Among Intel's ongoing reduction in open-source projects they maintain, their BigDL open-source project focused on running large language models across Intel XPUs from Core Ultra laptops to discrete GPUs to cloud / data center hardware all in a low-latency manner, is being ended.

Over the past year of Intel's shift of open-source strategy and corporate restructuring and cost cutting, they have ended many different open-source projects they formally maintained. Many of them not too actively maintained in recent months and the like, but as the latest surprising twist, they are preparing to sunset BigDL. Considering the ongoing market demand around AI and LLMs, it's rather surprising they are ending BigDL especially with it being an open-source project still routinely seeing new commits.

BugDL saw a lot of work in recent years with a focus to "seamlessly scales your data analytics & AI applications from laptop to cloud" with integration around TensorFlow, Keras, PyTorch, Apache Spark/Flink, and more. BigDL focused on both CPU and GPU acceleration, securing big data and AI with Intel SGX and TDX technology, and more.

This week the project was archived with the usual update noting the project is no longer being maintained by Intel. On Friday that archival notice was then updated to say it's going to be archived on 30 June 2026. So rather than right away, you now have until the end of the month to move off Intel BigDL.

Part of BigDL, it's BigDL-LLM component, was spun out previously to IPEX-LLM. But IPEX-LLM itself is no longer maintained and was already archived back in January.

Those wanting to check out Intel BigDL before it goes EOL can find the code via intel/BigDL on GitHub.