Video summary

Did The Bust Just Begin? | Jesse Felder

Main summary

Key takeaways

Finance

Market/Macro Thesis (AI bubble likely in late-stage / “topping process”)

  • Jesse Felder argues the stock market is in a prolonged topping process that has taken longer than he expected.
  • He suggests the AI boom is metastasizing into an AI bust, with the next major downside wave likely not limited to semiconductors but spreading to broader large-cap tech/hyperscalers due to:
    • Physical/operational constraints (data center build-out delays/cancellations)
    • Macro risk resembling stagflationary dynamics (rising prices + weaker labor demand)
    • Valuation risk (extreme multiples; earnings/FCF deterioration later)
    • Speculative excess (market structure/leverage)

Key Constraints & “Physics” of Data Center Buildout (catalyst for earnings revisions)

Felder repeatedly frames the core issue as a mismatch between what analysts are pricing and what the real world can deliver.

Data center delays/cancellations

  • He claims data centers have been delayed “for this basically this whole year” and that first cancellations are starting.
  • He cites a statistic: “two-thirds” of data centers projected for next year reportedly haven’t even broken ground.

Inventory bottlenecks

  • Example supply chain: Nvidia GPUs → suppliers like Super Micro build racks/systems → sold to data centers, but data centers aren’t ready to accept inventory.

Timing/Accounting effects boosting near-term earnings

  • Semiconductor/AI earnings strength reflects timing of revenue recognition vs. expense ramp, while hyperscalers’ heavy costs (e.g., depreciation/capex-related impacts) haven’t fully hit yet.

Implied forecast risk

  • He expects analysts’ forward earnings/revenue estimates must come down rapidly once it becomes clearer that the buildout can’t occur on the expected timeline.

Macro Backdrop: Stagflation signals and inflation/interest-rate pressure

  • NFIB small business survey (this week)
    • Small businesses are raising prices and dramatically reducing plans to hire.
    • Presented as a leading indicator for rising unemployment + rising prices → stagflation.
  • Iran war / geopolitical impact
    • Used as a driver for stagflationary conditions and energy market disruption.
  • Inflation data mentioned
    • He references CPI around 4%.
    • Suggests core PCE / PCE has been stronger than expected, implying the Fed will be pressured to tighten.

Specific Market/Trading Catalysts He Points To

  • Semiconductors down ~10% on Friday (explicit number).
  • Analyst behavior
    • He claims price targets have seen more reductions than increases in the second or third week of the year.
  • Company/data-center construction pauses
    • Mentions a large data center company placing a 1 gigabyte / 1.5 gigabyte project on hold at customer request (audio suggests “gigabyte,” but context implies a gigawatt-class facility).
    • Interprets customer holds as evidence hyperscalers are scaling back demand consistent with token-based pricing realities.

“Token-based pricing” and demand normalization (why compute demand may fall)

  • Felder argues earlier demand projections were inflated during the “token maxing” era when AI usage was effectively subsidized (by OpenAI/Anthropic/Microsoft/venture capital and low perceived marginal cost).
  • He claims hyperscalers now need more token-based pricing because they can’t subsidize compute to the same extent.
  • Result:
    • Demand is already dropping off
    • Analyst demand estimates likely need to come down

On-device AI as a further headwind to data-center intensity

  • Mentions Apple sold Mac minis allowing agents to run on-device without cloud assistance.
  • Notes that Dell and Nvidia have been developing chips for on-device AI, implying potentially less need for large data center capacity over time.

Hyperscalers/free-cash-flow deterioration (“2nd wave” risk)

Even if semiconductor revenue looks strong near term, he highlights a later wave:

  • Hyperscaler profitability deterioration as depreciation/costs rise and spending integrates.
  • Free cash flow is described as zero or negative (directionally, not specific figures).
  • He argues hyperscalers are moving from “monopolies” in their separate ecosystems into competing in the same capital-intensive arena, driving weaker profitability.

Portfolio/Asset Allocation Recommendations (explicit positioning ideas)

Core recommendation

  • Diversification and adding assets beyond US equities.
  • Emphasizes a “new bull market in diversification”, i.e., more importance of diversifying across:
    • International stocks
    • Real assets
    • Fixed income
    • Cash

Sectors he prefers within equities

  • Healthcare (value/contrarian framing)
  • Some consumer staples (mentions insider buying)

Real assets / commodities

  • Prefers energy over precious metals for the time being.
  • Claims energy is still cheap relative to history:
    • Energy is about 3% of the S&P 500
    • Typical range cited: 8–10–12%
  • Claims crude oil supply constraints imply higher prices.

Defensive macro hedges

  • Host suggests cash/T-bills/TIPS; Felder broadly agrees on diversification and real assets, with strongest emphasis on energy/commodities + diversification.

Valuation/price targets & rate expectations (key numbers)

  • Oil price target: $150/barrel (explicit).
  • 10-year Treasury yield target: could go to ~6% over 6–12 months (explicit).
  • Gold as a leading indicator
    • He states that gold’s rise over ~2 years has started to show up in broader commodities, oil, and interest rates over the next ~20 months (timeline mentioned).
  • Gold ETFs / precious metals sentiment
    • He cites a condition: when silver outperforms gold sharply, it can indicate speculation got too hot.
    • Wants gold fund inflows to reverse via meaningful outflows before being more constructive on precious metals.

Precious metals stance (why still under pressure)

  • Expects pressure while:
    • The Fed is tightening (he says higher real interest rates are unfavorable for precious metals).
  • Suggests a better precious-metals entry point may come after the tightening peak, if markets believe the Fed may shift to cutting.

Performance expectations for AI bust scenario

  • Comparison to dot-com:
    • Dot-com: NASDAQ fell ~90%, but the economy didn’t collapse the same way.
  • He argues AI is more broadly embedded now because ETFs and AI-linked exposure are more pervasive.
    • Scenario includes 40–50% equity drawdowns that would be difficult to avoid across broad markets.
  • Cites: >50% of stocks are “caught up” in the AI trade (explicit statistic).

Disclosures / disclaimers

  • The host states: “none of what Jesse said… is personal financial advice. Do your own research.”
  • No additional legal disclaimer details are included beyond that.

Instruments / tickers / assets mentioned

Semiconductors / AI supply chain

  • Nvidia (Nvidia)
  • Super Micro (Super Micro Computer mentioned)

Equity benchmarks

  • S&P 500
  • NASDAQ
  • “Magnificent Seven” (referred to as a group; no tickers listed)

Commodities / macro assets

  • Crude oil
  • Gold
  • Silver
  • Energy sector

Rates/income

  • 10-year Treasury yield
  • T-bills
  • TIPS

ETFs

  • General mention only; no specific ETF tickers.

Methodology / framework (step-by-step style claims)

AI bubble-burst framework

  1. Check physical feasibility: can data centers be built fast enough?
  2. Observe execution signals: delays → cancellations → customer holds
  3. Assess accounting/timing distortions in early earnings
  4. Adjust: analysts’ forward estimates must follow reality
  5. Expect macro feedback loop: stagflation + rate pressure + wealth effect
  6. Watch for profitability “2nd wave” at hyperscalers via depreciation and capex integration

Commodities/energy “signals”

  1. Use gold price trend as a leading indicator for broader commodities
  2. Translate implications into oil price and long-end rates
  3. Prefer energy when energy’s weight vs. S&P is historically low and oil fundamentals are tight

Presenters / sources mentioned

  • Adam Tagert (host, Thoughtful Money founder)
  • Jesse Felder (founder/editor, The Felder Report; macro analyst)
  • Jeremy Grantham (referenced by title; quoted on hyperscaler decade outlook)
  • “Sachin Nadella” (quote about buying chips but lacking “warm shells” / data center capacity)
  • “Sam Altman” / OpenAI (token/compute cost framing)
  • Tavi Costa (commodities/hard assets analyst; mentioned via live stream reference)
  • Warren Buffett / Berkshire Hathaway (cash/”more casino-like” framing referenced)

Original video