The AI bubble could be worse than the dot-com bust

[SamuelTheThird]

https://thehill.com/opinion/finance/5925202-tech-bubble-ai-driven-growth/

However, extrapolating from recent robust growth in semiconductor and memory chip revenues and expecting sustained long-term outperformance is risky, as it relies on the uncertain assumption that the current boom in AI infrastructure spending will continue unabated. Major hyperscalers are already resorting to additional debt and equity financing to support their elevated capital expenditures, since even their substantial free cash flow is becoming insufficient to fund the escalating investments needed for AI infrastructure.
----

If the AI bubble bursts, the economic fallout could be far more damaging than the dot-com crash. The U.S. economy is in a weaker position today, with growth increasingly dependent on high-income consumers, a trend highlighted by a recent New York Fed study. At the same time, substantial AI-related capital expenditures have become an important driver of recent U.S. GDP growth, raising the risk that any sharp pullback in investment could have broader economic repercussions.

Unlike today’s unbalanced economy, the late-1990s expansion was supported by stronger overall economic performance. Real GDP increased by more than 4 percent annually from 1997 to 2000, and the resultant prosperity was widely shared.

Furthermore, in contrast to the 1998 to 2001 period, when the federal government posted budget surpluses for four years in a row, the U.S. has experienced large deficits since 2020. Federal debt has climbed to more than 120 percent of GDP, compared with 54.5 percent in 2001. Such historically high levels of public debt create significant financial risks. If the benchmark bond yield rises above 5 percent, it could trigger a sharp repricing of risky assets and deflate the AI bubble.

[EdmondDantes_]

1. And unlike the dot com bubble, the infrastructure isn't as reusable

Everyone could reuse the network cables, but a 3 year old GPU isn't as readily usable.

[ultralite001]

2. Particularly if they've been worked hard...

Overheating + overclocked use + abuse will make much equipment worthless...

[edhopper]

3. The difference is the internet was a real economic

boom. There were just too many companies chasing the same thing. But the profits from the internet were there for those that survived. AI is unwanted and unneeded and it's costs don't warrant the expenses.

[MineralMan]

4. Yes. Yes, it could.

There is a lot of wishful thinking in the AI world. If only they could build an unlimited number of data handling facilities, success would be certain. Or so they believe.

They're wrong, I believe. They're investing in something that is unproven so far. If it works the way they hope it will, then they'll be in clover. However, if it gets out of control and runs into houses and kills old ladies, like a Tesla running on AI did, things might not work out as well as they hope.

We shall see. I'd not rush to invest in those data center ventures, frankly. They may pay off, but then, again, they may not pay off at all.

See this, from Texas:

"Tesla allegedly in driver-assist mode crashes into Texas house, woman killed: Investigators"
https://abcnews.com/US/tesla-allegedly-autopilot-mode-crashes-texas-house-woman/story?id=134062374

[Matthew28]

5. let's hope it doesn't happen

I don't like people getting hurt and it won't be good for our leadership in tech. China will become the leader without question in that case.

[CivicGrief]

6. That's the fear they want.

Too big to fail should not work this time.