We underwrite a 2–5x outcome driven by (1)
backlog monetization as powered capacity comes online through 2026–2028, and (2) attach of higher-value layers (production-readiness, security/verification, storage/data governance) that reduce pure $/GPU-hour commoditization. Versus direct
neocloud comps (e.g., NBIS/APLD/IREN), CoreWeave is further along on contracting and scale, but still carries “build-cycle” risk that caps the exit
multiple.