Finance Connect Podcast: Bernie Margulies on the risks and opportunities of financing the AI boom

As investment in artificial intelligence accelerates, concerns are mounting over the levels of debt being taken on to finance the infrastructure powering AI.

Morgan Stanley estimates that AI-linked debt issuance in the first nine months of 2026 was more than four times higher than during the same period in 2025, highlighting the scale of investment and raising questions about how lenders can manage their exposure to this rapidly developing market.

So what does this mean for equipment finance providers looking to finance the technology behind AI?

American Compute was recently featured in the Financial Times, in an article examining how Nvidia is exploring insurance solutions to help lenders manage the risks associated with financing AI infrastructure, particularly for newer cloud computing businesses known as NeoClouds.

With hyperscalers and NeoClouds continuing to invest heavily in computing capacity, financing these assets presents particular challenges around creditworthiness, equipment values and the pace of technological change.

American Compute specialises in financing solutions and risk mitigation for data centre equipment, with a particular focus on GPUs. Margulies explains how the business works with lenders and insurers to structure transactions and manage risks associated with financing this emerging asset class.

A central theme of the discussion is the role of residual value insurance in supporting GPU financing. Margulies outlines how insurance can help protect lenders against uncertainty over future equipment values, alongside other risk-management solutions including backup operators and lost revenue protection.

The conversation also explores how techniques traditionally associated with project finance are being adapted to support GPU financing, with lenders increasingly focused on contracted revenues, cash flow and the ability of borrowers to meet their financial obligations.

Another important consideration is the challenge of forecasting GPU residual values. Drawing comparisons with previous developments in technology leasing, Margulies explains why historical data alone may not provide a reliable guide to future asset values, particularly as new generations of chips and data centre infrastructure emerge.

Despite the potential opportunities, Margulies offers a note of caution to equipment finance providers considering entering the market. Rather than rushing to participate in the AI investment boom, he encourages lenders to take time to understand the risks, assess individual transactions carefully and identify opportunities that fit their risk appetite.

Listen to the full conversation on the Finance Connect European Equipment Finance Podcast, sponsored by Alfa, as well as Apple Podcasts, Spotify, YouTube and Amazon Music.