Macro & ratesGoldman Sachs Asset Management Standout
Investment Views from Our CIOs
AI infrastructureInterest ratesSemiconductors
Our summary
The global economy has moved from late-cycle conditions back into expansion, but full valuations and uncertain inflation, rates, and fiscal risks make relative-value allocation preferable to broad market exposure.
- AI infrastructure remains a durable cross-asset opportunity even if open-weight Chinese models and closed-weight US models coexist, although software winners will depend more heavily on the prevailing model regime.
- Chinese AI development is constrained mainly by compute, while US infrastructure faces power availability constraints; competitive pressure is expected to be strongest in memory, whereas US AI accelerators and leading-edge Taiwanese foundries appear better insulated.
- AI-related debt is expected to reach $2 trillion by 2028, with investment-grade hyperscaler bonds, leveraged neocloud credit, and off-balance-sheet data-center or chip financings representing distinct risk profiles.
- Hyperscaler investment-grade spreads are viewed as widening mainly because of heavy and unpredictable issuance, while off-balance-sheet financing could rise from less than $40 billion in 2025 to more than $850 billion in 2028, or about 44% of total AI debt outstanding.
- The Fed, ECB, and BoJ all raised rates in the same month, and rising yields alongside anchored inflation expectations suggest markets may be repricing the neutral rate; France's spring 2027 election adds to long-end fiscal risk.
Takeaway: Maintain conviction in AI infrastructure and selectively source income in high-yield, securitized credit, and structured AI debt, but remain cautious on broad rates and apply disciplined relative-value analysis to collateral, structure, and liquidity.
Published by Goldman Sachs Asset Management. Summary by HF Best Ideas.