Analysts have assessed the financial risks of the GPU infrastructure market

Analysts presented a report on the state of the market of specialized cloud providers, whose business model critically depends on the purchase of equipment from NVIDIA. The main problem is the mismatch between the schedules of debt repayment and the rate of depreciation of computing power, which serves as collateral for loans. The company CoreWeave was in the spotlight, whose strategy of aggressive expansion of the GPU fleet relies on debt financing. According to the study, the financial stability of such "neoclouds" is based on contracts with two main clients - OpenAI and Anthropic. In the event of a decrease in demand for rental of capacities from these companies, providers risk facing a liquidity shortage to service loans. The situation is complicated by the fact that the cost of graphics processors on the secondary market is falling faster than assets are depreciated on the companies' balance sheets. This creates a "domino effect": when the prices for renting computing resources fall, the value of the collateral decreases, which requires providers to make additional capital injections or restructure debts. Experts identify three scenarios for the development of events: from the consolidation of the market under the control of large technology corporations to the massive write-off of assets in the event of a correction in demand for training large language models. The current model of financing AI industry infrastructure is characterized by a high concentration of risks, where the stability of the entire supply chain depends on the solvency of several model developers.