The risk of a valuation bubble in AI has dwindled, writes Tim Shufelt.Xavier POPY/Supplied
While we’ve been peering through our fingers at the mess in the Middle East, the financial markets have quietly turned against artificial intelligence.
The Magnificent Seven stocks, which include the AI “hyperscalers” spending hundreds of billions of dollars on data centres, are all down by double-digits from their 52-week highs, wiping out roughly US$5-trillion in market capitalization.
Software stocks have been obliterated. But beyond that, the entire IT space, including the main beneficiaries of AI, has taken its lumps. U.S. tech sector valuations have sunk to the point that they are basically in line with the rest of the market. That hasn’t been the case since the depths of the pandemic, nearly six years ago.
So, where did the AI bubble go? If we’re focusing on valuations as a measure of excessive investor enthusiasm, “then any such bubble in the S&P 500 has already burst,” John Higgins, chief markets economist at Capital Economics, said in a note.
Small businesses are increasingly using AI to vet and hire candidates
Any decent stock bubble sees market valuations scale to scary heights. In a true speculative mania, when investors abandon all discipline, they become willing to buy a stock at ever greater multiples of a company’s earnings.
Back in the dot-com bubble of the late-1990s, for example, the tech-heavy Nasdaq 100 index sported a bloated price-to-earnings ratio of more than 60 at its peak. Though it never got anywhere close to that in the current cycle – it peaked at 32 last October – there has been a gnawing concern in recent months that some of the same dangerous forces were being stirred.
Maybe investors can now scratch “AI bubble” off their long list of fears. Valuations have cratered. Over the last five months, the S&P 500’s trading multiple is down from a peak of higher than 24 to roughly 20, which is right about where the 10-year average sits.
But why? It doesn’t really feel like we’ve passed a major turning point in the AI trade. It’s been more of a piecemeal turnabout that started with software stocks.
Lots of traditional software providers suddenly found themselves facing an existential threat from AI tools that could let companies build cheap alternatives in house.
U.S. software stocks collectively dropped by 30 per cent. The biggest casualty in Canada was Constellation Software Inc. CSU-T, which has a massive roster of smaller software companies catering to niche industries. Its stock suffered a max drawdown of 56 per cent.
But then investors got concerned that Big Tech was spending too much on AI. The top five hyperscalers have committed around US$700-billion in AI capital expenditures this year – double last year’s already lofty total – in what is a quickly accelerating data-centre arms race.
The conflict in the Middle East put new pressure on the tech sector. The disruption to the global oil supply from the closing of the Strait of Hormuz has awoken inflationary forces, which generally translate to higher interest rates. That’s bad news for valuations in a growth-oriented sector like technology.
Two of the biggest AI spenders – Microsoft Corp. MSFT-Q and Meta Platforms Inc. META-Q – have been hammered, their share prices down by 32 per cent and 27 per cent, respectively.
Hardware stocks have also joined in the downturn. Count Celestica Inc. CLS-T among the losers in this category. As a supplier of data-centre equipment, Celestica was one of the top performers on the TSX last year, when its stock price tripled. Its shares are now down nearly 20 per cent, despite trouncing the Street’s estimates in its most recent quarterly report.
Even semiconductor companies supplying the engines that make AI run are trading at their lowest valuations in 3½ years. Nvidia Corp., the poster child for the AI boom, now has a valuation of roughly half its level when ChatGPT was introduced.
The risk of a valuation bubble in AI, which until very recently took up a lot of the oxygen in financial markets, has dwindled.
However, there still could be a bubble in how much money the industry is expecting AI and chip companies to earn, “even if there isn’t one in the price that investors are willing to pay for them,” Mr. Higgins said.
“A pullback could occur if demand for AI turns out to be weaker than anticipated or if the Iran war drags on.”
There’s always a catch.