AL
4.8K posts
AL
@Almost_Finished
Technology Strategist and Architect. Everything not forbidden is compulsory: Murray Gell-Mann.



To clarify why it takes $500k & 4 months to get access to OpenAI on Azure if you are an AWS customer for a big org: If you are a small developer you can register, put a credit card down and get going (a bit harder with Azure, but still). If you are a large org with 10k developers, sensitive data, complex IT infrastructure - you just can't do that. Some things I remember we had to do (this was 2+ years ago so might be not 100%): - Create a gateway for access control & spend management - Change cloud regions as not all regions are created equal in AI land - Work out how we can send requests as we couldn't do it via the public internet - Open up proxies - Do pen testing (external company, that alone is like $100k) - Work out what data we can and cannot send and what's the governance around that - Get approvals from security, risk, privacy etc. To pay for all of this, we need to get contractors or pay internal IT teams (yes that's how it works). Azure experts, networking, engineers, PM - all of these people are not just lying around and we need to hire. There's probably 25 other things I'm forgetting - but this should give you a feel. And this was already with very strong senior sponsorship about doing this fast - lots of unblocking at a c-suite level. Lots of orgs would take a year to do this or not even bother. I wish you to not experience this yourself if you don't have to, but this is what is happening in thousands of orgs across the world.





















