Chainguard doubled its output from 500 million to more than 1 billion container build manifests in six months, and the company says that number reflects not a one-off surge but a redesigned system that makes daily, verifiable rebuilds routine.
What a “build manifest” represents
Chainguard counts a “build manifest” as each new, verifiable artifact the Chainguard Factory produces: a fresh image for go:1.26.5, a rebuild of nginx after a libc patch, a new architecture variant, or a regenerated SBOM after a dependency change. A single project such as Python may generate dozens of supported versions, multiple architecture builds, and repeated rebuilds as upstream changes or dependency patches arrive. The metric is intended to show how the catalog stays fresh across every project Chainguard supports, not merely secure at the instant an image is first published.
Chainguard OS and Chainguard Factory: the stack behind the numbers
Everything starts with Chainguard OS, a purpose-built Linux distribution designed for continuous integration and delivery, nano-updates, and rapid rebuilds. Rather than cutting infrequent releases, Chainguard OS uses a rolling release model so new artifacts "ship all day, every day." Artifacts from Chainguard Factory are built from source and include SLSA Level 3 provenance, Sigstore signatures, and full SBOMs. The factory’s declarative, reproducible builds are meant to eliminate hidden state and drift, enabling reliable regeneration of images on demand.
This site is the portfolio.
OSINTSights runs on Cloudflare Workers, D1, R2, and Vectorize, with an AI pipeline on Hetzner ARM. Nubivance designed, built, and operates it. We do the same for clients.
See what we buildFactory 2.0 and DriftlessAF: from event-driven to reconciliation
The original Chainguard Factory automated the basic mechanics of building—resolving dependencies, building, signing, and shipping—but as the catalog grew it became what the company called a "cascading mess." SREs were overwhelmed by event notifications, brittle queues, duplicate failures, and configuration drift that forced human intervention. Chainguard’s response is Factory 2.0, powered by DriftlessAF, an agentic, AI-assisted reconciliation layer atop deterministic automation.
DriftlessAF replaces a purely event-driven model with a continuous reconciliation loop that continuously compares desired state to actual state and works to close gaps when a CVE is reported, a new package version appears upstream, or a new best practice is adopted. A shared, continuous work queue assigns tasks to many reconciler bots; because work items are aimed at a target end state rather than one-off actions, failed items can be dropped or retried and the system converges on the correct outcome over time.
AI’s role: confined, purposeful, and learning
Chainguard positions AI as focused on the unstructured judgment calls that traditional automation could not handle—deciding whether a newly added component in a minor release warrants action, or whether backporting a CVE remediation to an older release is appropriate. Reconciler bots apply AI for those nuanced decisions while continuing to use structured, verifiable tools to prevent “hallucinations.” The system also learns from prior successes—such as successful backports—to inform future patches, making it progressively more capable of tackling complex tasks without a human in the loop for every decision.
What this means for technologists, affected enterprises, and adversaries
- Technologists and security teams: The emphasis is on continuous rebuilds and provenance—SLSA Level 3 attestations, Sigstore signatures, and full SBOMs—so operations teams will be watching whether the reconciliation model reduces manual triage and configuration drift while keeping artifacts verifiably reproducible.
- Affected enterprises and procurement leaders: A rolling-release distro and a factory that ships updates "all day, every day" claims to shorten the window between upstream change and a rebuilt, signed image. Procurement and engineering leaders will need to assess how those continuous outputs fit into their update and validation pipelines.
- Adversaries and threat actors: Chainguard frames the change as a response to attackers who now use AI to accelerate vulnerability discovery and exploit chaining. The company argues defenders must compress their rebuild cycle time at least as quickly, with rebuild speed reducing the hours attackers have to exploit a newly exposed weakness.
The company says Factory 2.0 has already halved its operational friction and doubled output in six months, and that the open source release of DriftlessAF will let other teams build on the same reconciliation model rather than starting from scratch. Chainguard also points readers to its container image catalog and to DriftlessAF documentation for those who want to inspect the artifacts or the system itself.
Matt Moore, Co-founder and CTO of Chainguard, contributed the account of these changes. The central test now is whether a reconciliation-first, agentic approach—open-sourced and aimed at reducing the "CVE doom loop"—will be adopted broadly enough to change defenders' cycle time as attackers continue to leverage AI.
https://thehackernews.com/2026/09/what-it-took-to-reach-1-billion-build.html
