Perplexity AI unveils hybrid local-cloud inference system

Perplexity AI has introduced a novel hybrid inference system that dynamically routes AI workloads between local devices and cloud-based models. This innovation, demonstrated at Computex 2026, allows sensitive data to remain on user devices while leveraging powerful cloud models for complex tasks, balancing accuracy, privacy, and cost. The system autonomously decides where each part of a task should execute, a capability not previously seen in AI orchestration. This development could significantly impact enterprise adoption, particularly in regulated industries, and potentially alter the demand for extensive data center infrastructure.