Create a cloud credits program for Jupyter to have dedicated cloud resources.

[fperez]

The PyTorch foundation has a cloud credits program that allows them to request contributions from industry in the form of cloud credits, that they can then allocate for various needs of the project (PyTorch).

At Jupyter we have a wide range of needs that could use this, historically the most visible ones have been Binder and NBViewer, but more recently the JupyterLab and other teams have talked about using Jupyter to host notes instead of google docs or HackMD, and one important bottleneck is the lack of a fully hosted service that can provide all the necessary pieces.

While credits alone don't solve everything (for some tasks there's also engineering needed that can be a bottleneck), credits can be a huge help. Having a core pool of credits for key tasks would relieve the pressure on people like @choldgraf and others who for so long did a huge amount of advocacy to get credits for Binder.

[choldgraf]

Just noting that this would be really helpful for Binder. Right now 2i2c is basically footing the bill for Binder and this isn't sustainable in the long term for us. We did find a way to host Binder sessions much more cheaply so it shouldn't be too expensive (maybe $6-10K a year at current volume on Hetzner or OVH, a bit more expensive on the bigger cloud providers).

We (2i2c) are also exploring deploying a Binder on federal infrastructure which would also be a nice option (though, is Binder-specific)