AWS was built from the ground up on new infrastructure. There was no reuse of existing Amazon infrastructure except tooling.

The whole "selling excess Q4 capacity" was a myth.

https://www.acquired.fm/episodes/Amazon-web-services goes into it in great detail.

Miscapitalized url, should be [0]. However, the page is mostly metadata pointing at a 2 hour youtube video [1].

[0] https://www.acquired.fm/episodes/amazon-web-services

[1] https://www.youtube.com/watch?v=APvj15_YCqk

Thanks for fixing! Mobile autocorrect fail.

It's a podcast first and foremost so the URL is still good for the direct links in various podcast players.

That doesn’t make sense in the context of google. Their own workloads are still way larger than their public cloud. Arguably their first offering, app engine, was the perfect fit because they shoved it into the cracks in between their other borg jobs. The fit of GCE in their overall platform is much worse.

It does if Google can double their footprint. They announced reduced profitability recently in part due to capex for data centers.

Last year google started warning some internal teams that they had to limit usage. It’s possible they’re running out of compute.

Google has been internally chopping budgets since decades. Is how they keep utilization up. Look up “steamroller” or “ucp” if you don’t know those.