diff --git a/content/blog/2026/stopping-core-dumps/index.md b/content/blog/2026/stopping-core-dumps/index.md
new file mode 100644
index 000000000..7e5a51f6d
--- /dev/null
+++ b/content/blog/2026/stopping-core-dumps/index.md
@@ -0,0 +1,51 @@
+---
+title: Benefitting all our communities
+slug: "benefit-all-communities"
+date: 2026-03-11
+categories:
+  - upstream impact
+tags:
+  - open source
+  - jupyterhub
+  - community
+---
+
+Scott Henderson [noticed](https://github.com/2i2c-org/infrastructure/issues/3321) [core dump files](https://en.wikipedia.org/wiki/Core_dump) appearing on home directories
+and taking up space when a program crashes in one of the hubs we run for the [CryoCloud community](https://cryointhecloud.com/).
+While core dump files can be useful in some circumstances, they are almost never useful on a hub. They can also
+be *huge*, costing real cloud spend. Ian Caroll [noticed
+this](https://github.com/2i2c-org/infrastructure/issues/3321#issuecomment-3191672965)
+happening again on both the CryoCloud and the
+[openscapes](https://openscapes.org/) hubs as well. Ian also [pointed out a fix](https://github.com/2i2c-org/infrastructure/issues/3321#issuecomment-3191672965),
+but it would need to be adapted to run on kubernetes. 2i2c is a small organization, and at that moment
+[we did not have the capacity](https://github.com/2i2c-org/infrastructure/issues/3321#issuecomment-3514101956)
+to do that work.
+
+Separately, [Alex Mandel](http://github.com/wildintellect) from [Development Seed](https://developmentseed.org/),
+working for the NASA VEDA project *also* noticed this in their hubs in late 2025, as part of their
+efforts to audit home directory usage and control costs. They were able to build on top of the work from the other
+communities we supported, and with help from [Chuck Daniels](https://github.com/chuckwondo) and [Tarashish Mishra](https://developmentseed.org/team/tarashish-mishra/) (also from Development Seed), they were able to:
+
+1. Clean up all core dumps in their communities, and reclaim 171G of space
+2. [Operationalize](https://github.com/2i2c-org/infrastructure/pull/7588) the kubernetes setup needed to disable core
+   dumps on an ongoing basis.
+
+We were able to roll this out to all our communities, and saved communities [a lot of space](https://github.com/2i2c-org/infrastructure/issues/3321#issuecomment-4028227976). In particular,
+the CryoCloud community was saved 717G, which translates to a lot of money! Also poignant that they were the first community to
+notice the problem.
+
+We're also working on contributing this upstream to zero to jupyterhub, as this is valuble to everyone running
+a JupyterHub, not just us.
+
+## 2i2c's Value
+
+This is a loop of value that we intentionally try to encourage and bring about at 2i2c:
+
+1. Community A notices a problem for them
+2. We recognize this is a general problem affecting a lot of communities, but just observed in Community A
+3. We informally put together a collection of communities that can contribute different things to solve the problem - observations,
+   engineering time, validation, etc.
+4. We solve the problem for *everyone*, including the folks who didn't know they had a problem!
+
+This is a generalization of the overall loop of open source software adding value to the world, and I'm excited to see
+it play out in a way that's organizationally supported by 2i2c.
\ No newline at end of file