Marsh's call for the Fed to ease capital rules on insured AI loans reveals a wider truth: regulators are still playing catch-up with how AI infrastructure financing actually works. UK regulated firms cannot afford to wait for rulebooks to catch up. They need to build robust governance now.
Regulatory Watch  Trovix ReachFinancial Services · Insurance

Marsh & McLennan's argument to US regulators is straightforward: allow banks to hold less capital against AI-related loans when those loans are covered by credit risk insurance. This matters urgently to UK insurers, banks and corporate counsel because the same tension exists here. The PRA's existing capital adequacy rules under PRA SS1/23 were designed for traditional lending and commercial real estate. They do not yet reflect the unique risk profile of data centre lending or the speed at which AI infrastructure finance is growing. When a UK bank writes a facility for a new GPU cluster or training facility, the regulatory treatment of insurance-backed hedging remains unclear. Clarity is coming, but not fast enough.

What Marsh is describing is not a one-off complaint. It is a symptom of a structural mismatch between how the financial system hedges new risk and how regulators measure it. Traditional UK banks have preferred interest rate swaps and collateral arrangements over insurance-backed capital relief because the regulatory treatment was understood. But AI infrastructure lending is different. The risks are operational, technological, and regulatory all at once. Insurance products designed to cover these risks — particularly those covering data centre outages, power supply failure, or regulatory shutdown — are emerging fast. Yet the FCA and PRA have not fully integrated insurance-backed hedging into capital frameworks. The result is that banks are hesitant to use the tools that actually exist, and insurers are unsure how their products will be treated by supervisors.

Here is Trovix's honest assessment: the problem is not that AI is too new for regulation. The problem is that regulators are treating AI infrastructure finance as a subcategory of traditional lending when it is structurally different. A regulated firm using off-the-shelf AI tools like Microsoft Copilot or Harvey for contract review gets clear guidance from the ICO and FRC on data handling and audit trails. But a firm trying to structure insurance-backed hedging for AI lending gets silence. This is backwards. The regulatory uncertainty is pushing mid-market firms toward one of two bad choices: either over-capitalize against AI loans (wasting capital) or under-hedge the risks (taking on hidden exposures). Neither is acceptable under the FCA Consumer Duty PS22/9. The answer is not to wait for regulators to catch up. It is to build transparent, auditable decision-making now. Trovix Watch exists precisely to track these regulatory gaps as they close — but firms also need Trovix Audit to document their own governance and capital decisions against AI-related lending so that when rules change, they can demonstrate they acted with care.

If you are a mid-market law firm advising on AI lending, an insurer writing cover for data centre risk, or a bank structuring these facilities, you need three things immediately. First, run a formal audit of how your current AI lending and hedging decisions are being classified under existing capital rules — do not assume your legal interpretations are shared by the PRA. Second, build a process for tracking regulatory change in this space with live monitoring — it will move fast once it moves. Third, document every material decision about insurance, hedging, and capital treatment through an AI governance dashboard that is built for regulatory scrutiny. Marsh is right that the rules will change. But firms that wait for change will be caught unprepared. Those that build governance now will be ready on day one.

Source: Bloomberg News

Related Trovix product:

Trovix Reach →Book a demo →