In 2018, Google open-sourced gVisor under the Apache 2.0 license. To the best of its contributors’ knowledge, it has ever since remained the second most mature implementation of Linux, after Linux.
\nThis year, the gVisor project is being donated to CNCF, and its governance\nmodel is shifting accordingly.
\nGoogle is donating the gVisor project, including its name and trademarks, to the Cloud Native Computing Foundation (CNCF), a subsidiary of the Linux Foundation. Like its name implies, CNCF is focused on cloud-native computing, with Google having seeded its creation by donating the Kubernetes project. Since then, Kubernetes has grown to become the industry-standard for container orchestration, and has grown a large and vibrant ecosystem around it. gVisor is now following the same footsteps.
\nYou can see gVisor’s CNCF application and process.
\nWhat has already happened:
\nOver the next few weeks:
\nThe project will move to CNCF “Sandbox ” status (quite
appropriately-named for a project like gVisor).
Over the next few months:
\nThe project will take the steps needed to move to CNCF “Incubation ”
status.
google GitHub organization.The gVisor project’s governance model will transition to a long-term model
that features org-based voting , thereby preventing Google from having\nunilateral control over governance decisions.
gVisor doesn’t fit neatly into the industry’s well-known boxes of the\nsandboxing/security landscape, which tends to separate “vanilla containers” from\n“virtual machines” with shades of gray in between. gVisor straddles this\nmiddle-ground, providing\nempirically-equivalent security but without\nchecking the familiar “virtualization” checkbox that security auditors,\nregulators, or security practitioners often treat as a one-to-one proxy for\n“secure”. This has caused adoption challenges over gVisor’s history, as it\nhas been difficult to communicate the value of the project to an audience that\nis used to this false dichotomy.
\nAnother challenge gVisor has faced is that of a performance perception\nproblem. Internally within Google (and other gVisor-using companies, such as\nAnt Group and Modal), there exist Linux kernel patches that improve gVisor\nperformance significantly. However, for other gVisor users, out-of-the-box\nperformance often shows performance degradation for certain I/O-intensive\nworkloads. This has led to poor first-impressions from potential adopters.\nWe have tried to address this by upstreaming Linux kernel patches that improve\nits performance, but have been turned down by kernel maintainers due to gVisor\nbeing a wholly-owned Google project.
\nLastly, gVisor as a project has potential that is difficult to prioritize when guided by corporate ownership alone. As a userspace implementation of Linux, gVisor has potential non-commercial applications such as:
\ngVisor-on-Mac : AllowingLinux programs to run on macOS , with a
similar experience as to how how Wine allows Windows programs on macOS.
Desktop Linux sandboxing : Allowing gVisor to be used as a**practical
option for desktop Linux application sandboxing** that is much more secure\nthan the current state of the art (bubblewrap/flatpak/nsjail/etc), yet much\neasier to integrate with than full-blown virtualization-based approaches\nlike that of Qubes OS. We have made some\nadvancements on this front with our\nrecently-introducedbwrap drop-in replacement,\nbut gVisor is capable of sandboxing\nso much more.
We see evidence of these problems by looking at the\ncurrent set of gVisor adopters, which are all either large tech\ncompanies with the ability to invest and customize gVisor to suit their own\nneeds (Google, Ant Group, OpenAI, Anthropic), and startups with a\nhighly-specific focus that exactly fits gVisor’s use-case, and where it makes\nsense to spend a startup’s limited resources specifically into making gVisor\nwork great for them (Modal, Tines). Who is not on this list?
\nHobbyist projects : See aforementioned non-commercial applications where
gVisor would be useful but isn’t currently adopted.
The “middle” of the industry : Individuals and companies that would
benefit from gVisor’s security, but either aren’t aware of its existence,\ndismiss it out of past perception problems, or don’t have the resources to\ninvest specifically into security but would happily adopt an off-the-shelf,\nwidely-available sandboxing runtime were it to exist as a widespread and\ncheap option already available as an offering by their computing\ninfrastructure provider.
Other non-Google hyperscalers : While gVisor is adopted internally by
nearly all large tech companies for their own at-scale sandboxing needs\n(e.g. code snippet execution, RL), only a small subset (Google,\nDigitalOcean, and Modal) directly sell general-purpose gVisor-powered\ncompute to their customers. This is in spite of gVisor’s competitive\noperational margins, as well as thedemonstrable demand for this,\nbecause issue reports to the gVisor repository show that a large number of\nentities are self-installing gVisor on non-Google hyperscalers. The\nremaining explanation of the lack of direct integration is likely the\nproject’s (pre-donation) governance risk.
By contributing the project to the CNCF, we aim to address all of these issues. This enables gVisor and application kernels to become part of the container ecosystem and security industry’s lingua franca, enabling integration and adoption beyond highly-motivated/sophisticated/resourceful corporate entities, and enables upstreaming Linux patches that solve gVisor’s performance for everyone.
\ngVisor is a drop-in-compatible container runtime that fits in the Cloud Native\ncontainer ecosystem. It integrates directly with CNCF technologies such as\nKubernetes, containerd, and\nAgent Substrate.
From a resource and efficiency standpoint, gVisor acts as a more cloud-native\ncontainer runtime than other security-focused container runtimes, thanks to its\ncontainer-like process model. This allows it to be efficiently and\ntightly-sized, enabling secure container binpacking at a resolution and density\nVM-based runtimes cannot match. It also does not require hardware virtualization\nor nested virtualization, enabling it to run anywhere Linux runs. That makes\ngVisor a good complement to CNCF’s existing portfolio. gVisor and is already\nusable on all major clouds, some of which offer it as a native offering, and\nothers for which cloud users can (and do) self-install it.
\nThe past few months have been pivotal for the security industry and\nfor secure sandboxing specifically.\nIt would make very little sense for Google to abandon its sandboxing technology\nat a time when the need for cheap and secure sandboxing has never been clearer.\nOn the contrary, we (the gVisor contributors at Google) have been pushing for\nthis move internally with the expectation that this will accelerate gVisor’s\ngrowth and adoption both within and outside of Google, in a similar manner as\nwhat has happened with Kubernetes.
\nGoogle is reaching out to potentially-interested parties. As of this writing, the following entities have committed to joining gVisor’s maintainers for the long haul: Ant Group, Modal, and Tines. Additionally, other companies including OpenAI, Tencent, and NVIDIA will continuing their ongoing contributions to gVisor.
\nSome gVisor contributors will be at KubeCon North America 2026. Come chat!
","headings":[{"level":1,"text":"gVisor is being donated to CNCF","id":"gvisor-is-being-donated-to-cncf"},{"level":2,"text":"What’s happening?","id":"what-s-happening"},{"level":2,"text":"What’s the timeline?","id":"what-s-the-timeline"},{"level":2,"text":"Why donate?","id":"why-donate"},{"level":3,"text":"Why donate to CNCF specifically?","id":"why-donate-to-cncf-specifically"},{"level":2,"text":"Will Google divest from gVisor development in the future?","id":"will-google-divest-from-gvisor-development-in-the-future"},{"level":2,"text":"Who is joining gVisor’s contributors beyond Google?","id":"who-is-joining-gvisor-s-contributors-beyond-google"},{"level":2,"text":"What does this mean for me?","id":"what-does-this-mean-for-me"},{"level":2,"text":"What’s next?","id":"what-s-next"}]}}