How the AI data center building boom could cause a financial crisis akin to 2008, including the rising use of short-term debt financing via private credit funds
This time let's think about a financial crisis before it happens. — The U.S. economic data for the last few months is looking decidedly meh.
NoahpinionNoah Smith
Context & Ripple Effects
AI infrastructure funding had already drawn large asset managers into the sector, with major firms committing hundreds of billions to data centers amid oversupply concerns. Later coverage of the buildout’s scale and ROI limits sharpened the question of whether projected demand can support the capital being deployed.
This analysis focuses on the financing layer: short-term borrowing through private-credit funds can make a long-lived construction asset dependent on continual refinancing. That concern sits alongside the IMF’s warning that a sharp reversal in the AI boom could threaten a resilient global economy.
First-order effects
Data-center developers and their private-credit lenders face greater rollover and liquidity exposure when construction is financed with short-term debt rather than matched-duration capital.
If funding conditions tighten or expected data-center cash flows disappoint, projects dependent on refinancing can face abrupt financing pressure before the underlying assets are fully productive.
Second-order effects
Private-credit funds may respond by tightening underwriting, raising financing costs, or reducing new commitments, slowing marginal data-center projects and affecting construction and equipment demand.
Asset managers and developers pursuing aggressive capacity plans would face stronger scrutiny of utilization and returns, extending concerns raised by the scale of planned and stalled U.S. capacity.
Third-order effects
If short-term private-credit financing becomes a standard way to fund long-duration AI infrastructure, the sector’s capital cycle could become more sensitive to liquidity conditions than to computing demand alone.
A synchronized pullback among lenders and developers could turn localized project stress into a broader financial-stability concern, though the available coverage does not establish that such a crisis is inevitable.
The trend: AI infrastructure is evolving from a capex boom into a financing-cycle risk, as private capital funds long-lived data-center assets with potentially fragile refinancing structures.
Really, when the bottom of the economy falls out (ai bubble pops and healthcare system collapses) I honestly don't think that there is enough white supremacy and racism the trump admin can do to prevent people from correctly recognizing that the newly minted autocrats are the cau…
the thing about the 2008 financial crisis is that in 2007, people generally still had food and disposable income. the global economy and the average citizen in 2025 is way more fucked, there's just no breathing room to have a financial crisis rn and, like, still survive
The latest figures show a trend: construction of office buildings in the US continues to fall - close to the low point of the fin. crisis. At the same time, data center construction is reaching a record high: over USD 40 billion in June, 28% more than in 24 and 190% more than wh…
Numbers on AI build out are mind-blowing. One new data center in Wyoming will more than double electricity consumption in the state. Majority of US growth is AI build out — One news program was discussing AI and the moderator asked if AI was a bubble. The response: of course…
My guess is the AI bubble popping will be similar to the dot com bubble popping at the turn of the century. There's a real technological advancement, it will have real long-term impact on the world, but a lot of the money now is hype, FOMO, and irrational exuberance. — But I g…
Overall market gains for years have been largely due to 7 tech companies, all of which are deeply, deeply sunk by the money they wasted on “AI.” The correction/revaluation is going to hurt way worse than 08-09. — I think reporters are trying hard *not* to burst the bubble. [e…
Crypto and AI are growing enmeshed into the banking system as leverage fuels both. This is the necessary precondition for turning the crypto and AI bubbles into a financial crisis. It is already larger and more enmeshed in the financial system than the dotcom bubble. It seems …
'If all the private credit funds are lending to data centers, then their correlations are probably pretty high — if there's a bust in AI, a lot of them will go bust at once.' www.noahpinion.blog/p/will-data- ...
Interesting. — We are in another giant tech hype cycle with AI — where people are throwing ungodly sums of money at something they don't understand, making promises they cannot keep — without thinking of the fundamentals. — All the ingredients of another crash. — www.noahpi…
No, The 1990s Internet Bubble Didn't Fuel A Productivity Boom And Neither Will AI - https://www.davidstockmanscontracorner. com/ ... Annual US Productivity Growth, 1948 to 2024 [image]
AI can both be real and be a “bubble” if “good enough” models can be run locally / outside of proprietary walled gardens, and that eats the data center market. I don't see strong signals to have an opinion here yet
Hot take but the dot com bubble is closer to the Generative AI bubble than any other bubble in history: - Massive capex (over) buildout - Massive retail buying - Everyone is max long - Deeply negative unit economics - 99th percentile valuations - 99% of startups are AI startups
Can't shake the feeling we are “too early” for AI, the same way dot com bubble companies like Kozmo (tried to do online deliveries like Doordash, in 1998) were too early. AI compute is realistically too expensive for the average consumer and GPU quantity, power and prices may not
So far, the danger doesn't scream “2008”. But if you wait until 2008 to start worrying, you're going to get 2008. It's good to start worrying early. https://www.noahpinion.blog/ ...
A reasonable take on what a crash in the AI market would actually mean for the wider economy, as CapEx for data centers continues to grow. (To be clear, there are no particular warning signs that this is a danger right now, but downside cases are always important to consider).
i like noah and he makes a bunch of good points but like, if the economy minus ai is in recession, and then the ai boom simply dials down a little bit in temperature, and the overly optimistic P/E ratios start plummeting with earnings, is it really AI data centers crashing the ec…
We now have: * A big story about why “this time is different” * Rapidly increasing debt, funding one single sector * An opaque corner of the financial system that has recently grown * Systemically important lenders enmeshed in the new sector https://www.noahpinion.blog/ ...