Akuity announced that it is expanding Kargo Enterprise with Fleet Management. Kargo Enterprise, the commercial edition of the open-source promotion project Akuity created, runs on the Akuity Platform. Fleet Management, now in private beta, lets platform teams move changes across a fleet of Kubernetes clusters, GPU nodes and edge devices as one release, with rollouts that are gated, verified and reversible. Enterprises running hundreds of clusters can now ship faster while protecting uptime.
“Managing partial failure is a real challenge when a release spans hundreds of clusters. Teams need to distinguish an isolated cluster problem from a bad change that should stop the rollout. Kargo Enterprise’s Fleet Management brings failure thresholds, release visibility and coordinated rollback into a shared process, reducing the orchestration work platform teams must maintain. As AI accelerates development, that matters. More code should not mean more release plumbing,” said Stephanie Walter, Practice Leader – AI Stack, HyperFRAME Research.
Production infrastructure no longer means one cluster or ten. It means hundreds, and AI workloads keep pushing that number higher. Each new cluster adds release risk and slows down delivery, breaks often and provides little control or visibility which delays it reaching production.
“Companies now run fleets of clusters, and the release process has become what limits how fast they can ship. Fleet Management lets them move faster while keeping each rollout under control. Platform teams that make fleet management a first-class capability now will have infrastructure that scales with the business. Those that wait will feel the gap on every release,” said Hong Wang, co-founder and CEO of Akuity.
With Fleet Management in Kargo Enterprise, platform teams group clusters into a fleet, set the rollout order and failure threshold once and Kargo promotes each change through it wave by wave. Each cluster reports its own health so the platform team sees where a release stands across the fleet throughout the rollout.
Fleet Management adds the controls existing delivery tools lack:
- Failure thresholds, not binary outcomes. A rollout stops only when the failure rate crosses a set threshold, so one failed cluster does not halt the fleet.
- Ordered rollout across waves. Changes move from pilot to canary to the general population, sequenced by risk, with a health check between each wave.
- The fleet as a unit. Platform teams can stop, roll back and investigate across the whole fleet at once, with rollback as deliberate and observable as rolling forward.
- Aggregate observability. One view across the fleet replaces individual dashboards for each cluster.
- Per-team RBAC. Teams that own a subset of the fleet get scoped control over it.
With Fleet Management, customers can:
- Contain a bad patch before it spreads. When a platform team rolls out a CVE fix for an add-on such as cert-manager, the rollout moves in waves with a health check on each so a bad update stops in its first wave instead of reaching the whole cluster estate.
- Clear the release backlog. With waves, health checks and failure thresholds built into the rollout, finished changes stop waiting on release windows and cluster-by-cluster sign-offs, and engineers spend less time maintaining separate pipelines.
Protect revenue as the business scales. As the fleet grows into new regions, markets and GPU capacity, a bad release stays contained, so a routine deploy does not turn into an SLA breach or a churn risk.
