Recent ructions were a dress rehearsal for an AI bubble burst
The past week brought a lively and very useful dress rehearsalfor what a bursting of the AI bubble in markets might look like.
First off, we got a hint of what will probably cause it: not cold hard economics, not a blown fuse in an arcane bit of financial wiring, but vibes.
Every investor on the planet knows the problems here. We have a highly concentrated global stock market that is trundling along almostentirely on the fumes of speculative excitement about this one big theme. We have eyewatering levels of borrowing, often laced with rather creative ways to shove leverage off balance sheets and into places where rating agencies and investors find it hard to spot.
The supposed finest minds in the sector blame “agents” and “worms” when their tech causes havoc, and still no one knows whether it can all make enough money to pay the bills when they fall due. We all know this. And we know global financial markets, and all our pensions and savings nestled inside them, are vulnerable if anything goes wrong.
Surveys of professional investors now place AI exuberance and the hyperscaler debt binge at the top of worry lists. Just to hammer the point home, it takes some effort to avoid hearing the constant warnings from central banks and bodies such as the IMF and the Bank for International Settlements that we must not assume this will go on forever and we must be ready for systemic shocks when the music stops.
All of this was true months ago just as much as in the second half of this week — nothing had really changed. But the trade wobbled significantly nonetheless. South Korea’s famously chip-heavy stock market sank 8 per cent in a day.
Bonds issued by pure-play data centre companies took a beating and, ominously, credit default swaps — a kind of derivative that acts as protection against debt defaults — on some issuers rocketed higher. A reasonable rule of thumb in markets is that whenever you hear the letters C, D and S in one sentence, something bad is happening.
It was not a full-on market reversal — conspicuously, the US still managed to hold up on the day. But the second big lesson was a taste of what a true AI shock might do to financial markets.
We saw two sure signs that investors were trulyspooked:nice, boring, nontechy stocks such as those in the UK’s FTSE 100 outperformed. Also, US government bonds, heavily bruised after a run of beatings, finally picked up — a sign that investors want safe assets, and that they think an AI implosion could scupper the miracle that is US economic growth. Data earlier in the week showed that US inflation was simply not backing down, but bonds still picked up handily, pulling benchmark 10-year yields back down towards 5 per cent.
Unusually, given the nerves, the dollar eased significantly — the DXY dollar index is down almost 1 per cent since the middle of the week — lending weight to the theory that the piercing of the AI bubble would lead to a global retreat from US markets and dollar assets.
So this is the outline of what end times might look like: a weaker dollar, stronger bonds, a hammering for markets hooked on the semiconductor boom and a preference for old-economy stocks. Good to know.
But determining when to reach for those safety plays is harder because traders and investors cannot pick out one solid event that rumbled markets this week.
They point to several things: the alarm ringing from inside the house, in the form of Broadcom’s chief executive warning that chip investors are treating markets like a “casino”; the breezy ambition of CoreWeaveto tap insurers for funding because “that’s all the money in the world”; and thefailure of efforts to list data centre companies on stock markets in Australia and Asia.
None ofthese things matter so much in isolation. But put them all together and you have a growing pool of evidence that the brightest days of the AI trade are behind us. The easy money has been made, and now investors are asking tougher questions.
That vague sense that “something could go wrong” is starting to really gnaw at investors, as George Saravelos at Deutsche Bank observed in a recent note. Systemic risks could stem from “a safety event, a failed [stock market listing], or disappointing revenues”, he said. “Concentration risk is immense, and it is this . . . event risk that is most underpriced in markets at the moment.”
The good news for investors who struggle to shake the feeling that we are overdue a big pullback in the runaway AI trade is that the classic safe retreat — developed-market government debt — is deep in the bargain basement.
Even after this week’s sizeable pick-up in bond prices, long-term yields are still around their highest level in decades. It is not clever or fancy, but sticking to tried and tested bets such as this does stand out among the options.
The trickier task is figuring out where the next wobble might come from and what might flip it from a stumble to a crash. The AI trade has lived by the vibes, and it will die by the vibes.
Recent ructions were a dress rehearsal for an AI bubble burst
The past week brought a lively and very useful dress rehearsalfor what a bursting of the AI bubble in markets might look like.
First off, we got a hint of what will probably cause it: not cold hard economics, not a blown fuse in an arcane bit of financial wiring, but vibes.
Every investor on the planet knows the problems here. We have a highly concentrated global stock market that is trundling along almostentirely on the fumes of speculative excitement about this one big theme. We have eyewatering levels of borrowing, often laced with rather creative ways to shove leverage off balance sheets and into places where rating agencies and investors find it hard to spot.
The supposed finest minds in the sector blame “agents” and “worms” when their tech causes havoc, and still no one knows whether it can all make enough money to pay the bills when they fall due. We all know this. And we know global financial markets, and all our pensions and savings nestled inside them, are vulnerable if anything goes wrong.
Surveys of professional investors now place AI exuberance and the hyperscaler debt binge at the top of worry lists. Just to hammer the point home, it takes some effort to avoid hearing the constant warnings from central banks and bodies such as the IMF and the Bank for International Settlements that we must not assume this will go on forever and we must be ready for systemic shocks when the music stops.
All of this was true months ago just as much as in the second half of this week — nothing had really changed. But the trade wobbled significantly nonetheless. South Korea’s famously chip-heavy stock market sank 8 per cent in a day.
Bonds issued by pure-play data centre companies took a beating and, ominously, credit default swaps — a kind of derivative that acts as protection against debt defaults — on some issuers rocketed higher. A reasonable rule of thumb in markets is that whenever you hear the letters C, D and S in one sentence, something bad is happening.
It was not a full-on market reversal — conspicuously, the US still managed to hold up on the day. But the second big lesson was a taste of what a true AI shock might do to financial markets.
We saw two sure signs that investors were trulyspooked:nice, boring, nontechy stocks such as those in the UK’s FTSE 100 outperformed. Also, US government bonds, heavily bruised after a run of beatings, finally picked up — a sign that investors want safe assets, and that they think an AI implosion could scupper the miracle that is US economic growth. Data earlier in the week showed that US inflation was simply not backing down, but bonds still picked up handily, pulling benchmark 10-year yields back down towards 5 per cent.
Unusually, given the nerves, the dollar eased significantly — the DXY dollar index is down almost 1 per cent since the middle of the week — lending weight to the theory that the piercing of the AI bubble would lead to a global retreat from US markets and dollar assets.
So this is the outline of what end times might look like: a weaker dollar, stronger bonds, a hammering for markets hooked on the semiconductor boom and a preference for old-economy stocks. Good to know.
But determining when to reach for those safety plays is harder because traders and investors cannot pick out one solid event that rumbled markets this week.
They point to several things: the alarm ringing from inside the house, in the form of Broadcom’s chief executive warning that chip investors are treating markets like a “casino”; the breezy ambition of CoreWeaveto tap insurers for funding because “that’s all the money in the world”; and thefailure of efforts to list data centre companies on stock markets in Australia and Asia.
None ofthese things matter so much in isolation. But put them all together and you have a growing pool of evidence that the brightest days of the AI trade are behind us. The easy money has been made, and now investors are asking tougher questions.
That vague sense that “something could go wrong” is starting to really gnaw at investors, as George Saravelos at Deutsche Bank observed in a recent note. Systemic risks could stem from “a safety event, a failed [stock market listing], or disappointing revenues”, he said. “Concentration risk is immense, and it is this . . . event risk that is most underpriced in markets at the moment.”
The good news for investors who struggle to shake the feeling that we are overdue a big pullback in the runaway AI trade is that the classic safe retreat — developed-market government debt — is deep in the bargain basement.
Even after this week’s sizeable pick-up in bond prices, long-term yields are still around their highest level in decades. It is not clever or fancy, but sticking to tried and tested bets such as this does stand out among the options.
The trickier task is figuring out where the next wobble might come from and what might flip it from a stumble to a crash. The AI trade has lived by the vibes, and it will die by the vibes.
Comments
Post a Comment