Goldman Sachs: Credit Spreads Hit Generational Lows, Recommends Three AI Supply Shock Hedging Strategies
Alina Collins
U.S. investment-grade bond issuance in H1 2026 already matches each of the past three full years, yet spreads sit at generational lows — Goldman sees the disconnect as unsustainable and lays out three hedging playbooks.
Issuance has doubled — why haven't spreads moved?
U.S. investment-grade credit issuance hit $1.3 trillion in H1 2026, matching the full-year totals for 2022, 2023, and 2024.
Despite the supply surge, IG credit spreads — the yield premium over Treasuries — remain pinned near generational lows.
This means → the market is pricing as if the supply shock does not exist. Buyers are absorbing everything, but if that bid falters, the room for spread widening is enormous.
How fast is hyperscaler leverage climbing?
Goldman's Adam Crook notes that hyperscalers — Microsoft, Google, Amazon and peers — face an estimated $1.4 trillion funding need by 2027, nearly exhausting their entire operating cash flow.
Total hyperscaler leverage rose from 0.9× in Q3 2025 to 1.8× today — doubling in under three quarters, now exceeding the entire energy sector, and still climbing roughly 0.3× per quarter.
In plain terms = Big Tech is borrowing faster to build AI infrastructure than traditional "heavy-asset" oil companies ever did — and their credit-default swaps (CDS, the market's price tag for default risk) have hit new highs.
Why are spreads and stock prices telling opposite stories?
Goldman frames the current picture: issuance, leverage, and perceived credit risk are all rising — yet overall spreads are falling.
High-yield tech credit spreads have visibly diverged from tech equity prices. Rising stock prices normally signal improving fundamentals and tighter spreads, but the two are now moving in opposite directions.
This reflects an extreme market conviction in the AI narrative — so extreme that credit investors are accepting historically minimal compensation for risk. Goldman argues this is precisely why hedging is needed now.
What are Goldman's three hedging tools?
First — CDX IG payer spread options: Buy the IG46 August-expiry 55/70 payer spread, currently offering a 9.5× total payout ratio. In plain terms = spend $1 in premium, and if spreads widen far enough, you collect up to $9.50 — a small-cost insurance policy against a sudden spread jump.
Second — Goldman 30-year custom basket short: Sell $500 million of the GSIG30AS basket via total-return swap (TRS) while buying $443 million in Treasury swaps, betting that long-end concentration risk surfaces as new issuance hits.
Third — European iTraxx crossover payer spread: Buy the iTraxx Xover S45 August-expiry 275–350 payer spread. European credit implied volatility has fallen to lows, but skew remains elevated — heavy demand for tail options makes this payer spread a high-value hedge.
The "tip of the iceberg" — how much risk is still unpriced?
Goldman warns that public debt markets and disclosed capex represent only the "tip of the iceberg" of the AI super-cycle; roughly $1.8 trillion in off-balance-sheet exposure remains insufficiently priced.
This means → the $1.3 trillion already issued and the $1.4 trillion still to be funded account for less than half the story. Once off-balance-sheet commitments surface, absorption pressure on credit markets steps up again.
Whether the current spread-versus-fundamentals disconnect can persist hinges on whether the AI capex thesis faces a material challenge — if AI returns disappoint, credit repricing will be far more violent than equity repricing.
Content is for reference only, not financial advice.