Data Center CMBS Boom Raises AI Infrastructure Risks
Data center commercial mortgage-backed securities (CMBS) lending has surged from below $500 million before 2020 to an estimated $27 billion–$30 billion in 2025. JPMorgan expects annual issuance to reach $30 billion–$40 billion in 2026 and 2027, or about 7%–10% of combined CMBS and asset-backed securities issuance.
The rapid expansion, driven by artificial intelligence infrastructure demand, is introducing new risks for CMBS investors. Atrium estimates that $128 billion in US data-center debt will mature between 2025 and 2028, rising to $213 billion by 2029. That maturity wall is larger than the one facing US office CMBS.
Key risks include dependence on a small number of hyperscale tenants, rapid technology obsolescence as GPU generations change, limited power-grid capacity and potential overbuilding in major markets. Loans issued today could mature under significantly different interest-rate, demand and technology conditions.
Risk premiums on data-center CMBS have widened, indicating that investors are demanding greater compensation for risks that remain difficult to model. For crypto traders, the development is an indirect indicator of AI infrastructure financing stress rather than a direct digital-asset catalyst. It may, however, affect sentiment toward AI-related tokens, technology equities and broader risk assets if refinancing pressure increases.
Neutral
The news is neutral for the cryptocurrency market because it concerns commercial real estate credit rather than a specific blockchain, token or crypto regulation. The sharp rise in data-center CMBS issuance and the large 2025–2029 refinancing wall point to growing financial risks, but the article provides no evidence of immediate stress in digital-asset markets.
In the short term, wider data-center CMBS spreads could weaken sentiment toward AI-related investments and encourage traders to reduce exposure to high-beta technology and AI-linked tokens. If refinancing costs rise or a major hyperscale tenant cuts spending, broader risk-off trading could pressure crypto prices alongside technology equities. However, these effects would likely be indirect and depend on evidence of defaults, falling data-center demand or tighter credit conditions.
Over the long term, power constraints, GPU obsolescence and overbuilding could create winners and losers across the AI infrastructure sector. Continued AI investment could support blockchain projects linked to computing, decentralised infrastructure and data services, while a financing downturn could hurt speculative tokens. Similar to past commercial real estate and technology-credit concerns, market impact would likely become significant only if refinancing problems spread into banks, private credit or broader liquidity conditions.