I Analyzed 163K Lines of Kuzu’s Codebase. Here’s Why Apple Wanted It. | by Tanmay Deshpande | Data Science Collective | MediumSitemapOpen in appSign up<br>Sign in

Medium Logo

Get app<br>Write

Search

Sign up<br>Sign in

Data Science Collective

Advice, insights, and ideas from the Medium data science community

Member-only story

I Analyzed 163K Lines of Kuzu’s Codebase. Here’s Why Apple Wanted It.

How a 10-person startup built the graph engine Apple’s on-device AI strategy was missing.

Tanmay Deshpande

7 min read·<br>Feb 14, 2026

Listen

Share

Press enter or click to view image in full size

Source — AppleI spent the last week reading through 163,000 lines of C++20 that make up Kuzu, the embedded graph database Apple quietly acquired in October 2025.<br>The acquisition surfaced through an EU Digital Markets Act filing — 10 employees, Waterloo Ontario, GitHub archived the same day.<br>Every tech outlet ran the same angle: will Apple add a database to iWork? That reading misses what’s actually in this codebase.<br>By the end of this piece, you’ll understand three architectural decisions — * factorized execution, worst-case optimal joins, and an Umbra-style storage engine — that make this acquisition strategic rather than accidental.<br>These aren’t incremental improvements over Neo4j or other graph databases.<br>They’re fundamental design choices that solve the exact problem Apple Intelligence faces: reasoning about relationships across your entire digital life, on-device, without a server.<br>What’s Actually in the Codebase<br>Kuzu is an embedded graph database.

Published in Data Science Collective<br>927K followers<br>·Last published 3 hours ago

Advice, insights, and ideas from the Medium data science community

Written by Tanmay Deshpande

4.7K followers<br>·2.4K following

Book author (7 titles) - Weekly deep dive articles on things that matter

Help

Status

About

Careers

Press

Blog

Store

Privacy

Rules

Terms

Text to speech