• AI tenants pay up to 30% more per megawatt than crypto miners (2025).
• U.S. data center occupancy rates near 95% in key hubs, up 8% YoY.
• Crypto miners offered short-term leases as power is reprioritized.
• Power constraints drive risk reassessment in new lease underwriting.
• Grid expansion lags, limiting capacity for both AI and crypto sectors.

A sharp rise in artificial intelligence (AI) workloads is transforming the U.S. data center landscape. With occupancy rates nearing record highs, operators are actively negotiating short-term leases with cryptocurrency miners to optimize surplus capacity. This marks a pivotal shift as firms prioritize AI tenants—who pay a premium for power—over less lucrative crypto clients. These power-driven tenant dynamics are shaping underwriting, lease terms, and long-term capital allocation.

AI’s appetite for computing power has outpaced projections, with average rates for AI tenants reaching 30% above those paid by cryptocurrency miners, per WSJ 2025 data. By contrast, crypto mining—once a dominant source of variable revenue—now faces shorter, more flexible lease terms as operators seek to maximize margins. “It’s a game of megawatts, not square feet,” notes one data center executive. The result is a heightened focus on power allocation, not just floor space.

In major U.S. data center markets, occupancy rates have climbed to nearly 95%, an 8% increase year-over-year, according to sector surveys. Meanwhile, grid expansion initiatives lag behind demand, particularly in Northern Virginia and Texas. This grid bottleneck forces operators to triage tenants, often favoring AI contracts with longer terms and higher power density. Consequently, crypto miners find themselves facing rising rental costs or relocation pressures. On balance, competition for limited energy supply is driving more aggressive lease negotiations.

The shift in tenant composition is altering risk profiles across the sector. Lenders and investors are scrutinizing lease durations and the creditworthiness of tenants tied to volatile industries like crypto. As AI tenants secure longer-term, higher-value contracts, stabilized cash flows become more attractive for institutional capital. By contrast, facilities with a high share of short-term crypto leases may see increased credit risk premiums. Ultimately, underwriting is becoming more sensitive to power reliability and tenant stability.

If grid investments accelerate or power pricing stabilizes, data center operators could regain flexibility in tenant selection. However, as of Q4 2025, the persistent power shortage means capital is likely to flow toward projects with high-efficiency cooling, backup generation, and AI-oriented infrastructure. Regulatory policy—particularly on energy allocation—remains a wild card for underwriting. Should states impose stricter prioritization, lease pricing and risk spreads could widen further. For now, disciplined capital allocation favors long-term, power-intensive tenants over transient crypto demand.