AI data centre finance: building a bigger and bigger bomb
The AI data centre boom may carry some sort of financial risk. You might have heard this theme here from time to time. The Bank of England was warning about AI exuberance last October. The Bank of International Settlements — the central bank for central banks — warned in June this year that what…
The AI data centre boom may carry some sort of financial risk. You might have heard this theme here from time to time. The Bank of England was warning about AI exuberance last October. The Bank of International Settlements — the central bank for central banks — warned in June this year that what goes up has some chance of coming down: [BIS] The current surge in capital expenditure could prove unsustainable if supply bottlenecks restrain production. Intense competition for market leadership may fuel overinvestment further, as seen in previous innovation waves, increasing the risk of a sharp reversal if AI payoffs disappoint. … Easy financial conditions could tighten and become a potent amplifier in adverse scenarios where interest rates rise and AI payoffs disappoint. The incredibly obvious is still happening. The data centre buildout needs several trillion dollars over the next few years — and that’s if things go well and all the numbers keep going up. The data centre developers have to borrow enough money to fund these things. And the lenders are worrying this might be a bit risky. [FT, archive] If the lender uses the Nvidia GPUs as collateral, the chips have a short life. They’re obsolete in a few years — if they don’t just fail from overheating. The general public hate data centres. Politicians from both US parties have worked out they need to say no to data centres if they want to get reelected in the midterm elections in November. If a data centre falls to local activism, how does the lender get their money back? This is not always clear! Lenders try to protect against the risks with insurance. But the insurers think the projects are too risky as well. Richard Myers of Morgan Stanley has an idea who can take on the risk: Pensions and insurance are obvious candidates for this type of stable and low-risk investment profile. Stable and low risk in your pension, eh. But even those investors are not getting in. Edwin Wilches of PGIM Credit told the FT: We don’t know what this data centre is going to be worth in 10 years. The world got a rude awakening … Investors were like, that’s not what I thought it was, and they started to reprice the risk. The risk is getting pricier. Standard & Poor’s, the largest credit rating agency, says: [FT, archive] a full half of the economic growth coming from the US private sector was linked to AI-centric activity over the past year. A lot of that “growth” is the same alleged dollars going in a circle. It’s not clear it’s real. The hyperscalers do not quantify their returns on AI. You might think that’s because the returns are really bad. But S&P makes a generous assumption: the largest companies in the world probably know what they’re doing and that their current and future investment all works out. That’s quite the assumption. But even then, the numbers work out badly: Every time we take a deep dive into this sector, we find that capex is rising faster than we anticipated, financings are becoming more complicated and less transparent, and that returns on investment will take years to realize. The hyperscalers risk having their credit rating downgraded. Which means it costs them more to borrow money. And if their credit rating goes down to a “junk” rating, that means that some juicy lenders — such as the pension funds — cannot, for the most part, lend them money or buy their bonds. Oracle is one rating drop away from this one. S&P also calls out circular financing: The broader question is whether circular financing is creating leverage collectively that is individually manageable but could become highly correlated should demand fall sharply. The scale of overlap and interconnectedness is vast. In a downturn, even the best capitalized and most profitable firms may incur substantial pain. That is, they’re all building a great big bomb. There remains the teensy risk — very minor, hardly a thing at all — that the AI bubble won’t work out: Finally, there is the risk looming in the background that the data centres and the power plants being built to electrify them will one day be stranded assets. But I’m sure it’ll all be fine. VIdeo — PodcastSource: Pivot to AI — Published — Category: Business