The Bond Market Is Flashing Danger — And AI Is A Bubble | Jesse Felder

Macro thesis: inflation is entering a “secular” resurgence; long-end yields may break higher

• Jesse Felder argues the world is in an inflationary age (2020s onward) that will persist, not fade quickly—even beyond temporary war/oil shocks (e.g., the Iran war as a catalyst).
• He emphasizes the “most important chart” = the 10-year Treasury yield (and watches the 30-year closely), because asset pricing across markets is driven by the level/trend of longer yields.
• He suggests bond markets have not fully priced this inflation-regime shift, creating risk that long-end yields “break out” to new highs.

Key leading indicators he highlights

• Gold and commodity prices
  • Presented as a leading indicator for inflation and interest rates.
  • He claims gold’s strong move over the last ~18 months foreshadowed a broader commodity upturn.
• Commodities → later yields
  • Commodity strength implies a vulnerable period for Treasuries.
  • He frames it as: commodities have been “soaring” for about ~six months, which (per his relationship) points to yields trending higher about ~6 months later.

Potential magnitude/timing of rate risk

• Starting point mentioned: 10-year Treasury yield ~4.4% (described as “still under four and a half”).
• Forecast/threshold language:
  • Yields could rise “pretty easily” to over 5% in a short period of time.
  • More severe scenario: 10-year over 5% → ~5.5% → pushing ~6%, which he says would likely bring a recession.

Why inflation expectations could become “unanchored”

• Core drivers beyond oil
  • Demographics: aging societies / a smaller labor force vs a larger population → wage pressure over the long run.
  • Deglobalization: reshoring and supply-chain reconfiguration → structurally higher costs.
  • Underinvestment in commodity production (oil & gas, mining) for ~10–15 years → structural shortages.
  • US deficits/debt: cites deficit ~5–6% of GDP, described as creating substantial new issuance and inflationary financing pressure.
• Fed credibility risk
  • He cites reporting that the Fed may shift from discussing rate cuts to “laying the groundwork for rate hikes.”
  • He warns that if people stop believing the Fed will reach target after missing it for ~five years, inflation expectations can de-anchor, creating feedback loops.

---

Equity/AI angle: why stocks can rally despite bond danger (risk-of-crowding argument)

Why stocks might hold up

• He attributes stock strength to speculative “dip buying” and crowding:
  • Retail participation is high; “animal spirits” remain elevated since the pandemic/meme-stock era.
  • He cites an overlap: for the first time on record, some portion of institutions overlaps with retail—about a third of hedge funds and growth mutual funds hold the same stocks as retail.
• He argues the rally is heavily linked to AI, but claims AI is “a bubble.”

His AI bubble argument (earnings quality / capex timing)

• Earnings quality deteriorating at hyperscalers:
  • Earnings may be up, but free cash flow collapses toward zero due to aggressive capex (data centers / GPUs).
• Accounting/timing effects
  • AI infrastructure spend is recognized as revenue/income by some companies, while depreciation/expenses may lag until data centers come online.
• He expects a later earnings downdraft when capex timing catches up and depreciation surges.
• Historical parallel:
  • Similar setup in 2001–2002: revenue growth slowed, expenses rose, and the NASDAQ fell ~90%+ (as referenced).

Additional AI skepticism

• Data center buildout constraints
  • Only about ~one-third of announced projects have broken ground (his claim).
• Examples
  • Mentions Nvidia and hyperscalers.
  • Also cites OpenAI/Oracle canceling some compute/data center contracts as a sign of capacity/cost constraints.
• LLM/agentic reliability
  • He claims LLM accuracy around ~95–96% is not enough for “six sigma” style reliability (implying a need for 99.999% class reliability).
• AI wages / inflation suppression
  • He argues wage-inflation suppression via AI is unlikely because replacing labor with AI costs more (per company narratives).
• IPO/monetization angle
  • Providers must raise prices (OpenAI/Anthropic) due to heavy cash burn.
  • He references Anthropic switching from flat monthly fees to token-based pricing, implying higher costs could trigger customer backlash.
  • He cites Deepseek releasing a model with comparable performance at ~one-tenth the cost as an example of commoditization pressure.

---

What he recommends to own in an inflationary regime (portfolio tilt)

Core recommendation: real assets over financial assets

He recommends increasing exposure to real assets, explicitly listing:

• Commodities
• Precious metals (gold)
• Energy (including energy stocks)
• Treasury Inflation-Protected Securities (TIPS)

Regime logic he provides

• When debt is rising and central banks were historically ultra-dovish, real assets can outperform.
• He cites a 5-year lookback where energy and gold/precious metals outperformed the S&P 500, and even tech.
• Under-allocation concern:
  • Energy is mentioned as only ~3% of portfolios (his claim), despite strong relative performance.

---

Capital cycle: the engine behind commodity/real-asset outperformance

Framework

1. Capital flows into a theme/sector.
2. Later, future returns get depressed due to oversupply.
3. Capital leaves → the sector becomes underinvested.
4. Supply shortages emerge → prices rise and returns recover.

His application

• Over the last 10 years, he claims:
  • Real assets (especially mining) have been underinvested.
  • Technology/data centers/AI have attracted massive capital.
• He argues future commodity/energy returns depend on catching early capital turning points.

---

Risks/cautions he emphasizes

• Bond/equity linkage
  • If long-end Treasuries jump (he cites >5%), it can be bearish for stocks due to higher discount rates and volatility.
• Profit-margin risk
  • He suggests profit margin expectations are too high given refinancing cycles:
    • Corporates often refinance around ~five-year windows.
    • He connects today’s refinancing strain to private credit stress, not necessarily defaults yet (with “payment-in-kind” and accounting games mentioned).
• AI bubble timing risk
  • He warns earnings estimates may be “heroic” due to a catch-up that may lag depreciation and operating expense surges.
• Data center/grid/electricity and “local approval” risks
  • Political/regulatory resistance and electricity costs could increase inflationary pressure and constrain rollout.

---

Instruments / tickers / sectors mentioned

Rates / Treasuries

• 10-year Treasury yield
• 30-year (no specific ticker provided)

Sectors

• Energy
• Gold / precious metals
• Commodities
• Tech / semiconductors
• S&P 500
• NASDAQ

Companies referenced (examples)

• Nvidia
• Meta (Yan LeCun mentioned)
• OpenAI
• Anthropic
• Google
• Oracle
• Deepseek
• Microsoft (implied via ecosystem references; not explicitly named as a ticker)

Other instruments

• TIPS (explicitly mentioned)

---

Step-by-step / methodology elements explicitly shared

1) Bond-market “anchor” method

• Track the 10-year Treasury yield (and 30-year closely) as a pricing driver/ripple source across assets.

2) Inflation/interest-rate leading-indicator method

• Use commodity price trends as a leading indicator for interest rates.
• He cites a relationship: the Bloomberg commodity spot index moving about ~6 months ahead of the 10-year Treasury yield direction.

3) Real-assets tilt framework (regime logic)

• In an inflationary regime, own real assets (commodities, precious metals, energy, TIPS) rather than relying solely on S&P 500 and Treasuries.

4) Capital cycle framework

• Identify where capital is flowing now.
• Predict supply/demand impacts:
  • capital inflows → later returns can get compressed
  • underinvestment → supply constraints → price support and outperformance
• He suggests positioning before the capital “flood” arrives (catching the transition).

---

Key numbers & thresholds mentioned

• 10-year Treasury yield: ~4.4% (under 4.5%)
• Breakout/risk levels:
  • >5% expected “pretty easily” in a short period
  • severe scenario: 5.5% → ~6% (likely recession)
• Relationship timing:
  • commodities lead yields by about ~6 months
• Inflation/deficit framing:
  • US deficit ~5–6% of GDP
• Inflation history reference:
  • “three major inflation waves” in the 1970s (historical framing)
• Portfolio allocation claim:
  • energy exposure ~3% of portfolios (his claim)
• AI/earnings risk history:
  • NASDAQ down ~90%+ in 2001–2002
• Data center buildout constraint:
  • only about ~one-third of announced data centers broken ground (his claim)
• LLM accuracy:
  • ~95–96% accuracy vs an implied 99.999% target for “six sigma”
• AI pricing/commoditization:
  • Deepseek cost advantage: about ~one-tenth vs top frontier firms (as stated)

---

Disclaimers / disclosures

• No explicit “not financial advice” disclaimer appears in the provided subtitles summary.
• A promotional disclosure exists for a free portfolio review via “Wealthy” / “Wealthy London,” but it is not presented as a formal investment disclaimer.

---

Presenters / sources mentioned

• Maggie Lake (host)
• Jesse Felder (author of the Felder Report)
• Named external references:
  • Nick Timiraos (Wall Street Journal)
  • Neil Kashkari (Minneapolis Fed)
  • Warren Buffett (Berkshire Hathaway meeting interview)
  • Financial Times (RER Sharma referenced)
  • Reuters (BlackRock portfolio manager referenced)
  • Wall Street Journal (rare-earths / Chinese EV narrative referenced)
  • Jim Pollson (investment relative to GDP piece; not fully identified in subtitles)
  • Garing and Rosenwag (commodity experts referenced)