Production Experiences with uv

Yesterday on Reddit, batman-iphone asked:

Anyone used uv package manager in production

Is it reliable to use it in production as it is comparatively new in the market.

Also, are there any disadvantages that I should be aware of before pitching it to my manager?

Given that the tool was introduced just over a year ago, its rate of adoption has been wild.

Here is my summary of the discussion:

uv Pros

Real Performance Gains

Multiple developers reported dramatic performance improvements, e.g.,

“With a warm cache you’re likely to see closer to the stated 10-100x [speed improvement].” (source)

Support for Workspaces

One of the most appreciated features is uv’s support for workspace management:

“uv has a concept called workspaces which are analogous to workspaces in Rust’s package manager cargo. In effect, you can have multiple Python packages in one repo, and have dependencies between these packages.” (source)

This capability allows teams to maintain compatibility between interdependent packages while keeping them individually installable, solving significant challenges in managing code across repositories.

I have not paid attention to this feature of uv, but I’m going to start experimenting with it soon.

All-in-One Tooling

Many developers praised uv’s comprehensive approach:

“I love that it’s sort of like an all-in-one tool. Makes working with multiple Python versions a breeze.” (source)

uv consolidates functionality that previously required multiple tools (pip, pip-tools, pipx, poetry, pyenv), creating a more streamlined development experience.

uv Concerns

Configuration Learning Curve

Some developers noted initial challenges:

“It takes time to get hang of the flags if you use the lock file. Like don’t install the current project as editable, use virtual env in a different location and so on.” (source

The command structure and various options require some adjustment for teams accustomed to other tools.

Dependabot Support

“One disadvantage is that dependabot support isn’t here yet, so dependabot won’t yet scan your lock files.” (source)

Several users mentioned that Renovate already supports uv, and Dependabot support is currently under development.

Industry Adoption

The discussion showed surprisingly broad adoption across various sectors:

“Been using it in production for over 6 months. It’s pretty much a standard now in Python industry it seems.” (source)

Even teams in regulated industries reported successful migrations:

“Work in a pretty regulated industry, we’ve almost all switched over. Mostly because we all grown to hate poetry.” (source)

Conclusion

For teams considering uv, the consensus suggests it’s stable enough for production use despite its pre-1.0 status. Migration appears straightforward, especially from Poetry, with most users reporting a “straight upgrade” experience.

The performance benefits alone make it compelling for CI pipelines and Docker builds, while workspace support offers significant advantages for complex multi-package projects. Most challenges appear to be minor learning curve issues rather than fundamental limitations.

As one user summarized it: “It’d be like turning down a free Ferrari because you don’t like the air freshener.” (source)