What's More Likely: A Rally Or Rout From Here? | New Harbor Financial

Summary of main arguments and reporting

• AI-driven market valuations may be based on “bubble” assumptions. Host Adam Tagert leads a discussion with New Harbor’s partners about a potential reckoning in AI-related equities and capex spending. A key theme is that market prices are discounting a near-permanent earnings surge, akin to historical bubbles—only now the “bubble” may be showing up not just in asset prices but also in earnings expectations.

• Fred Hickey’s warning (as discussed by the hosts) is that earnings are overstated. Adam recounts Fred Hickey’s argument that hyperscalers’ apparent profits look strong partly because depreciation and future costs aren’t fully reflected in current earnings, due to heavy data-center capex. When adjusted, forward valuation multiples could be far higher than commonly believed (Hickey’s figure cited: ~65x on adjusted terms).

• Earnings growth forecasts are extreme vs. history. John Lodra presents charts suggesting that analyst projections for earnings growth over the next several years are the most aggressive in decades—even exceeding peaks around the tech bubble (with context that COVID affected the baseline). Another chart compares the S&P 500’s projected earnings to a cyclically adjusted framework (CAPE-like), showing a parabolic/record deviation.

• Semiconductors/hardware are treated like “utilities,” but are still cyclical. The hosts argue that bullish narratives that chips and AI infrastructure have become permanently high-demand “staples” are likely overstated. They also point to AI/data-center constraints (compute costs, infrastructure rollout limits, memory/supply considerations, land/water/permits, etc.) as limiting the ability of the buildout to scale smoothly.

• AI capex spending may be outpacing returns, pressuring cash flow and credit. Mike Preston emphasizes that free cash flow has dropped while companies have ramped spending. Examples cited include large debt issuance (e.g., Google raising $80B in debt) and increasing credit risk indicators such as CDS moves and credit spreads. The implication: if monetization lags, financing costs and growth prospects deteriorate.

• The “AI economics” may be breaking down—compute is getting more expensive. The hosts argue that the cost of compute is rising rapidly while benefits aren’t appearing at the same pace. They mention real-world effects such as firms slowing hiring or reshoring staff after plans to fully automate with AI didn’t pan out as expected, and shifting toward cheaper model alternatives.

• Cheaper models (including from China) could erode Western AI business models. Mike argues that China’s constraints on leading chips may push it toward more compute-efficient software, and that large language models from China may deliver much of the desired output at far lower cost—threatening margins/pricing for higher-cost Western AI providers.

• Market positioning is extremely concentrated in the AI trade. Adam notes that AI-related stocks (chips + AI ecosystem) represent a large share of major indexes, so any downward repricing of AI earnings expectations could spill over broadly into the entire market and economy (via lost capex-driven GDP growth).

“Rally or rout?”—their market outlook

• They do not claim certainty, but expect potential blowoff/rip higher first, then rollover.

  • Mike says if the AI trade breaks, the whole market could break, but he’s watching for signs of broadening beyond mega-cap tech/chips.
  • He suggests a likely path could be a blowoff move (“fireworks”)—a rapid rally—followed by a decline/rollover, rather than an immediate straight drop.

• Need “breadth” to sustain highs. Mike and John discuss that while the S&P 500 has been near highs, leadership has largely been chips/semis. They argue that for a further advance to be durable, other sectors (not just AI/chips—e.g., banks/healthcare/homebuilders/financials) must start breaking out too. Early signs are described as weak/early.

• If this turns into a “com bubble” style bust, downside could be severe. Mike gives a numeric risk estimate: if the market reprices aggressively, he expects >50% drawdown, roughly 60–80% as a plausible magnitude based on his historical framing (illustrated as a move from ~8,000 down toward the low 3,000s).

• Impact on inflation vs. deflation. When asked what would happen to inflation in such a bust, Mike argues deflation (and disinflation turning into downturn dynamics) would be a first-order concern, since hyperscaler capex and overall economic growth would likely contract sharply.

Long bear market / “lost decade” risk framing (retirees in particular)

• John argues “lost decade” risk is real even if bear markets aren’t always long in calendar time.

  • He says historically major drawdowns can be brief, but can still be devastating in real (inflation-adjusted) terms—especially for retirees and near-retirees.
  • He critiques passive “set-and-forget” stock-heavy portfolios in this context, suggesting many investors could face a decade of low/negative real returns.

• He warns about behavioral traps.

  • Even when investors intend to reduce risk, they often get trapped by FOMO or refusing to sell after declines, waiting for prices to recover to previous levels.

Action recommendations: de-risking and risk management

• Their practical advice is “derisk / rebalance” and use hedging rather than hope.

  • John suggests that for passive investors approaching retirement, an equity allocation around ~30% stocks (vs. 60–80%) may be more appropriate in valuation-stretched conditions.
  • They emphasize this is not perfect timing, but tactical risk management to avoid catastrophic drawdowns.
  • They highlight hedging/option-based risk tools (presented as defensive, not speculative) as part of their approach.

• Psychology matters as much as strategy. Mike frames this as a “financial war” against uncertainty created by central bank regime changes and the psychological effect of market narratives/wealth effects. The practical takeaway: reduce risk enough that you can “sleep” through volatility and avoid emotional mistakes.

Precious metals update (gold/silver)

• Mike says the precious metals charts remain weak.
  • He frames the pullback as bigger and longer than expected and says technical signals look sick: below key moving averages, with potential “death cross”/bearish structure on silver.
  • Despite a modest reaction to inflation news, he suggests the charts haven’t yet broken downtrend levels that would signal a durable recovery.
  • He does not call it “fatal,” but portrays it as testing faith and still needing confirmation.

Inflation discussion (shelter / shelter disinflation)

• They discuss that CPI dynamics are heavily driven by shelter, which has been disinflating, making it harder for inflation to re-accelerate sharply—unless oil rises materially (e.g., above $100/bbl was mentioned as a possible upside risk).

Presenters / contributors

• Adam Tagert (host; “Thoughtful Money” founder)
• Mike Preston (New Harbor Financial; lead partner)
• John Lodra (New Harbor Financial; lead partner)
• Fred Hickey (referenced; “High-Tech Strategist,” discussed via prior conversation/interview)
• Ed Yardini (referenced; chart contributor)