Specifications
Vendor
Qualcomm
Product
Dragonwing AI On-Prem Appliance
AI Compute
Up to 870 TOPS
Power
~150 W system
Silicon
Qualcomm Cloud AI 100 Ultra accelerators
Model Support
Large vision and language models up to 120 billion parameters
Workloads
Multi-user inference, RAG, video analytics, forensic search, natural-language interaction
Availability
Through partners including Aetina, Advantech and Lanner
Deployment
On-premises / edge appliance
Overview
The Qualcomm Dragonwing AI On-Prem Appliance delivers up to 870 TOPS of AI compute in a compact system drawing about 150 W, built on Qualcomm Cloud AI 100 Ultra accelerators. It is designed to run large vision and language models — up to 120 billion parameters — on premises rather than in the cloud.
Target workloads include multi-user inference, retrieval-augmented generation (RAG), video analytics and forensic search, and natural-language interaction. The appliance is offered through partners including Aetina, Advantech and Lanner.
Key Benefits
Up to 870 TOPS in ~150 W - Cloud AI 100 Ultra accelerators - models up to 120B parameters - multi-user inference and RAG - on-prem deployment.
Applications
On-premises generative-AI inference, video analytics and forensic search, multi-user RAG serving and edge data-center AI.
Request a Quote — Qualcomm Dragonwing AI On-Prem Appliance
QS Compute — global B2B supply of AI computing hardware, edge AI systems and accelerators. Volume pricing, 15-day sample lead time.
Get Your Quote →