Lol is this the Hacker News version of "I am Groot".

Sidenote, I attempted a startup in the "Google Colab for GPUs" space.

You approach won't work unless you give a generous free tier and bankroll content creators to create CUDA/Mojo tutorials for the noobs.

The professional GPU devs mostly have ChatGPT writing CUDA kernels for them that they debug on their personal computers.

Your best at growth-hacking would be forking VS Code and building an IDE solely for GPU programming. Again, I'm super familiar with the space and if you'd like to collaborate then reach out on Twitter and I'll share all the growth-hacking tricks I learnt while building a "Google Colab for GPUs" startup.

Isn't Google Colab already "Google Colab for GPUs"?

Also VS Code for GPU programming is VS Code with Nvidia's Nsight plugins (look up the new Nsight Copilot), there is no gap to be filled.

> Isn't Google Colab already "Google Colab for GPUs"?

Yes it is. Same way Llama is also open. A considerable % of devs don't wanna use anything that comes from them