NOT another Ai bubble post, just questions about hope, truth and direction

Lately I have been getting a lot of posts into my timeline from the likes of Ed Zitron and others who claim that Ai and all the companies involved is a massive bubble. The accusations range from straight up accounting fraud (millions of unused GPus in Warehouses) to raising questions about IMO legitimate concerns like how OpenAI will honor a trillion dollars in spending commitments on a couple billions of revenue. We all know how Sam Altman responded to the question and its anything but reassuring. Now he was asked this question again and his answer basically came down to: "People have a hard time understanding exponential growth"

Now, if the Ai industry consisted of 2-3 companies one could shrug this off and see what would eventually happen. The odd thing to me is that ALL of the most successful companies in TECH that are in amazing shape and print billions of dollars every year basically bet their shirt on Ai. Oracle is literally risking a 700 billion dollar business on this one trade, even worse on one company, which is OpenAi and who they do this massive buildout for.
Amazon, Alphabet, Meta and Microsoft all invest tens to hundreds of billions into Ai in the coming years. They are turning themselves from free cashflow giants to some of the biggest debtors in the world. This can't all be based on hope. They must know something that we don't but what could that be?

Some things seem just true no matter how you spin them:
\- There is no real path to turn compute expense into revenue (Depreciation not even considered)
\- Most Ai features that are tried in a larger corporate context are rejected (Nobody wants to use Copilot)
\- The models are hardly differentiated and it seems incredibly difficult to even make money with them at all
\- the power needed to keep scaling the buildout is not available
\- Some of the most prominent figures in Ai say that the current approach to improve Ai will never work
\- The Ai startup eco system basically is a insert 5 dollars and get 2 dollars out scheme financed by VCs (what are they thinking??)

So what could it all be about? Is Ai the great distractor from something else? If so, why would the tech giants participate and risk their companies?

Is Ai really much better than we know already and there is a clear path to creating some sort of AGI using the compute buildout? If so, rather than competing the companies must be colluding to keep this "hidden". IMO unlikely.

Is it something much more banal where early AI seemed promising but now these companies are stuck in a narrative they can't get out of? Imagine what ANY scaling back from AI, from on of those companies would do to the market.

TLTR: Why does the AI ecosystem show so many obvious red flags while attracting levels of investment that suggest those red flags don’t exist?