The WSJ's Sam Goldfarb (26 Sep) frames the debate in one line: "Investors panicked in 1994, cheered in 2016." With the 10-year at 5.18%, its highest since 2007, and the S&P 500 still up 12.6% since the Iran war began on 28 February, here is our read.
1. A real-rate shock, not an inflation scare. Since 27 Feb the 10-year is up 121bp, of which 113bp is the real yield (2.85%, highest since 2008). Inflation breakevens barely moved. Three borrowers are competing for the same savings: the Treasury (net interest now above $1tn a year), the AI hyperscalers, and a world repricing its own debt (German Bunds at 17-year highs).
2. AI has flipped from self-funded to externally funded. Alphabet bought back $28bn of stock in H1 2025 and none in H1 2026; in June it raised $49.6bn of new equity. Meta stopped buybacks and borrowed $24.9bn. Capex builds capacity, which buybacks never did, and faster growth is the only painless exit from the US debt arithmetic. But capex lifts rates before it lifts growth.
3. History: the macro rhymes with 2006 (oil shock, Fed hiking, housing rolling over), the market structure with 1999 (ten AI-linked names are 40% of the S&P 500). The difference with 1999 is earnings: 19x forward, with Q3 EPS expected +29%. And it is not 2022: credit spreads are tighter than before the war.
4. The US will not "collapse". A credibility crisis looks like yields up, dollar down, gold up. We see the opposite: since the war began the dollar is +3.4% and gold -18%. The risks are dated: Iran after the 3 Nov midterms (the WSJ reports Trump expects to resume bombing after the vote), the 11 Dec funding deadline, and any hyperscaler bond deal that struggles.
5. Own earnings, not duration. Favour AI that funds itself (Microsoftpple) over debt-reliant AI (Oracle, CoreWeave); energy and US LNG as insurance (ConocoPhillips, EOG, Cheniere); large-cap pharma and biotech on M&A (Merck, AbbVie, IBB rather than XBI); front-end Treasuries; gold as the tail hedge. Avoid utilities, homebuilders, REITs and small caps. Moderna (+574% YTD) is a clinical re-rating, not a macro trade: the binary is the full melanoma data at ESMO on 24 Oct.
Scenarios to year-end (our probabilities): base 55%, index flat to +5%; bull 25% (Iran deal, Brent towards $80), +8 to +12%; bear 20% (escalation or an AI funding accident), -10 to -15%.
Verdict: constructive, conditionally. Stay invested, rotate, and watch the dollar and hyperscaler credit, not the S&P, for the turn.
Chart attached. For professional investors only; not investment advice.
Best regards,
Laurent Chekroun
1. A real-rate shock, not an inflation scare. Since 27 Feb the 10-year is up 121bp, of which 113bp is the real yield (2.85%, highest since 2008). Inflation breakevens barely moved. Three borrowers are competing for the same savings: the Treasury (net interest now above $1tn a year), the AI hyperscalers, and a world repricing its own debt (German Bunds at 17-year highs).
2. AI has flipped from self-funded to externally funded. Alphabet bought back $28bn of stock in H1 2025 and none in H1 2026; in June it raised $49.6bn of new equity. Meta stopped buybacks and borrowed $24.9bn. Capex builds capacity, which buybacks never did, and faster growth is the only painless exit from the US debt arithmetic. But capex lifts rates before it lifts growth.
3. History: the macro rhymes with 2006 (oil shock, Fed hiking, housing rolling over), the market structure with 1999 (ten AI-linked names are 40% of the S&P 500). The difference with 1999 is earnings: 19x forward, with Q3 EPS expected +29%. And it is not 2022: credit spreads are tighter than before the war.
4. The US will not "collapse". A credibility crisis looks like yields up, dollar down, gold up. We see the opposite: since the war began the dollar is +3.4% and gold -18%. The risks are dated: Iran after the 3 Nov midterms (the WSJ reports Trump expects to resume bombing after the vote), the 11 Dec funding deadline, and any hyperscaler bond deal that struggles.
5. Own earnings, not duration. Favour AI that funds itself (Microsoftpple) over debt-reliant AI (Oracle, CoreWeave); energy and US LNG as insurance (ConocoPhillips, EOG, Cheniere); large-cap pharma and biotech on M&A (Merck, AbbVie, IBB rather than XBI); front-end Treasuries; gold as the tail hedge. Avoid utilities, homebuilders, REITs and small caps. Moderna (+574% YTD) is a clinical re-rating, not a macro trade: the binary is the full melanoma data at ESMO on 24 Oct.
Scenarios to year-end (our probabilities): base 55%, index flat to +5%; bull 25% (Iran deal, Brent towards $80), +8 to +12%; bear 20% (escalation or an AI funding accident), -10 to -15%.
Verdict: constructive, conditionally. Stay invested, rotate, and watch the dollar and hyperscaler credit, not the S&P, for the turn.
Chart attached. For professional investors only; not investment advice.
Best regards,
Laurent Chekroun