# skywater-pdk: an open process kit for the SKY130 node > Google and SkyWater Technology Foundry published the design rules, cell libraries and EDA support files for a real 130nm process. The repository is still labelled an experimental preview, and the README says plainly what that means for tapeout. **google/skywater-pdk** — Open source process design kit for usage with SkyWater Technology Foundry's 130nm node. - Repository: https://github.com/google/skywater-pdk - Website: https://skywater-pdk.rtfd.io - Stars: 3,726 · Forks: 488 - Language: Python - License: Apache-2.0 - Published: 2026-10-06 · Updated: 2026-10-06 · Language: en - Canonical page: https://hysenlabs.com/projects/google-skywater-pdk ## What a process design kit actually is A process design kit is the file-level contract between a designer and a fab. It tells you the design rules a process imposes, supplies the models that let a simulator predict what your circuit will do after it is manufactured, and provides the standard cell libraries that a synthesis tool draws from. Without it you can draw a schematic, but you cannot honestly predict the silicon. The SkyWater Open Source PDK is a collaboration between Google and SkyWater Technology Foundry intended to produce that contract for the SKY130 process node in full. The README describes the node as a mature 180nm to 130nm hybrid technology, originally developed inside Cypress Semiconductor, spun out into SkyWater, and then made accessible to general industry through this collaboration. As of May 2020 the repository targets SKY130, and the README notes that if that release succeeds, more advanced nodes may become available later. What lands in the repository is listed explicitly: documentation covering the design rules required to create manufacturable devices on SKY130, EDA tooling support files for both open source and proprietary design flows, primitive cell libraries and models for analog work, several standard digital cell libraries aimed at different use cases, and documented examples of using the kit. Documentation, libraries and tool support in one place is the shape of a real PDK, not a subset of one. ## What the SKY130 process gives a designer The process stack is spelled out in the README as a bulleted list, and it is worth reading closely because it explains which designs are a natural fit. The node supports internal 1.8V with 5.0V I/O that can be operated at 2.5V, has one level of local interconnect and five levels of metal, is inductor capable, includes a high sheet resistance poly resistor, offers optional metal-insulator-metal capacitors, includes a shrunken SONOS cell, supports a 10V regulated supply, and provides extended-drain high voltage NMOS and PMOS devices. Several of those are normally optional paid features at other foundries. The local interconnect, SONOS functionality and MiM capacitors are called out in the README as standard rather than add-ons, and the framing is that this flexibility widens the range of designs worth attempting. If you need something beyond that set, the README points to SkyWater directly, which specializes in process customization including materials such as niobium, germanium, vanadium oxide and carbon nanotubes. The high voltage devices and the 10V supply capability are the reason 130nm keeps coming up. The node was first commercialized around the 2001 to 2002 timeframe, and the README says it is now used mainly in research, small microcontroller development and mixed signal embedded designs such as IoT devices. That is an accurate picture: this is the process for a chip that does a little analog work around a small digital core, not for a large SoC. ## Why the experimental preview label is load bearing The README puts a warning near the top and the distinction it draws is the single most important thing to understand before using this kit. Google and SkyWater are treating the current content as an experimental preview, described as an alpha release. The underlying SKY130 process and the PDK the open release was derived from have produced many designs manufactured commercially in significant quantities, but the open source PDK is explicitly not intended for production settings at this time. What it should be usable for is spelled out next: test chips and initial design verification, and even that carries no guarantee. The explanation offered is that internal validation and test designs, including silicon validation, are under way by Google, SkyWater and their partners, and that the results are planned for publication. The versioning scheme gives you a concrete thing to watch. The PDK gets tagged with a production version when it is ready for production design, and the README points to a versioning section describing how the numbers are assigned. There is also a badge on the repository that counts commits since the latest release, and the release tag it references is v0.0.0, which is an honest signal of where the project sits. For staying informed, the README directs readers to a skywater-pdk-announce mailing list. ## Getting the environment set up with conda and make The repository is built around a conda environment and a small set of Python tooling. The `Makefile` at the root names the two inputs it needs: `requirements.txt` for pip and `environment.yml` for conda. It pulls in a `third_party/make-env` submodule for the conda glue, then includes `scripts/make/git.mk` for the version-related targets, so a fresh clone needs its submodules initialised before make will do anything useful. That submodule step is a single command, taken from the `Makefile` itself. ```bash cd $(TOP_DIR); git submodule update --init third_party/make-env ``` The Python side is small and readable. `requirements.txt` lists flake8 for linting and wavedrom for rendering timing diagrams out of the documentation, plus a comment explaining that `rst_include` exists because GitHub does not render `.. include::` directives in previews. The last line installs the project's own Python API straight from the tree rather than from an index. ```text flake8 wavedrom rst_include -e scripts/python-skywater-pdk ``` In `requirements.txt` the package carries no leading marker, because the trailing `-e` is what makes the editable install work. The API lives in `scripts/python-skywater-pdk`, so a reader who wants to query cell data programmatically starts there, and `rst_include` sits between the two for documentation reasons rather than runtime ones. ## Documentation generated rather than written by hand The README itself is generated, and the `Makefile` shows the whole pipeline. The source is `README.src.rst`, and the target rule depends on `README.src.rst`, `docs/status.rst` and the `Makefile` itself. It removes the old file, runs `rst_include` inside the conda environment, and pipes the result through `sed`. The substitutions are informative on their own. A placeholder token for the version string gets replaced with the actual tag version, and two cross-reference directives get rewritten into plain external links pointing at `docs/versioning.rst` and `docs/known_issues.rst`, presumably so the generated file renders correctly on GitHub where Sphinx-style references do not resolve. Warning admonitions get converted into bold text for the same reason. This tells you where the documentation of record lives. `docs/` is the real content, `README.src.rst` composes it, and the rendered `README.rst` is a build artifact. A `.readthedocs.yml` sits at the root and the published site is at skywater-pdk.rtfd.io, which is the version to read if you want the design rules themselves rather than the summary. ## Where to look when something does not work Two places, both named in the README. The first is a known issues section in the documentation, the second is the GitHub issue list for the repository, which the README points at for more detail on problems currently known to exist. With roughly 200 open issues, the tracker is where a question about a specific cell or a specific flow will meet other people who hit the same thing. The repository also points at two mirrors besides GitHub itself, a Google CodeSearch interface at cs.opensource.google and a repository on foss-eda-tools.googlesource.com. The second of those matters more than it first appears, because a foundry design kit is exactly the kind of content that large organisations are reluctant to depend on from a single commercial forge. The README lists three places to view the latest SKY130 design resources and the duplication is deliberate. One caution the README states directly. A living Google document collects inspiration from what researchers and commercial entities have done at similar process sizes, but the results in it cannot be assumed to reproduce identically on SKY130, because the manufacturing process and materials differ. Treat that document as a source of project ideas rather than as a set of designs waiting to be copied. ## Conclusion The unusual thing about skywater-pdk is not that it is open source, since foundries have shared PDKs under NDA for decades. It is that this one is readable, installable and usable by people who will never sign a foundry agreement, which is why the repository has accumulated thousands of stars from a field that is not known for open anything. Read the known issues page before a first install, treat the documented examples as the real entry point, and keep tapeout plans built on the announcement of a production tag rather than on the current experimental preview label. ## FAQ ### What is SkyWater PDK? It is a collaboration between Google and SkyWater Technology Foundry to provide a fully open source process design kit, which means the design rules, device models, cell libraries and EDA tool support files needed to design a chip that SkyWater can manufacture. The repository currently targets the SKY130 process node. ### What is a PDK in the semiconductor industry? A process design kit is the collection of rules, models, libraries and tool configuration files that a specific foundry provides for one process node. It encodes the design rules that keep a design manufacturable and supplies the standard cell and primitive device libraries that synthesis and simulation rely on. skywater-pdk is one of the few that is published openly rather than under a foundry confidentiality agreement. ### Is skywater-pdk ready for a real tapeout? Not according to the README, which labels the current content an experimental preview and states that the open source PDK is not intended for production settings. It should be usable for test chips and initial design verification, though the README does not guarantee that either. A production version tag is planned for when the kit is ready for production design. ### What process node does skywater-pdk cover and what can it do? It covers SKY130, a mature 180nm to 130nm hybrid technology. The stack includes internal 1.8V with 5.0V I/O, five levels of metal plus one level of local interconnect, high sheet resistance poly resistors, optional MiM capacitors, a shrunken SONOS cell, a 10V regulated supply, and extended-drain high voltage NMOS and PMOS devices. ## Sources - [google/skywater-pdk on GitHub](https://github.com/google/skywater-pdk) - [Issues](https://github.com/google/skywater-pdk/issues) - [License: Apache-2.0](https://github.com/google/skywater-pdk/blob/main/LICENSE) - [Project website](https://skywater-pdk.rtfd.io) - [README](https://github.com/google/skywater-pdk/blob/main/README.md) --- Hysen Labs editorial analysis, written from the project's own repository and release notes. Cite the canonical page: https://hysenlabs.com/projects/google-skywater-pdk