Let's, let's give a little bit of background on, like, what Apache Arrow is and maybe why, um, Grok, in particular, was interested in it before we, we delve too much into this story.

Yeah.

It's a columnar in-memory data store format that allows you to- Uh, allows different applications to all access the same, uh, memory and sort of interoperate, and it's all wonderful, and kumbaya says all the applications, and it's, it speeds up a bunch of the, the tools that, uh, you and I m- may use to do data science and machine learning and, uh, it sort of really underpins, uh, a lot of the AI technology that we use today.

Um, it goes hand-in-hand with Parquet very often.

Parquet is sort of like the on-disk format, and Arrow is this, uh, is the in-memory format.

Uh, and so we are heavily invested in it whether we like it or not, and we were always on the sidelines trying to make things better.

W- Like one of the very first things that we did in Groq was get nanosecond timestamps back into Parquet V2, uh, which, uh, required a whole lot of plumbing.

You know, I promise you that we're gonna be that, uh, like, at least me, as a person who is, is at the endpoint of one of those dollars that get put into open source by gross, uh, we take it really seriously, and it's, it's kind of a privilege almost to be one of the people getting to work that much upstream.