Sebastiao Pereira implements k-means clustering in T-SQL:
K-means clustering is an unsupervised machine learning algorithm used to group data into k distinct clusters based on their similarity, allowing for customer segmentation, anomaly detection, trend analysis, etc. The most common machine learning tutorials focus on Python or R. Normally, data is stored in SQL Server, and it is necessary to move data out of the database to apply clustering algorithms and then, if necessary, to update the original data with the cluster numbers. Is it possible to do it directly in SQL Server?
Given the work you have to do to implement this, I can’t imagine that it would be particularly fast. But it is neat to see that it’s possible.