{"paper":{"title":"A Techno-Economic Framework for Cost Modeling and Revenue Opportunities in Open and Programmable AI-RAN","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"GPU-based RAN hardware can deliver up to 8x return on investment by leasing idle capacity to AI inference workloads.","cross_cats":[],"primary_cat":"cs.NI","authors_text":"Gabriele Gemmi, Michele Polese, Tommaso Melodia","submitted_at":"2026-03-30T16:59:15Z","abstract_excerpt":"The large-scale deployment of 5G networks has not delivered the expected return on investment for mobile network operators, raising concerns about the economic viability of future 6G rollouts. At the same time, surging demand for Artificial Intelligence (AI) inference and training workloads is straining global compute capacity. AI-RAN architectures, in which Radio Access Network (RAN) platforms accelerated on Graphics Processing Unit (GPU) share idle capacity with AI workloads during off-peak periods, offer a potential path to improved capital efficiency. However, the economic case for such sy"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"across a range of scenarios encompassing token depreciation, varying demand dynamics, and diverse GPU serving densities, the additional capital and operational expenditures of GPU-heavy deployments are offset by AI-on-RAN revenue, yielding a return on investment of up to 8x.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"Publicly available benchmarks of 5G Layer-1 processing on heterogeneous platforms combined with realistic traffic models and AI service demand profiles for LLM inference form an accurate foundation for the joint cost and revenue projections.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Techno-economic framework shows that GPU AI-RAN deployments can offset extra costs via AI revenue for up to 8x ROI across scenarios with varying token depreciation, demand, and GPU densities.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"GPU-based RAN hardware can deliver up to 8x return on investment by leasing idle capacity to AI inference workloads.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"97b1af2d407fa3107263ba3a99e4935c734bdc12708010612d181d5fd1e8db38"},"source":{"id":"2603.28680","kind":"arxiv","version":3},"verdict":{"id":"e5c14c95-5c01-404d-a7cf-4c0c8478b672","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-19T16:50:16.099514Z","strongest_claim":"across a range of scenarios encompassing token depreciation, varying demand dynamics, and diverse GPU serving densities, the additional capital and operational expenditures of GPU-heavy deployments are offset by AI-on-RAN revenue, yielding a return on investment of up to 8x.","one_line_summary":"Techno-economic framework shows that GPU AI-RAN deployments can offset extra costs via AI revenue for up to 8x ROI across scenarios with varying token depreciation, demand, and GPU densities.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"Publicly available benchmarks of 5G Layer-1 processing on heterogeneous platforms combined with realistic traffic models and AI service demand profiles for LLM inference form an accurate foundation for the joint cost and revenue projections.","pith_extraction_headline":"GPU-based RAN hardware can deliver up to 8x return on investment by leasing idle capacity to AI inference workloads."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2603.28680/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":45,"sample":[{"doi":"","year":2023,"title":"Global mobile trends 2023","work_id":"050aed19-4549-43da-ba3a-352764352d6a","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2023,"title":"The economic potential of generative AI: The next productivity frontier","work_id":"5159cd23-6615-4985-9a6d-a3d857e3abed","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"10.1145/3711896.3737413","year":2025,"title":"BurstGPT: A real-world workload dataset to optimize LLM serving systems","work_id":"f8d8d362-841a-46f1-9a17-7cf3545abc39","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2025,"title":"Beyond connectivity: An open architecture for AI-RAN convergence in 6G","work_id":"b4336994-8ecc-45ea-9d29-9f7f38bbf450","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2024,"title":"Industry leaders form AI-RAN alliance","work_id":"ff5cf23a-cb99-4bc8-9b16-b2af64d1e68a","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":45,"snapshot_sha256":"a8c1ad3d1f37cffb7fd8dde7a14d8d50378c4ff6e8aeec675e08e8954e974743","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}