Asking for sources is always welcome with me.

Here’s a deep dive from Ed Zitron into the whole AI/LLM industry that details the heavy investment from several key banks (Deutchebank being one), and the shrinking finance availability from traditional means (bank loans, hedge funds, managed funds). It’s long but it’s really worth a read if you have a spare hour or so.
https://www.wheresyoured.at/the-enshittifinancial-crisis/

A glaring tell that I don’t recall him highlighting is that the hyperscalers have largely outsourced the risk of AI investment to others. META, Google, and Microsoft are making small bets on AI comparitively - they’re using cash assets they have as profits from other business models, which are still significant (measured in low billions) but dont require them to take loans or leverage themselves. This means they are playing it very cautiously, all the while they’re shoving AI into all their products to try to make it seem like they’re all-in and it’s ‘the next big thing’, which is helping their stock prices in the investor frenzy. Most of the investment capital required for the AI boom is going into hardware, datacenters and direct investment in the software development - and that’s mostly being avoided by the big guys. This allows them to minimize risk and still having a decent win if it takes off. Conversely If/when the bubble bursts they’ll still take a hit, but they’ll also still be making money via other streams so it’ll be a bump in the road for them - compared to what will happen to OpenAI, Athropic, Stability, the datacenters and their financiers.
https://archive.is/WwJRg (NYTimes article).