According to this[0] podcast some captains of industry in SV consider that bubbles are good, actually. At least, that is the thesis of this[1] book published by Stripe Press. Not because they'll deliver returns for their investors, but because they drive innovation in the long run. But obviously, if they are perceived to have terrible ROI, they aren't getting over the line, which leads to a dilemma – how do we reach that moonshot without fooling a bunch of people and destroying the economy along the way?

All that said, anyone who spends a lot of time with AI knows the current direction of generative AI isn't really that moonshot. But the scale of compute that is being unlocked right now, along with technologies like photonics, might be what's key to AGI.

[0] https://shows.acast.com/the-david-mcwilliams-podcast/episode... [1] https://www.stripe.press/boom

they have 2-5 years... then we're gonna need to replace all those GPUs