Ads will certainly generate revenue but it won't miraculously help OpenAI generate profits.

> 2/ Stop training new models

Google et al won't stop training new models.

I remember reading about the cooling effects of aerosols about a decade ago. It's not like this was a mystery even back then.

See this article from 2013:

https://news.mit.edu/2013/the-global-warming-conundrum-green...

What about 50-100 years from now when most of the ice will have melted?

There are tons of options for sharing state between islands.

The docs show how to use nanostores but you can use other libs like vue refs, etc.

https://docs.astro.build/en/recipes/sharing-state-islands/

It's not and will probably never be.

If you want some precedent look at Hono. Initially it was just for the CF Workers runtime (not developed by CF). Then CF started using Hono internally and hired the dev to work on Hono full time. Hono works on any JS runtime.

https://hono.dev/docs/concepts/web-standard

I suppose we’ll see — I’d be surprised if the whole promise of nothing will change don’t hold over the long term. I don’t love that Astro is tied to a narrow set of hosting providers either.

Astro runs on many JS runtimes. It can be deployed anywhere.

A fix is coming: https://github.com/withastro/astro/pull/15227

Substack only really makes sense if you want to monetize your content.

Probably better support for CF Workers/Pages and better integration with Wrangler.

Astro is amazing. I've been using it for a couple of years now. Initially only for static sites but now I'm building the UI of all my web projects with it.

I wonder if there will be some sort of collab between Hono and Astro given that Yusukue also works at Cloudflare.

Yup, there's a lot of overlap especially with the HonoX framework that has island architecture as well!

This is the main thing for me. If I can keep the cf workers backend in the same repo and deploy them together I will consider leaving Next.js for good.