The answer was always yes. The question was the speed and cost. AI has changed the speed, but increased the cost. Not just in the actual price of the tokens you burn, but in the increased cost of securing the code, increased review time, and increased cost of downtime because AI can build an app, but it cannot build the infrastructure to scale that app.
Will it get there eventually? Possibly. Is the added cost worth it for people trying to build something new? Yeah, maybe. But all the talk I hear of people leaping forward in their delivery contrasts heavily with the real-world experience I'm seeing of massive backlogs and breakage.
That's the right assessment, as a result of cost the final decision would be we cannot. The cost to at least prototype and experiment is dramatically lower which lets you try things you'd never have thought was possible.
> it cannot build the infrastructure to scale that app
Why do you say this? I haven't seen proof of this, inversely teams with already scaled infrastructure are able to launch greenfield at scale faster with AI.