UBS: AI Capital Expenditure Is Reshaping the Logic of the Global Credit Cycle
0xBroomberg
The global credit impulse has risen to +1.3% of GDP, driven almost entirely by the US — where the engine is not the bond market but bank lending tied to AI infrastructure.
What does a +1.3% credit impulse actually mean?
The credit impulse — the change in new credit as a share of GDP over the past 12 months — now reads +1.3%. The global economy is borrowing at an accelerating pace.
This means → demand momentum in the real economy is picking up. This indicator correlates with global real demand growth at roughly 0.6 — a widely tracked leading signal.
But nearly all the improvement comes from developed markets, and within DM it is almost entirely a US story. The US alone contributed +2.6%, or roughly $800 billion in additional credit.
Who is accelerating and who is slowing?
Japan shows the strongest momentum, with a credit impulse of +5.6% of GDP.
Canada is the only major economy where credit growth is decelerating.
Emerging markets are flat to weakening at -0.3% of GDP — China, Russia, Brazil, and Mexico all contribute negatively.
What is actually driving US credit expansion?
The driver is bank lending, not bond issuance — roughly 45% from household loans and 55% from corporate loans.
The single largest source of corporate loan growth is AI infrastructure buildout. US credit growth has re-accelerated from about 1.8% year-on-year at mid-2025 to roughly 4% now.
In plain terms = hyperscale cloud companies — Google, Microsoft, Amazon — are borrowing heavily to build data centers, and that money is flowing through the banking system rather than the bond market.
Why did the bond market "go missing"?
UBS flags a statistical anomaly: hyperscalers issued roughly $300 billion in net new investment-grade bonds, so the bond channel should show a positive contribution — yet the data shows it at near zero.
This means → AI companies are issuing bonds, but other forces are offsetting the credit gain across the broader bond market. The banking system is doing the heavy lifting.
One more number to watch: hyperscalers alone are projected to issue $600 billion in debt in 2026, while the trillions in off-balance-sheet obligations they have previously disclosed appear to remain largely off investors' radar.
What happens if the AI credit wave reverses?
UBS's core warning: AI-linked financing is now both the primary driver of US credit expansion and the single largest positive contributor to the global credit impulse.
This means → if AI capex slows or reverses, the biggest positive line item in the global credit impulse flips negative — a direct drag on global growth.
This reflects a deeper fragility: the S&P 500's entire year-to-date gain is concentrated in the top ten companies. The AI narrative props up both equities and credit expansion — any doubt about AI investment returns could hit capital markets and real-economy credit simultaneously.
Content is for reference only, not financial advice.