CoreWeaveRegister for a premium account to gain access to Sterling AI.
Get StartedThings you can ask Sterling:
| Region | Sites + Detail | Strategic Context |
|---|---|---|
| North America - US Central | Plano TX (flagship DC) + Dallas TX + Las Vegas NV + Chicago IL + Wilmington DE | Largest concentration of CoreWeave capacity; ~50%+ of total GPU count |
| North America - US Northeast/Southeast | Atlanta GA + Reston VA + Boston MA + smaller satellites | Growing footprint to support enterprise + hyperscaler co-located customers |
| North America - Canada | Toronto ON + Montreal QC | Cross-border expansion; renewable-energy positioning |
| Europe - UK + Western Europe | London UK + Frankfurt DE + Stockholm SE | European AI sovereignty + GDPR-compliant deployment for EU customers |
| Total facilities | ~30+ sites globally | Scaling toward 50+ by mid-2026 + 100+ by 2027-2028 per company commentary |
| Site lease vs own mix | ~90% leased / ~10% owned | Capital-efficient scaling; trades off long-term cost vs short-term capital deployment |
| Editorial. Why this matters | Strategic context | Data center footprint is the physical substrate of CoreWeave's GPU-cloud business + the constraint on capacity growth. Compared to peers: AWS + Azure + GCP own most of their data centers; pure-play neoclouds (CoreWeave + Crusoe + Lambda) primarily lease. The marquee question is whether scale economics force CoreWeave toward an owned-DC model over time, or whether the leased model remains structurally competitive |