Could rising AI debt derail the investment boom?

The AI buildout, which has already absorbed $1.6 trillion in capital since 2023, faces growing financing risks. While demand for compute remains exceptionally strong, with projections of over $1 trillion in annual spending, the expansion is increasingly reliant on debt markets.

This shift is pressuring specialized "neoclouds," whose capital-intensive models depend on upfront financing against long-term revenue. Higher borrowing costs, concentrated customer bases, and rapidly depreciating GPUs make them vulnerable. We're seeing this reflected in credit default swaps for providers like CoreWeave approaching 1,200 basis points.

However, the larger hyperscalers remain well-positioned with strong balance sheets and substantial cash flow, enabling them to absorb higher borrowing costs. The interconnectedness of the AI ecosystem also provides support, as chip suppliers, hyperscalers, and AI developers have incentives to keep projects moving. Investors should monitor the cost of capital and any signs of weakening compute utilization or customer commitments, as these could signal broader issues.