HP management says enterprises shift to hybrid AI as cloud costs rise
HPQ•HP management said rising AI cloud costs are driving enterprises toward hybrid architectures that move more inference to PCs, workstations and edge devices. It cited projections that more than 20% of enterprises will run AI locally by 2028 and inference will exceed half of AI workloads by 2030.
1. Inference moves on-device
HP management flagged rising AI cloud costs as a driver of hybrid architectures that shift more inference to PCs, workstations and edge devices. It said local inference can improve token economics, lower latency and reduce security exposure for agentic workflows handling sensitive data.
2. Cloud remains in mix
Management cited growing use of lower-cost open-weight models and an estimate that 95% of enterprise AI still uses frontier models. It expects cloud inference to remain for the largest, compute-intensive workloads, with focus shifting to orchestration across devices, edge and cloud.
3. Adoption projections
Management cited Gartner projecting that more than 20% of enterprises will run AI locally by 2028, and McKinsey projecting that inference will exceed half of AI workloads by 2030.
