# Installation The Qualcomm® Cloud AI SDK includes two components, Platform SDK and Apps SDK. Choose the deployment mode below to see which components apply to your environment. ## Deployment modes Note If you are using the Dragonwing AI On-Prem Appliance, the hardware and software are preconfigured. The instructions in this section are provided for reference only. Choose the mode that matches your setup: - Cloud instance - Both the Platform SDK and Apps SDK are preinstalled on supported cloud instances (AWS, Cirrascale). No SDK installation is required. - Single-host - The Platform SDK and Apps SDK are both installed on the same machine. Compilation and inference run on the same system. This is the standard path for local server and on-premises appliance installations. - Offline compilation (separate build host) - The Apps SDK is installed on a dedicated build host **without** Cloud AI hardware for model compilation only. Compilation doesn’t require Cloud AI hardware. Compiled QPC artifacts are then transferred to a hardware-attached host running the Platform SDK for inference. Use this mode when your build environment is separate from your inference environment. SDK requirements by deployment mode | Host Role | Platform SDK | Apps SDK | | --- | --- | --- | | Cloud instance (preprovisioned) | ✔ Preinstalled | ✔ Preinstalled | | Local server (single-host) | ✔ Required | ✔ Required | | Offline build host (no hardware) | ✘ Not required | ✔ Required | | Inference host (hardware only) | ✔ Required | ✘ Not required | Note Systems running Linux kernel 6.10 or later include KMD support and firmware for Qualcomm Cloud AI accelerators. If you are running kernel 6.10+ and deploying with containers, the Platform SDK installation is optional because the Cloud AI container images bundle the necessary runtime libraries and tools. ## Choose your installation path Select the environment that matches your setup: Tab Cloud instance Tab Local server Tab Offline build host Tab Docker SDKs are preinstalled on supported cloud instances (AWS, Cirrascale). Verify your installation before proceeding. Install both the Platform SDK and Apps SDK on the same host. Follow the steps in order: 1. [Review prerequisites](https://docs.qualcomm.com/doc/80-99100-3/topic/prerequisites.html). 2. [Download SDKs](https://docs.qualcomm.com/doc/80-99100-3/topic/download-sdks.html). 3. [Install Platform SDK](https://docs.qualcomm.com/doc/80-99100-3/topic/platform-sdk.html). 4. [Install Apps SDK](https://docs.qualcomm.com/doc/80-99100-3/topic/apps-sdk.html). 5. [Verify Cloud AI hardware and software functionality](https://docs.qualcomm.com/doc/80-99100-3/topic/verification.html). Install the Apps SDK on a host without Cloud AI hardware for model compilation then transfer the compiled QPC to an inference host with Cloud AI hardware for execution. 1. [Review prerequisites](https://docs.qualcomm.com/doc/80-99100-3/topic/prerequisites.html). 2. [Download SDKs](https://docs.qualcomm.com/doc/80-99100-3/topic/download-sdks.html). 3. [Install Apps SDK](https://docs.qualcomm.com/doc/80-99100-3/topic/apps-sdk.html). 4. [Verify Cloud AI hardware and software functionality](https://docs.qualcomm.com/doc/80-99100-3/topic/verification.html). Containerized deployments use prebuilt Cloud AI Inference containers or custom images built with the Apps SDK. Use the prebuilt images unless you need to customize the image for your workflow; prebuilt images reduce setup time and include validated dependencies. 1. [Install Platform SDK on the host](https://docs.qualcomm.com/doc/80-99100-3/topic/platform-sdk.html). 2. [Install Docker Engine](https://docs.docker.com/engine/install/) on the host system that will run the container. 3. Choose a [prebuilt Cloud AI image](https://docs.qualcomm.com/doc/80-99100-3/topic/index_Docker.html#reference-to-docker), or use the Apps SDK on the host to [build a custom Docker image](https://docs.qualcomm.com/doc/80-99100-3/topic/build-image.html). 4. [Launch the container](https://docs.qualcomm.com/doc/80-99100-3/topic/launch-container.html#reference-to-docker-launch-container) with device access. 5. [Verify Cloud AI hardware and software functionality](https://docs.qualcomm.com/doc/80-99100-3/topic/verification.html). Note Install the Docker Engine on the host system, not inside the container. ## Next steps Complete the [Prerequisites](https://docs.qualcomm.com/doc/80-99100-3/topic/prerequisites.html#reference-to-pre-requisites) before installing the Cloud AI SDKs. Last Published: Aug 25, 2026 [Previous Topic Verify setup](https://docs.qualcomm.com/bundle/publicresource/80-99100-3/topics/verify-setup.md) [Next Topic Prerequisites](https://docs.qualcomm.com/bundle/publicresource/80-99100-3/topics/prerequisites.md)