The European Central Bank's November financial-stability analysis of AI puts shared technology suppliers alongside model risk. Separate financial applications can depend on the same cloud infrastructure or model provider, leaving institutions exposed to correlated disruption.
Source-backed brief: The analysis concerns banks, asset managers and the wider economy, with a euro-area focus. It neither measures US credit-union dependencies nor creates a US supervisory requirement.
The dependency behind the contract
Section 4.1 describes how concentrated cloud and model supply can increase switching costs and transmit outages, model changes, pricing changes or API restrictions across institutions. It also recognizes the benefits of large providers' infrastructure and security resources. Concentration is an exposure to understand, not proof that a particular supplier is unsafe.
For a credit-union board, the useful distinction is between the company signing the contract and the infrastructure required to keep its service working. Counting vendor names alone cannot answer the continuity question.
A shared failure path
Consider a hypothetical credit union using one provider for employee document search and another for a member-service assistant. If both rely on the same upstream model endpoint, a second contract may not provide the independent fallback management expects. This is an illustration, not a reported outage or a finding about a named credit union.
A usable dependency record would connect each critical service to its upstream model, hosting region and fallback arrangement, then distinguish a documented alternative from a merely promised one. An owned manual service route may matter more during a disruption than another supplier logo.
The immediate decision for technology and strategy leaders is whether their continuity evidence reaches past the direct vendor. The ECB paper supplies a reason to ask; only the credit union's contracts, architecture and tested operating arrangements can establish its own exposure.