Flox unveils 'Kubernetes, Uncontained' to slash container image overhead. Image Credit: kerenby/Bigstockphoto.com Flox announced "Kubernetes, Uncontained" at KubeCon NA 2025, a solution that eliminates the need for massive container images that slow AI/ML deployments and drive up cloud costs. Teams deploying PyTorch, TensorFlow, or CUDA workloads are stuck managing 5-10GB+ container images, creating friction in CI/CD pipelines, registry storage, and security reviews. Flox's approach lets teams deploy immutable, hash-pinned environments directly to Kubernetes with no image rebuilds required. Pods pull only the packages they need from node-local caches, delivering 50% faster deployments with zero base-image overhead and the same reproducibility guarantees. Integrates at Kubernetes CRI layer (like Kata Containers or gVisor) and provides SBOMs-by-default with cryptographic provenance for every deployment. It works with existing GitOps workflows, CI/CD pipelines, and security tools. Kubernetes delivers a rock-solid foundation for modern infrastructure. Flox adds a small, vendor-maintained shim that allows teams to declare their workload dependencies as Flox environments, powered by Nix. These are declarative, immutable stacks where every package and version is pinned to SHA256-hashed artifacts. At pod startup, the shim activates the Flox environment and fetches the specified artifacts from the node's local read-only store. Flox's approach integrates cleanly with existing deployment workflows, including CI/CD pipelines, GitOps practices, and security tools. The only change is what gets deployed - Flox environments instead of container images - while Kubernetes's orchestration, networking, and runtime behavior remain unchanged. Benefits include a smaller attack surface, lower cloud costs, faster releases, and Software Bills of Materials (SBOMs) by default. * Building on Proven Patterns at Scale: Flox draws on Kubernetes's proven extensibility to make "imageless" deployments a viable pattern for mainstream infrastructure. Instead of bundling all dependencies into multi-gigabyte container images, Flox uses Nix's input-addressed storage to load only required dependencies at runtime. * Key Advantages: Organizations can use existing CI/CD systems, GitOps tools, and security scanning processes. Kubernetes manifests require minimal changes - in most cases just the image reference. Existing Kubernetes security controls, networking policies, and admission controllers work unchanged. The Flox shim integrates at the CRI layer, so it's a drop-in option. * Other advantages include reduced attack surface, simplified compliance, reproducible behavior at build-time and runtime, lower cloud costs, fast deployments, and portability across architectures and platforms. * Real-World Impact: Teams running AI/ML workloads and complex multi-language stacks face major packaging challenges. Container images can reach 5-10 GB or more. With Kubernetes "Uncontained," there's zero base-image overhead, enabling pre-warmed caches and hash-pinned dependencies, making deployments reproducible and auditable. * Market Momentum: Since launching Flox 1.0 in March 2024, the company has delivered 40+ releases and attracted Fortune 5 enterprises alongside high-growth tech companies. The recent $25M Series B led by Addition underscores their long-term commitment. * Availability: Flox Kubernetes support is available today for all Flox users. Michael Brantley, CTO and co-founder of Flox Kubernetes gives teams resilient, reliable infrastructure for scheduling, networking, and rollout mechanics. What The Fast Mode is doing is simplifying how workloads get packaged and deployed. Teams define applications as immutable, versioned Flox environments, then deploy them using a minimal Flox-maintained shim. The same environment that runs on a developer's laptop runs identically in production. Kelsey Hightower, distinguished engineer and Flox advisor Just like Kata Containers used the CRI to give you VM-level isolation, or gVisor to give you user-space kernel isolation, Flox's shim gives you Nix-style reproducibility with the ability to run only the software you actually need

Flox secured $25M in Series B funding led by Addition, bringing total funding to over $50M. The company aims to solve software inconsistency issues with its platform built on Nix technology, offering deterministic environments across devices. Flox plans to enhance cross-OS consistency, software governance, and security infrastructure. The platform supports AI applications by providing reproducible environments, addressing infrastructure challenges in modern software development.