Classified AI compute for the US government now has a commercial path. CoreWeave and Leidos unveiled the plan on July 30: a joint offering that puts CoreWeave’s cloud platform inside accredited classified data centers, with Leidos managing the integration and security work for intelligence and defense customers.
The planned menu runs to five categories. Classified AI cloud services span the full model lifecycle. Analyst augmentation bundles intelligence fusion and imagery exploitation. Cyber AI ranges support simulation and adversarial testing. Synthetic data and digital twins enable mission-relevant generation, and edge-to-cloud orchestration extends a central platform to disconnected forward locations.
Accreditation is the bottleneck, not silicon. Classified workloads require cleared facilities, cleared staff, isolation from the public internet and auditability built in from day one. Commercial multi-tenant clouds are engineered for the opposite conditions, which is why the integrator sits at the center of the deal.
CoreWeave’s half of the deal is the machine room itself: compute, storage, networking and orchestration for training and inference inside Sensitive Compartmented Information Facilities. Leidos’ half is the program delivery, mission integration, accreditation support, cyber operations and data engineering for the intelligence community and the Department of War.
The federal channel also diversifies CoreWeave’s concentrated revenue base. First-quarter revenue hit $2.08B, up 112% year over year, against a $99.4B backlog that includes a $21B Meta commitment. Leidos reported roughly $17.2B in revenue for its latest fiscal year with more than 50,000 employees, and both companies say deployment will track mission requirements and federal appropriation priorities.
