Skip to content

Back to the catalog

GitHub saw 7.4B commits in Sep., adding 4M CPU cores+160 PB storage to keep up
thenewstack.ioarticle

GitHub saw 7.4B commits in Sep., adding 4M CPU cores+160 PB storage to keep up

GitHub reported 7.4 billion commits per month in September 2026, up from 2.9 billion in August and 1.4 billion in April, which it attributes mostly to agentic demand. It has added 4 million CPU cores, 160 PB of storage and 50 Tb/s of network capacity this year, mostly on Azure. It is also rearchitecting to separate compute and storage, with internal benchmarks showing up to 35x higher write throughput.

TL;DR GitHub says AI agents pushed it to 7.4B monthly commits, so it added 4M CPU cores and 160 PB of storage and is splitting compute from storage.
Document

What it claims

GitHub announced at its Universe conference that it now handles 7.4 billion commits per month. In 2024 the figure was 65 million monthly commits, so the current rate is roughly 100 times higher. GitHub attributes the growth mostly to "agentic demand" and links it to its recurring outages.

The capacity added this year is 4 million CPU cores, 160 PB of high-speed storage, 50 Tb/s of network capacity and more compute for GitHub Actions, mostly on Azure. In August, CTO Vlad Fedorov had reported 3 million added cores, so about 1 million more came online between August and the Universe announcement.

How it works

The page gives two responses to the load: buying capacity and changing the architecture. The new architecture separates compute and storage. Jay Parikh, Microsoft's EVP for Core AI, writes that "capacity alone won't get us there."

The New Stack's author says GitHub was scaling on a legacy foundation. The author also says moving to Azure was "essentially the only option" because GitHub's own data center had run out of capacity.

Evidence

The only performance figure is "up to 35x higher write throughput" from GitHub's internal benchmarks of the new architecture. The page gives no workload, baseline system or test setting for it. GitHub first described this result earlier in October 2026, and the Universe announcement repeated it.

Limits

  • Architecture status: The page does not say whether the compute and storage split is in production, partly rolled out or only planned. It gives no timeline.
  • Outages: The article says outages have become "something of a recurring theme." It gives no incident counts, uptime figures or affected services.
  • Capacity figures: The core, storage and network numbers are totals for the year. They are not broken down by service, apart from a mention of Actions.
  • Sources: Every number comes from GitHub or Microsoft announcements and is not independently verified. The "legacy foundation" and "only option" judgements are the article author's, not GitHub's.

For teams that depend on GitHub or Actions, the article's main point for adoption is this: GitHub links its reliability problems to agent-driven load, and the architectural fix has no stated delivery date.

Organisations named here

  • GitHub
  • Microsoft
  • GitHub Universe
  • Jay Parikh
  • Vlad Fedorov

Technologies named here

  • GitHub
  • GitHub Actions
  • Azure

Techniques named here

  • Separating compute and storage