical market structure issue: Nvidia accounts for approximately 8% of the S&P 500, meaning nearly all 401(k) holders are directly exposed to its volatility.

The concentration has reached historic levels. Goldman Sachs data shows the Magnificent Seven represented 21% of hedge fund net exposure as of June 2024, before declining to 15.5% by January 2025 as managers implemented concentration caps [15]. Yet even this “de-risking” leaves market breadth dangerously narrow: 70% of S&P 500 gains since 2023 stem from just seven companies [16].

Bank of England officials and 54% of surveyed fund managers now label current pricing a bubble, yet paradoxically, positioning remains elevated relative to historical norms [16]. This creates asymmetric risk: further concentration increases fragility, while de-risking removes the marginal buyer supporting valuations.

The Magnificent Seven: Earnings Deceleration and Margin Pressures

Aggregate Performance Trajectory

While Nvidia continues to astonish, the broader Magnificent Seven cohort faces structural headwinds. Q4 2024 aggregate earnings reached $131.2 billion (an all-time high), growing 31.7% year-over-yearbut this represented the lowest growth rate since Q1 2023[17]. Q1 2025 growth is forecasted at just 18.5%, and quarter-over-quarter growth is expected to collapse from 19.3% to -17.4%[17].

Period Q1 2024 Q2 2024 Q3 2024 Q4 2024
Mag-7 Earnings Growth 117.1% 46.9% 19.8% 31.7%

Table 2: Magnificent Seven earnings growth deceleration[18]

The deceleration is not uniform. Nvidia, Amazon, and Meta contributed 73% of Q4 growth, while Apple, Alphabet, Microsoft, and Tesla showed weaker momentum [17]. This internal divergence suggests AI beneficiaries are pulling away from the broader cohort. 

Margin Compression Dynamics 

Net profit margins for the Magnificent Seven reached 25.8% in Q4 2024, an all-time high since data collection began in Q3 2020, and nearly double the S&P 500’s 13.4%[17]. Yet sustaining these margins faces three structural challenges: 

  1. Capital intensity shift: Meta now spends 30% of revenue on capex ($65B on $185B revenue), a dramatic departure from the asset-light model that defined the 2010s tech boom[19] 
  2. Interest rate environment: Borrowing at near-zero rates to buy back shares, a key margin support mechanism, is impossible with policy rates at 3.75-4.00%[20]
  3. Tariff uncertainty: Reciprocal taxes from the EU and Asia, while not directly hitting software, increase operational complexity and costs[20]

Goldman Sachs strategists peg the Magnificent Seven forward P/E multiple near 30x, a 60% premium to the rest of the S&P 500. With bond yields above 4%, every expansion point carries a heightened opportunity cost [21]. Systematic models blending earnings revisions with valuation have begun automatically downgrading their weightings. 

OpenAI and the Pure-Play AI Valuation Conundrum

Revenue Reality vs Valuation Expectations 

OpenAI represents the most transparent test case for the viability of AI business models. The company’s trajectory: 

  1. 2024 revenue: Approximately $4 billion[22]
  2. H1 2025 revenue: $4.3 billion (16% more than all of 2024)[23]
  3. June 2025 annualised run rate: $10 billion[4]
  4. 2025 projected revenue: $13 billion[6]
  5. 2029 revenue target: $125 billion[6]
  6. March 2025 valuation: $300 billion (30x current revenue)[5]

The mathematics are both impressive and concerning. To justify a $300 billion valuation at a 20x revenue multiple (typical for mature SaaS), OpenAI must reach $15 billion in revenue, achievable within 12-18 months at current growth rates. However, to maintain valuation at that revenue level, margins must approach 30-40%, requiring a dramatic transformation of the cost structure. 

The profitability gap is severe. OpenAI lost approximately $5 billion in 2024 and expects $8.5 billion in cash burn for 2025[6]. R&D spending alone reached $6.7 billion in H1 2025, doubling the $2.5 billion spent in all of 2024[23]. The company held $17.5 billion in cash and securities at mid-year, providing a runway but not an indefinite cushion. 

The Unit Economics Challenge 

As the Equidam analysis emphasises, AI companies face compute costs that scale superlinearly with usage, unlike traditional software, where marginal costs approach zero [24]. This creates three critical questions revenue multiples cannot answer: 

  1. Training costs: What is the all-in cost to develop and continuously improve models? GPT-4’s training reportedly cost $100+ million; GPT-5 estimates range into billions
  2. Inference costs: Per-query GPU costs remain substantial. At $10 billion annual revenue and 500 million weekly users, average revenue per user (ARPU) is just $20/year, leaving little margin for compute, R&D, and sales[25]
  3. Scaling dynamics: Do costs grow linearly, sub-linearly, or super-linearly with user growth? Current evidence suggests super-linear due to model size increases

Anthropic, OpenAI’s closest competitor, recently surpassed $3 billion in annualised revenue tenth of OpenAI’s scale[4]. This gap illustrates winner-take-most dynamics but also raises questions about sustainable differentiation. If open-source models or hyperscaler platforms commoditise capabilities, first-mover advantages evaporate quickly. 

The Perplexity Warning Signal 

Perplexity AI has become a cautionary symbol within venture circles. The conversational search startup raised multiple rounds through 2024-25, reaching reported valuations of $18-20 billion despite minimal revenue[26]. At a recent San Francisco summit, investors named Perplexity among the startups “most likely to fail,” citing unclear revenue models and unsustainable burn rates [26]. 

The concern is not specific to Perplexity but systemic: dozens of generative AI startups command billion-dollar-plus valuations with negligible revenue, betting entirely on land-grab strategies and eventual monetisation. This mirrors 1999-2000 dot-com dynamics, where “eyeballs” and “engagement” substituted for business fundamentals. 

Cross-Investment Amplification: Correlated Risk or Diversification?

The Interconnected AI Ecosystem 

A unique feature of the current AI boom is the dense web of cross-investments among participants. Examples include: 

  1. Microsoft + OpenAI: $13 billion invested, 49% profit share up to return cap, exclusive cloud provider[27] 
  2. Amazon + Anthropic: $4 billion investment, AWS compute partnership 
  3. Google + Anthropic: $2 billion investment, Google Cloud integration 
  4. Nvidia + Multiple startups: Equity stakes in dozens of AI companies via NVentures 
  5. Oracle + OpenAI + Microsoft: $500 billion “Stargate” data centre project[28] 

Additionally, OpenAI pays Microsoft approximately 20% of its revenue for Azure compute services, meaning a significant portion of OpenAI’s $10 billion run rate flows back to its primary investor [29]. This creates a circular dependency: Microsoft’s returns depend on OpenAI’s success, which in turn depends on Microsoft’s infrastructure reliability and pricing. 

Systemic Risk Interpretation 

Does this interconnection diversify risk across the ecosystem or concentrate it? The answer depends on the failure mode: 

Diversification scenario: If one AI startup fails, others continue independently. Microsoft and Amazon both have exposure to AI, so losses at one company don’t trigger correlated collapse. 

Concentration scenario: If the fundamental AI business model proves uneconomic (compute costs exceed revenue potential), all participants fail simultaneously. In this case, cross-investment amplifies lossesMicrosoft loses both its OpenAI equity and Azure revenue, while OpenAI loses access to compute. 

Current evidence leans toward concentration risk. All major AI labs face similar unit economics challenges, rely on the same GPU supplier (Nvidia), and compete for overlapping customer segments. A broad loss of confidence in AI ROI would trigger simultaneous markdowns across the portfolio. 

The Bank for International Settlements specifically warned about hedge fund exposure creating “displacement risk “, the binary outcome where technical leadership retention preserves value, while falling behind causes overnight collapse[30]. With 60% of hedge fund selling in April 2025 concentrated in the Magnificent Seven stocks, positioning sensitivity is extreme [21]. 

Monetary Policy Crosscurrents: Fed Dynamics and the Trump Variable

Current Policy Trajectory 

The Federal Reserve cut its policy rate to 3.75-4.00% in October 2025, the second reduction of the year[31]. Goldman Sachs Research expects a December cut despite Chair Jerome Powell’s hedged language (“not a foregone conclusion”), followed by two 25-basis-point cuts in March and June 2026 to a terminal rate of

3.00-3.25%[31]. 

However, internal divisions at the Fed have intensified. Recent speeches reveal sharp disagreement between inflation hawks (who prioritise persistent 2%+ inflation) and labour market doves (who emphasise cooling hiring)[32]. Market-implied odds of a December cut have fallen to 50-50 after sitting near 80% weeks earlier[33]. 

Key data points framing the debate: 

  1. November CPI: 2.7% year-over-year, up from 2.6% in Octoberwrong direction from the Fed’s 2% target[34]
  2. November retail sales: +0.7%, exceeding 0.6% expectations, suggesting consumer resilience[34]
  3. October unemployment: Steady but no longer declining rapidly, indicating labour market normalisation [31]

Cleveland Fed President Beth Hammack articulated the hawkish view: “Strong growth, a robust labour market, and elevated inflation indicate to me that it is appropriate to maintain a somewhat restrictive monetary policy for the time being”[35]. 

The Trump Administration Wildcard 

President Donald Trump’s aggressive criticism of the Fed adds unprecedented uncertainty. On November 18, 2025, Trump stated: “I think I already know my choice” for the next Fed chair to replace Powell, whose term ends in May[36]. Trump has repeatedly said he wants Powell out immediately, though Treasury Secretary Scott Bessent and others have counselled against firing him to avoid market chaos[37]. 

Trump’s November 19 comments were particularly revealing: “The only thing Scott is blowing it on is the Fed because the Fed, the rates are too high, Scott. If you don’t get it fixed fast, I’m going to fire your ass”[38]. While firing Bessent would not directly affect Fed policy (the Fed is independent), the pressure campaign signals potential institutional stress ahead. 

The next chair will face an impossible balancing act: appeasing Trump’s demands for lower rates while maintaining investor confidence in Fed independence and inflation-fighting credibility[36]. If markets perceive the Fed as politically captured, long-term rates could rise even as policy rates fall, a perverse outcome that tightens financial conditions. 

Implications for AI Valuations 

Higher-for-longer rates directly pressure AI valuations through two channels: 

  1. Discount rate effect: Long-duration growth stocks (with cash flows years away) are most sensitive to changes in the discount rate. A 100-basis-point increase in discount rates translates to
15-20% valuation compression for companies expected to turn profitable in 2027-2028 
  1. Competing alternatives: With 10-year Treasuries yielding 4.3% and investment-grade corporate bonds at 5.5%+, the hurdle rate for equity risk premium rises. Growth stocks must demonstrate clearer paths to profitability to justify premium multiples

Conversely, aggressive rate cuts (Trump’s preferred approach) could reignite speculative fervour, prolonging the AI bubble phase. The 2020-2021 experience showed how zero rates eliminate opportunity cost for long-duration bets, inflating unprofitable tech valuations. 

Monetary Policy Crosscurrents: Fed Dynamics and the Trump Variable

Scenario 1: Managed Normalisation (50% probability) 

Key developments: 

  • Nvidia and hyperscalers continue robust capex, reaching $450 billion combined in 2026 
  • OpenAI achieves $13-15 billion revenue but remains unprofitable; valuation holds at $250-300 billion on growth trajectory
  • Magnificent Seven earnings growth stabilises at 12-15% annually, down from 30%+ but above the S&P 500 average
  • Fed executes 2-3 cuts to 3.00-3.25% terminal rate without market disruption
  • No major accounting scandals materialise; depreciation controversy fades as GPU secondary markets remain healthy Market impact: 
  • AI stocks consolidate in a 10-15% range around current levels
  • Sector rotation into profitable AI beneficiaries (Nvidia, Microsoft, Amazon ) and away from unprofitable startups
  • Private market valuations decline 20-30% for pure-play AI companies in 2026 funding rounds
  • S&P 500 returns 6-8% as breadth improves beyond Magnificent Seven
Scenario 2: Burry Vindication – Accounting Reckoning (30% probability) 

Key developments: 

  • SEC or auditors force accelerated depreciation recognition at 2+ major hyperscalers
  • Earnings restatements reveal 15-20% profit overstatement; CFOs resign, credibility crumbles
  • Secondary GPU markets weaken as newer architectures (Blackwell, next-gen) commoditise H100/A100 capabilities
  • One major AI startup (potentially Perplexity or similar) fails, triggering down-round contagion Fed pauses cuts due to sticky inflation, keeping rates at 3.75-4.00% through Q2 2026  Market impact: 
  • Magnificent Seven stocks decline 25-35% from peaks as multiple compression compounds earnings hits
  • Nvidia holds relatively well (down 15-20%) due to fundamental demand, but suffers sympathy selling 
  • Credit spreads widen; high-yield bonds of unprofitable AI companies face distress
  • S&P 500 declines 12-18%; VIX spikes above 35
  • M&A activity accelerates as weaker players seek acquirers before cash runways end
Scenario 3: AI Winter – Fundamental Model Failure (15% probability) 

Key developments: 

  • Evidence emerges that LLMs have hit scaling limits; GPT-5/Gemini Ultra fail to show meaningful improvements over GPT-4/Gemini
  • Enterprise AI adoption stalls as ROI studies show minimal productivity gains relative to costs
  • Hyperscalers announce capex cuts of 30-40%; Nvidia revenue guidance drops 25%+ 
  • Multiple AI unicorns declare bankruptcy; venture funding collapses 70%+
  • The recession begins due to overleveraged credit conditions and the tech sector contraction Market impact: 
  • Tech-heavy indices enter bear market; Nasdaq down 35-45%
  • Magnificent Seven loses $3-4 trillion in market cap; Nvidia falls 50%+ 
  • Credit crisis emerges as CLO exposure to tech debt creates contagion
  • Fed forced into emergency cuts to 2.00-2.50%, but ineffective due to credit freeze
  • Global growth contracts 0.5-1.0% as AI capex reversal creates a negative multiplier
Scenario 4: Renewed Euphoria – Trump Fed Capitulation (5% probability) 

Key developments: 

  • Trump successfully pressures Powell replacement to cut aggressively; rates fall to 2.50-2.75% by mid-2026
  • Fiscal stimulus (tax cuts, deregulation) reignites growth expectations
  • AI hype cycle extends; new applications (autonomous agents, scientific discovery) validate investment thesis
  • Nvidia exceeds guidance by 20%+; stock breaks new all-time highs 
  • Retail investor FOMO returns; crypto and AI stocks surge in tandem Market impact: 
  • Magnificent Seven rallies 40-60%; S&P 500 gains 25-30%
  • Valuation multiples expand to 2021 levels; P/E ratios hit 35x+ for growth stocks
  • Credit quality deteriorates as cheap money fuels malinvestment; zombie AI companies survive on continuous fundraising
  • Inflation re-accelerates to 3-4%; bond vigilantes sell Treasuries, pushing 10-year yields above 5%
  • Set up for a larger crash in 2027 as unsustainable dynamics eventually resolve 

Strategic Considerations for Investors

Risk Management Imperatives 

Given the scenario distribution and elevated uncertainty, investors should consider: 

  1. Concentration limits: Cap exposure to Magnificent Seven at 15-20% of equity allocation, down from 30%+ for many portfolios. Use forced rebalancing to prevent drift
  2. Quality bias within AI: Favour cash-generative AI beneficiaries (Nvidia, Microsoft, Amazon ) over cash-burning pure-plays (OpenAI, Anthropic, Perplexity)
  3. Valuation discipline: Avoid companies trading at 30x+ revenue multiples unless the path to 40%+ operating margins is demonstrable. Most AI software companies will face 20-30% margin ceilings due to computing costs.
  4. Duration hedge: Increase allocation to shorter-duration value stocks and sectors (financials, energy, industrials) to hedge against AI growth disappointment
  5. Tail risk protection: Maintain 3-5% portfolio allocation to put options or volatility strategies, sized for a 25-35% Magnificent Seven drawdown scenario
Opportunity Identification 

Paradoxically, structural concerns create an asymmetric opportunity if navigated correctly: 

  1. Infrastructure beneficiaries: Power utilities, data centre REITs, and electrical equipment manufacturers benefit from AI capex regardless of the success of the software layer. Constellation Energy, Vistra, and Digital Realty offer compelling risk-reward[28]
  2. Nvidia volatility selling: Implied volatility remains elevated (40-50%) despite strong fundamentals. Selling 15-20% out-of-the-money puts captures premium while establishing entry points at attractive valuations.
  3. Enterprise software survivors: Companies with genuine AI-driven productivity improvements

(ServiceNow, Salesforce ) should separate from speculative plays as markets differentiate 

  1. Contrarian quality in de-rated names: Apple, Alphabet, and Oracle trade at 22-25x earnings despite strong cash flow and manageable AI investment potential value plays if market panic creates indiscriminate selling 
The 2026 Inflexion Point 

Multiple factors converge in 2026 to force resolution of current ambiguities: 

  1. Profitability deadline: OpenAI and peers must demonstrate clear paths to breakeven, or face down-rounds in 2026-2027 funding cycles
  2. Capex normalisation: Hyperscalers have guided to “significantly higher” 2026 capex, but 60%+ growth rates are unsustainable. Any deceleration will test Nvidia multiple times 
  3. Enterprise AI ROI data: By late 2026, sufficient deployments will exist to measure genuine productivity impact. If results disappoint, adoption will stall
  4. Fed regime change: New chair begins tenure in May; if inflation remains above 3%, credibility-building rate hikes become likely regardless of political pressure
  5. Depreciation schedules are maturing: GPUs purchased in 2024 are approaching mid-life; if performance degradation or obsolescence accelerates, secondary-market weakness will validate Burry’s thesis.

Conclusion: To AI or Not to AI?

The AI sector in November 2025 embodies Hemingway’s observation about bankruptcy: it happens “gradually, and then all at once.” Multiple indicators suggest we remain in the “gradual” phase, Nvidia demand remains robust, hyperscaler spending accelerates, and equity markets show resilience. Yet the structural stresses are accumulating: unsustainable burn rates, questionable accounting practices, extreme concentration risk, and valuation disconnects from profitability timelines. 

Michael Burry’s warning merits serious consideration, not because depreciation fraud has been definitively proven, but because it highlights a broader pattern: markets have systematically underpriced the risk that AI business models prove fundamentally uneconomic. The $405 billion annual capex run rate demands revenue growth and margin profiles that no technology sector has ever achieved at a comparable scale. 

For investors, the binary framing of “AI bubble or AI revolution” is false. Both can be true: transformative technology can coexist with massive capital misallocation and valuation excess. The challenge is distinguishing companies that will compound capital at 15-20% annually for a decade (Nvidia, Microsoft, select infrastructure plays) from those that will consume capital before succumbing to competition or economic reality (most AI unicorns). 

The next 6-12 months will not provide definitive answers; AI’s full impact will take years to manifest. But this period will likely separate survivors from casualties, reward discipline over speculation, and penalise concentration regardless of the quality of the individual companies. In an environment where 70% of market gains come from seven stocks, the greatest risk may be not what investors own, but what they fail to diversify away from. 

As Burry’s November 25 promised update approaches, markets would be wise to heed the lesson from 2007: when leverage unwinds, it happens faster than it built up. The AI infrastructure boom has lasted 24 months; the unwind, if it comes, may take 24 weeks. 

References

[1] CNN Business. (2025, November 18). Nvidia beats earnings expectations, even as bubble fears persist. https://www.cnn.com/2025/11/19/tech/nvidia-earnings-ai-bubble-fears

[2] Business Insider. (2025, November 16). Michael Burry Warns of AI Bubble With ‘Lord of the Rings’ Meme. https://www.businessinsider.com/big-short-michael-burry-ai-tech-bubble-capital-expenditures-lotr-2025-11

[3] IO Fund. (2025, November 12). Big Tech’s $405B Bet: Why AI Stocks Are Set Up for a Correction. https://io-fund.com/ai-stocks/ai-platforms/big-techs-405b-bet

[4] Reuters. (2025, June 9). OpenAI’s annualised revenue hits $10 billion, up from $5.5 billion in December 2024. https://www.reuters.com/business/media-telecom/openais-annualized-revenue-hits-10-billion-up-55-billion-december-2024-2025-06-09

[5] CNBC. (2025, June 9). OpenAI hits $10 billion in annual recurring revenue fueled by ChatGPT growth. https://www.cnbc.com/2025/06/09/openai-hits-10-billion-in-annualized-revenue-fueled-by-chatgpt-growth.html

[6] Entrepreneur. (2025, October 9). OpenAI Saw More Revenue in Six Months Than All of Last Year. https://www.entrepreneur.com/business-news/openai-saw-more-revenue-in-six-months-than-all-of-last-year/497774

[7] Fortune. (2025, November 12). The ‘Big Short’ investor betting $1 billion against the AI bubble explains his position. https://fortune.com/2025/11/13/the-big-short-investor-closing-scion-ai-bubble-depreciation-explained/

[8] AICerts. (2025, November 13). Michael Burry Flags Looming AI Valuation Bubble. https://www.aicerts.ai/news/michael-burry-flags-looming-ai-valuation-bubble/

[9] IO Fund. (2025, November 12). Big Tech’s $405B Bet: Why AI Stocks Are Set Up for a Correction. https://io-fund.com/ai-stocks/ai-platforms/big-techs-405b-bet

[10] Yahoo Finance. (2025, October 6). Big Tech’s Infrastructure Spending to Nearly Triple Amid AI Push. https://finance.yahoo.com/news/big-techs-infrastructure-spending-nearly-211211451.html

[11] Business Standard. (2025, November 18). Is the AI bubble on the brink of a reckoning as dot-com fears return? https://www.business-standard.com/technology/tech-news/ai-bubble-dotcom-fears-investors-nvidia-oracle-perplexity-markets-125111900159_1.html

[12] AICerts. (2025, November 13). Michael Burry Flags Looming AI Valuation Bubble. https://www.aicerts.ai/news/michael-burry-flags-looming-ai-valuation-bubble/

[13] Fortune. (2025, November 12). The ‘Big Short’ investor betting $1 billion against the AI bubble explains his position. https://fortune.com/2025/11/13/the-big-short-investor-closing-scion-ai-bubble-depreciation-explained/

[14] NVIDIA. (2025, November 19). NVIDIA Announces Financial Results for Third Quarter Fiscal 2026. https://investor.nvidia.com/news/press-release-details/2025/NVIDIA-Announces-Financial-Results-for-Third-Quarter-Fiscal-2026/

[15] Resonanz Capital. (2025, August 14). De-Crowding the Magnificent Seven. https://resonanzcapital.com/insights/de-crowding-the-magnificent-seven

[16] AICerts. (2025, November 13). Michael Burry Flags Looming AI Valuation Bubble. https://www.aicerts.ai/news/michael-burry-flags-looming-ai-valuation-bubble/

[17] LSEG. (2025, November 3). Magnificent-7 Q4 2024 Earnings Review: Growth Holds, Rotation Awaits. https://www.lseg.com/en/insights/data-analytics/magnificent-7-q4-2024-earnings-review-growth-holds-rotation-awaits

[18] Boston Partners. (2024, September). Are the Magnificent Seven losing momentum? https://www.bostonpartners.com/uploads/2024/09/e8c98a27ac7b18aadc20ee4c82714581/sep-2024-magnificent-7-wp-final.pdf

[19] T. Rowe Price. (2025). The “Magnificent Seven,” equity market concentration, and portfolio considerations. https://www.troweprice.com/institutional/us/en/insights/articles/2025/q2/magnificent-7-equity-market-concentration-and-portfolio-considerations

[20] T. Rowe Price. (2025). The “Magnificent Seven,” equity market concentration, and portfolio considerations. https://www.troweprice.com/institutional/us/en/insights/articles/2025/q2/magnificent-7-equity-market-concentration-and-portfolio-considerations

[21] Resonanz Capital. (2025, August 14). De-Crowding the Magnificent Seven. https://resonanzcapital.com/insights/de-crowding-the-magnificent-seven

[22] TechCrunch. (2025, November 13). Leaked documents shed light on how much OpenAI pays Microsoft. https://techcrunch.com/2025/11/14/leaked-documents-shed-light-into-how-much-openai-pays-microsoft/

[23] Entrepreneur. (2025, October 9). OpenAI Saw More Revenue in Six Months Than All of Last Year. https://www.entrepreneur.com/business-news/openai-saw-more-revenue-in-six-months-than-all-of-last-year/497774

[24] Equidam. (2025, June 26). AI Startup Valuation: Revenue Multiples, 2025 Challenges and Insights. https://www.equidam.com/ai-startup-valuation-revenue-multiples-2025-challenges-insights-2/

[25] CNBC. (2025, June 9). OpenAI hits $10 billion in annual recurring revenue fueled by ChatGPT growth. https://www.cnbc.com/2025/06/09/openai-hits-10-billion-in-annualized-revenue-fueled-by-chatgpt-growth.html

[26] Business Standard. (2025, November 18). Is the AI bubble on the brink of a reckoning as dot-com fears return? https://www.business-standard.com/technology/tech-news/ai-bubble-dotcom-fears-investors-nvidia-oracle-perplexity-markets-125111900159_1.html

[27] TechCrunch. (2025, November 13). Leaked documents shed light into how much OpenAI pays Microsoft. https://techcrunch.com/2025/11/14/leaked-documents-shed-light-into-how-much-openai-pays-microsoft/

[28] Business Insider. (2025, October 6). AI Capex Boom: Payoff Uncertain for Infrastructure, Data Centers. https://www.businessinsider.com/big-tech-ai-capex-infrastructure-data-center-wars-2025-10

[29] TechCrunch. (2025, November 13). Leaked documents shed light into how much OpenAI pays Microsoft. https://techcrunch.com/2025/11/14/leaked-documents-shed-light-into-how-much-openai-pays-microsoft/

[30] Bank for International Settlements. (2024, September 2). Hedge fund exposure to the carry trade. https://www.bis.org/publ/qtrpdf/r_qt2409x.htm

[31] Goldman Sachs. (2025, November 4). The Fed Is Forecast to Cut Rates in December as Employment Cools. https://www.goldmansachs.com/insights/articles/the-fed-is-forecast-to-cut-rates-in-december-as-employment-cools

[32] Finance Commerce. (2025, November 18). Fed split on December rate cut as inflation, jobs diverge. https://finance-commerce.com/?p=548101

[33] Finance Commerce. (2025, November 18). Fed split on December rate cut as inflation, jobs diverge. https://finance-commerce.com/?p=548101

[34] NBC News. (2024, December 18). Fed cuts key interest rate but signals elevated inflation is likely to persist. https://www.nbcnews.com/business/economy/federal-reserve-interest-rate-cut-december-2024-much-economy-rcna184586

[35] NBC News. (2024, December 18). Fed cuts key interest rate but signals elevated inflation is likely to persist. https://www.nbcnews.com/business/economy/federal-reserve-interest-rate-cut-december-2024-much-economy-rcna184586

[36] Business Standard. (2025, November 18). Trump signals he has identified his pick for next Federal Reserve chair. https://www.business-standard.com/world-news/trump-signals-he-has-identified-his-pick-for-next-federal-reserve-chair-125111900099_1.html

[37] CNN. (2025, November 18). Trump teases he’ll fire Bessent if Fed doesn’t cut interest rates. https://www.cnn.com/2025/11/19/politics/bessent-trump-federal-reserve-powell

[38] CNN. (2025, November 18). Trump teases he’ll fire Bessent if Fed doesn’t cut interest rates. https://www.cnn.com/2025/11/19/politics/bessent-trump-federal-reserve-powell