Some AI-related stocks have been pummelled lately, while others are doing just fine. For all the talk of an AI crash, market behaviour suggests something more nuanced, pointing less to a bursting bubble than to a selective sorting of high-risk and lower-risk names.
Take Oracle. In September, shares went skyward on AI enthusiasm, briefly making founder Larry Ellison the richest man in the world, before then giving up all their gains as investors examined the financial plumbing beneath the story.
Facebook parent Meta, too, has been bloodied, losing a quarter of its value since peaking in August. The company is pouring billions into AI talent and infrastructure, yet no clear product or monetisation path has emerged, leaving investors – who have not forgotten Mark Zuckerberg’s ill-fated bet on the so-called metaverse – decidedly wary.
In contrast, Google parent Alphabet, has done just fine recently and is up more than 50 per cent in 2025.
READ MORE
Generating abundant free cash flow and with a credible route to monetising AI, it has even tempted Warren Buffett’s Berkshire Hathaway – hardly a known bubble-chaser – into taking a $4.9 billion (€4.24 billion) stake in the company (disclosure: I own shares in Berkshire).
Credit markets, too, show the distinction: the cost of insuring Oracle’s debt has roughly doubled since September, notes FT columnist Robert Armstrong, while Alphabet’s remains largely unchanged.
It all suggests markets are increasingly wary of big AI promises unsupported by cash flow. Instead, investors are distinguishing sharply between two types of companies – those with strong balance sheets, positive free cash flow and credible paths to monetising AI, and those that are debt-funding vast infrastructure without clear profitability.
In other words, there isn’t an indiscriminate, frothy AI bubble where everything with an AI label floats. Recent moves suggest a selective recalibration rather than a panic, reinforcing that not all AI bets are created equal.
