Global debt surged by about $29 trillion in 2025, marking the biggest annual rise since the pandemic era, and pushing total world debt to a new record of $348 trillion, according to the Institute of International Finance (IIF).
The IIF links the surge to governments and companies spending heavily on national security and the accelerating AI infrastructure buildout, with AI-related investment emerging as a fresh driver of borrowing and capital markets activity.
AI and defense spending are reshaping the debt cycle
The IIF notes that public borrowing did most of the heavy lifting in 2025, as governments expanded deficits and issued more sovereign debt to fund security priorities and strategic tech programs.
At the same time, Big Tech’s AI arms race keeps pulling forward massive capex for data centers, chips, and energy capacity. That “capex supercycle” strengthens demand for debt financing even when rates stay elevated, because these projects require upfront spending at a scale equity can’t always cover efficiently.
U.S. bonds: AI capex becomes the big issuance catalyst
On the corporate side, analysts expect AI to be a major reason bond supply keeps climbing. A Reuters report citing Barclays analysts projects U.S. corporate bond issuance rising in 2026, with AI hyperscaler capex flagged as the biggest upside risk for even more jumbo deals than normal.
This isn’t abstract. Oracle, one of the major infrastructure financiers tied to the AI boom, has laid out plans to raise tens of billions to fund cloud expansion and AI-related capacity.
Why “AI bubble risk” stays on the table
The risk case is straightforward: debt-funded capex can overshoot real demand. If growth cools or financing conditions tighten, markets can see valuation stress, refinancing pressure, and sudden pullbacks in spending—especially where multiple tech giants rely on interconnected, circular dealmaking to justify ever-larger buildouts.
Eco-friendly SEO angle
Debt-fueled AI expansion also raises an energy question. If the next wave of borrowing funds efficient data centers, cleaner power, and higher utilization, it can reduce waste per unit of compute. If it funds overbuilt capacity and low-value workloads, it increases emissions and stranded infrastructure. The sustainability win depends on efficiency, clean energy sourcing, and realistic demand planning.
