About us

The compute already exists. We're putting it to work.

Since 2022 we've built software that makes many machines work as one. Today we build Cascadia, private AI infrastructure for the computers enterprises already own, co-developed with Intel.

our mission

Powerful AI shouldn't depend on owning a data center. We're building the systems that let the machines you already own do the work.

They’re racing to build the largest machine.
We teach the small ones to think together.

  1. 2022

    Community Labs is founded

    A $30M strategic round led by Lightspeed Venture Partners funds infrastructure for distributed networks.

  2. 2023

    Nine companies built and launched

    Our studio designs, engineers and ships products with founding teams, from developer tools to consumer apps.

  3. 2024

    Cascadia begins in our incubator

    Community Labs starts incubating Cascadia: our distributed-systems work, pointed at private AI.

  4. 2026

    Cascadia launches, built for Intel AI PCs

    We co-developed Cascadia with Intel's AI PC team, and published three research papers. Pilot programs with design partners began.

  5. now

    Frontier AI on the machines you own

    We're consistently raising the size of model a fleet of ordinary computers can run, so every company can run powerful AI locally.

leadership

Meet the leadership team behind Cascadia.

  • Tate Berenbaum, Chief Executive Officer

    Tate Berenbaum

    Chief Executive Officer

  • Kyla Witte, Chief Operating Officer

    Kyla Witte

    Chief Operating Officer

  • Matias Parij, Chief Technology Officer

    Matias Parij

    Chief Technology Officer

  • Saif Nashi, Head of Sales

    Saif Nashi

    Head of Sales

backed by

Funded by investors who back infrastructure for the long run.

Lightspeed Venture Partners
Bain Capital
Distributed Global
CMT Digital
Road Capital
1kx
Blockchain Capital
Permanent Ventures

Private AI on your own machines, in a week.

We set up your first model on your network in under a week. Prompts and data never leave the building, and you pay per machine, never per token.

See it run on hardware like yours.

Book a demo

Check the engineering first.

See our research