What changes?
GitHub last week announced its new Git storage architecture at GitHub Universe. The company says the design is built to handle heavy use by AI agents and to support agentic development for hundreds of millions of developers.
Test results
In internal testing, GitHub ran 5,000 concurrent agents on a single repository. It sent 7,200 pushes per second to that repo while serving read traffic with consistently low latency. GitHub notes that today's busiest repositories receive at most dozens of writes per second, and it expects that to become a sustained rate as agents write more code.
Two core principles
The architecture rests on two principles: minimize coordination and decouple storage from compute. Git objects are immutable and named by their contents, so two writers can store data at the same time without agreeing on anything. The only part that needs agreement is a tiny reference update.
What it means for developers
GitHub stresses that Git stays Git for developers: push, fetch, and branch work the same way, with the same protections. What changes is that storage, request serving, and repository maintenance can now scale independently.
