Microsoft integrates DeepSeek AI models into Copilot+ PCs and Azure amid competition concerns

Microsoft, which supports OpenAI, recently announced in its blog that it will integrate the DeepSeek-R1 model into the Windows Copilot+ PCs. It will first be available on PCs with Qualcomm Snapdragon X, followed by computers with Intel Core Ultra 200V and other devices.

The initial release, DeepSeek-R1-Distill-Qwen-1.5B, will be offered in the AI Toolkit, with 7B and 14B variants coming soon. These optimised models are designed to help developers build and deploy efficient AI applications on Copilot+ PCs, leveraging the powerful Neural Processing Units (NPUs).

In addition, Microsoft is adding the R1 model to its Azure AI Foundry platform, placing it alongside AI models from OpenAI, Mistral AI, and Meta.


The company emphasised in its blog that DeepSeek has applied lessons learned from Phi Silica — a small language model for Windows 11 — to the DeepSeek models, focusing on minimising battery consumption and optimising PC resources.

The blog noted, “The optimised DeepSeek models for the NPU take advantage of several of the key learnings and techniques from that effort, including how we separate out the various parts of the model to drive the best tradeoffs between performance and efficiency, low bit rate quantisation and mapping transformers to the NPU.”

This move comes amid ongoing allegations from Microsoft and OpenAI that DeepSeek used ChatGPT's proprietary models without permission, with both companies reportedly investigating the matter.

Microsoft reported results for the quarter ending December 31, 2024, with shares dropping 4.5% in after-hours trading due to concerns over high spending and competition from cheaper Chinese AI models like DeepSeek.

Also Read: Microsoft, OpenAI allege DeepSeek used proprietary data to power its breakout models