Executive Summary

The macro landscape since our last brief (Feb 18, Brief #8) has entered a consolidation phase where every thesis is simultaneously being tested and reinforced by new data. Gold’s recovery above $5,100 after the $4,900 correction we used as a buying opportunity validates NEM as maximum conviction — the pullback we identified as “shaking out leveraged speculators” played out exactly as described. NVDA’s Meta deal expansion remains the strongest possible pre-earnings signal for Wednesday. The Iran escalation that we flagged as reversing diplomatic progress continues to build, with VP Vance keeping military options explicitly open. And the AI disruption sell-off has now gone genuinely global, with Indian and European markets experiencing the same structural repricing we identified as having roots in 2022.

The new developments worth specific attention: (1) Gold at $5,100 vs. $4,900 in our prior brief represents a meaningful upgrade to NEM’s FCF assumptions and validates that the correction was temporary, as we predicted. (2) The convergence of business investment recovery (AI-driven capex rebound) with industrial semiconductor strength (ADI earnings beat) signals that the “long atoms” thesis is broadening from a commodity trade to an industrial cycle recovery story. (3) Housing data provides the clearest possible confirmation of the DHI thesis — starts at a five-month high while existing sales collapse 8%, empirically proving that builders ARE the market. (4) Cleveland Fed Hammack’s “hold for quite some time” vs. Einhorn’s “substantially more than two cuts” creates the widest rate-path uncertainty range we’ve seen, with PCE Friday as the arbiter. (5) Anthropic facing potential supply chain risk list placement reshapes the AI competitive landscape in ways that benefit PLTR and MSFT at Anthropic’s expense.

The framework is performing. The pair trades are working. The tactical adjustments from Brief #8 — PANW downgrade, airline conviction reduction, defense overweight — are all confirmed by subsequent data. The question for this week is binary: does NVDA validate the AI infrastructure thesis on Wednesday, and does WMT confirm the consumer isn’t cracking on Thursday? Everything else is secondary.

Key Events & Analysis

Gold at $5,100: The Correction We Bought Is Already Over

Prior call tracking: In Brief #8, we wrote gold at $4,900 was a buying opportunity driven by temporary factors (CME margin hikes, Chinese holidays). Gold has recovered above $5,100 within days. This is our cleanest call in the series.

Goldman’s research adds structural reinforcement: central bank buying as the primary driver, with governments extending commodity stockpiling beyond gold into agricultural and energy markets. This validates our CF Industries, ADM, and BG positions in the agricultural input complex. The stockpiling thesis is broadening from gold into a multi-commodity sovereign hoarding pattern that has historical precedent only during pre-war periods — consistent with Dalio’s “end of post-war order” framework from the Munich conference.

NEM’s FCF at $5,100 gold is approximately 15-20% higher than at $4,900. The miner-to-bullion ratio remains historically compressed, meaning gold equities have not yet repriced to reflect the commodity move. This creates additional upside as the equity market catches up to the commodity price.

The risk we flagged in Brief #8 — that gold is the most crowded institutional trade — remains live. But fund managers being “uber-bullish” while simultaneously hoarding cash and gold is the textbook setup for a volatility event that resolves in gold’s favor: the cash gets deployed, and gold’s safe-haven bid intensifies during any equity sell-off.

NVDA Wednesday: Meta Deal Converts Narrative to Evidence

The Meta deal expansion is analytically important for a reason most commentators are missing. Meta committing to “millions” of NVDA chips is a contractual revenue commitment that flows through to NVDA’s Q2 and Q3 guidance. This means NVDA can guide UP even if non-Meta hyperscaler demand is flat — a scenario that would be interpreted as broadly bullish but is actually concentrated in a single customer relationship.

NVDA exiting its entire ARM stake simultaneously is the confidence signal we predicted in Brief #8: GPU demand is so overwhelming that CPU architecture diversification is irrelevant to NVDA’s near-term business. NVDA management doesn’t need the strategic hedge anymore.

AMD and AVGO falling on the news deserves emphasis. Meta chose to consolidate around NVDA rather than diversify across GPU suppliers (AMD) or invest in custom ASICs (AVGO). This is a customer behavior data point that is more valuable than any analyst report. When the customer with the most AI compute expertise in the world chooses NVDA exclusively, the competitive moat is wider than the market prices.

NVDA miss probability: 15%, down from 20%. The Meta commitment provides a revenue floor that didn’t exist before. But a miss DESPITE this commitment would be more violent because expectations are now higher.

Iran Escalation: The Portfolio’s Most Dangerous Macro Scenario

The Iran escalation from Brief #8 continues to build. VP Vance keeping military strikes on the table while oil jumped 2%+ creates a bifurcated oil outlook: Saudi flooding the market at 5-year low prices (bearish base case) vs. Strait of Hormuz disruption (very bullish tail case). These forces cannot coexist indefinitely — one resolves into the other.

The second-order impact is what matters for our portfolio. If oil sustains above $80:

  • CPI trajectory reverses → rate cuts repriced from 2-3 to 0-1
  • DHI, VICI, LEN face indirect but material headwinds from mortgage rate expectations
  • Airlines lose the fuel cost advantage that supported our triple-tailwind thesis
  • Gold benefits from both geopolitical uncertainty AND inflation persistence

Our Brief #8 adjustments (airline conviction reduced, defense overweight) are exactly right for this scenario. The new question is whether to add explicit oil-upside hedging via XLE calls as a tail-risk position. With Saudi actively suppressing prices, the premium on these options should be cheap, making the asymmetry attractive.

Business Investment Rebound: The Industrial Cycle Inflection We Called

The MarketWatch report that AI-driven business investment is powering economic growth confirms our Brief #8 addition of ADI and CAT as new positions. The growth composition shift from consumer spending to capex has three implications:

First, it validates Hammack’s “hold for quite some time” stance — if capex is driving growth, the economy doesn’t need rate cuts. Einhorn’s aggressive cut thesis depends on consumer weakness being sufficient to override capex strength. The data currently favors Hammack.

Second, it means NVDA, DELL, and EQIX earnings are literally GDP data points. Their revenue growth IS economic growth in the current composition. This gives NVDA Wednesday added macro significance beyond the company-specific story.

Third, it supports the industrial semiconductor cycle inflection (ADI, TXN, NXPI) that we added to watchlists in Brief #8. If business investment is rebounding broadly, analog chip demand follows with a 1-2 quarter lag.

Credit Market: GOOG’s 100-Year Bond Is the Most Underappreciated Event

The credit data continues to reinforce our framework from Briefs #7 and #8. But I want to emphasize GOOG’s 100-year bond at 10x oversubscription, which the market has not fully processed for its strategic implications.

GOOG has locked in a century of AI funding at the lowest real cost of capital in its history. No other company — not Amazon, not Microsoft, not Apple — has this balance sheet flexibility. GOOG can now fund AI projects with 20-30 year payback periods because its annual cost of servicing this debt is negligible relative to its cash generation. The competitive advantage compounds: projects that require patient capital (quantum computing, autonomous vehicles, drug discovery) become viable for GOOG and not for competitors who fund AI with shorter-duration obligations.

This strengthens our GOOG/INTU pair trade from Brief #8. GOOG buys the future at nearly zero cost; INTU faces AI disruption that erodes its present.

Housing: The Empirical Proof

Housing starts at a five-month high while existing sales collapse 8% is the single cleanest data point supporting our DHI thesis. Builders are filling the inventory vacuum created by the rate-lock effect. The math: at 6.12% mortgage rates, someone locked in at 3.5% faces a ~70% increase in monthly payments on equivalent housing. This rate-lock doesn’t break until mortgage rates drop below 5%, requiring Fed funds below 3%, which is a 2027+ dynamic per our analysis.

DHI at 14x P/E with empirically confirmed monopoly-like competitive positioning in the most important asset class in the economy remains the most mispriced large-cap in our coverage universe after gold miners.

Anthropic’s Government Trouble: PLTR Benefits

Defense Secretary Hegseth considering placing Anthropic on a supply chain risk list would redirect government AI contracts to competing providers. PLTR, as the established government AI platform, is the primary beneficiary. Microsoft’s OpenAI relationship also benefits. GOOG, as Anthropic’s largest investor, faces mixed implications — some contracts redirect to Google Cloud, but the investment’s value declines.

At 199x P/E, PLTR doesn’t need new catalysts to justify caution on valuation. But Anthropic’s troubles remove a competitive threat that was real.

Portfolio Implications

Positions Confirmed by New Data

NEM — Maximum conviction FURTHER STRENGTHENED. Gold at $5,100 vs. $4,900 in prior brief. Goldman confirming central bank buying as primary driver. European defense bonds weakening dollar. Iran escalation. Every driver converging. Add on any pullback below $5,000.

NVDA — Maximum conviction maintained. Meta deal provides contractual revenue floor. ARM exit is confidence signal. Wednesday earnings remain portfolio-defining. Size for 15% drawdown.

DHI — High conviction STRENGTHENED. Housing starts at 5-month high + existing sales -8% = empirical proof builders ARE the market. 14x P/E.

CRWD — High conviction maintained. PANW guidance cut confirms CRWD as sole cybersecurity conviction name. Sympathetic selling creates entry.

GOOG — High conviction STRENGTHENED. 100-year bond competitive advantage deepens with each passing day of AI investment.

APO — Near-maximum conviction. 2% real bond returns + credit spreads at 1998 lows = mathematical certainty of private credit migration.

Defense (LMT, RTX, NOC) — Overweight confirmed. France-India $40B Rafale + Japan $36B + Iran escalation + European defense bonds.

New Additions This Brief

MP Materials (MP) — Rare earth prices above government floor price. Critical mineral independence thesis aligned with defense and Japan investment themes.

ADI — Industrial semiconductor recovery confirmed by broader business investment rebound. Upgrade from watchlist to buy.

DPZ — Buffett increasing stake provides floor signal. Franchise model + delivery infrastructure = AI-resistant consumer play.

Key Pair Trade Updates

All nine pair trades from Brief #8 remain active. The GOOG/INTU pair is strengthened by the 100-year bond analysis. The NEM/CRM pair is strengthened by gold at $5,100 + Mistral CEO’s 50% enterprise software replacement claim. The MPC/DVN pair needs monitoring given Iran oil dynamics.

Critical Events This Week

Day Event Portfolio Impact
Wed NVIDIA earnings Meta deal de-risks. Expect beat-and-raise. THE event.
Wed Fed minutes Waller dissent language. Balance sheet debate.
Thu Walmart earnings Consumer bifurcation test. OTC drug tailwind.
Fri PCE data Inflation trajectory vs. Fed study skepticism. Iran oil risk compounds.

Risk Scenarios

Risk 1: NVIDIA Misses Despite Meta Deal (15%). Miss probability reduced by Meta commitment but consequence increased because expectations are higher. If NVDA acknowledges Chinese AI pricing pressure, reprices the entire medium-term margin outlook.

Risk 2: Iran Military Action (15-20%). VP Vance’s rhetoric is a policy signal. Oil above $80 kills rate cuts, crashes DHI/VICI. NEM provides natural hedge. Defense positions benefit.

Risk 3: Credit Cascade (25%). $63B near-junk IG at 1998 spreads. A single high-profile downgrade triggers forced index selling. APO benefits structurally.

Risk 4: PCE Friday Shows Inflation Persistence (25%). Combined with Fed study skepticism, a hot PCE print could reprice rate path from 2-3 cuts to 0-1. Rate-sensitive positions (DHI, VICI) most exposed. Hammack’s “hold” stance validated.

Risk 5: Coordinated Institutional Bearishness Self-Fulfills (20%). Grantham, Constan, Dalio, Tepper all expressing US equity skepticism. If institutional de-risking begins, EM rotation accelerates and US large cap flows decline.

Risk 6: Gold Correction Resumes (15%). Fund managers crowded in gold. CME margin hike risk. But seven structural drivers and $5,100 momentum suggest pullbacks are buying opportunities.

$10,000 Model Portfolio

Ticker Company Allocation ($) Shares Thesis
NEM Newmont Corporation $2,000 36 Maximum conviction gold miner at $5,100 gold with seven converging structural drivers and compressed miner-to-bullion ratio
NVDA NVIDIA $2,000 11 Meta deal provides contractual revenue floor ahead of Wednesday earnings; 40x fwd P/E at sentiment low
APO Apollo Global Management $1,500 10 2% real bond returns mathematically force institutional migration to private credit; 16x vs BX 54x
DHI D.R. Horton $1,200 8 14x P/E with housing starts at 5-month high while existing sales collapse 8% — builders ARE the market
CRWD CrowdStrike $1,000 3 Sole cybersecurity conviction name after PANW guidance cut; AI agent proliferation increases attack surface
GOOG Alphabet $1,000 5 100-year bond at 10x oversubscription creates permanent cost-of-capital advantage for AI investment
LMT Lockheed Martin $800 2 Defense sector secular growth: France-India $40B deal, Japan $36B US investment, Iran escalation, European rearmament
FCX Freeport-McMoRan $500 12 Japan $36B critical mineral investment + copper supply tightness + gold thesis cross-correlation

Portfolio construction logic: This portfolio expresses the “long atoms, short software” thesis through concentrated positions in physical asset producers (NEM, FCX), AI infrastructure (NVDA, GOOG), alternative asset management (APO), physical housing (DHI), cybersecurity (CRWD), and defense (LMT). The portfolio is deliberately overweight commodities and physical assets (30%) relative to a typical equity allocation because the convergence of stealth QE, de-dollarization, sovereign stockpiling, and defense spending creates a multi-year secular bid for physical assets that most portfolios underweight.

NEM and NVDA each receive $2,000 as the two highest-conviction positions with asymmetric risk-reward: NEM benefits from seven converging macro forces with gold at $5,100, while NVDA has the Meta contractual commitment de-risking Wednesday’s earnings. APO at $1,500 captures the institutional migration to private credit that Apollo’s own research proves is mathematically inevitable. DHI at $1,200 exploits the housing oligopoly mispricing confirmed by this week’s data. CRWD, GOOG, LMT, and FCX are smaller positions reflecting high conviction with slightly wider outcome distributions. The portfolio hedges its own risks: NEM and LMT benefit from the Iran escalation scenario that would pressure DHI through rate-cut repricing. NVDA benefits from the AI capex cycle that creates the disruption pressure on software (not held). The key exit trigger would be NVDA missing Wednesday with downward guidance despite the Meta deal, combined with a hot PCE print Friday — this double failure would force a reassessment of both the AI infrastructure and Goldilocks theses simultaneously.

Changes from prior model portfolio: This is the first explicit model portfolio. Compared to the positioning described in Brief #8, the notable addition is LMT (defense overweight thesis) and FCX (Japan critical mineral catalyst), while we exclude the pair-trade short legs (CRM, BX, ACN) that would require a more complex portfolio structure. In a long-only $10,000 allocation, concentrating on the highest-conviction longs maximizes the thesis expression.