I suppose one should expect ‘time to agi’ to be a function of:

1. Researchers

2. Productivity-enhancing things in the org (either technical stuff like frameworks/optimisations or nebulous social things like better collaboration)

3. Amount of available compute.

4. Luck

And then the past experience with openAI should help a lot with doing a good job of #1 and #2 even if it is just being a competent executive.