Warren Buffett often said that the stock market is a voting machine in the short run, but a weighing machine over longer horizons. The outgoing Berkshire Hathaway CEO also claimed not to understand technology stocks. Yet his dictum does a good job of explaining the valuations of Apple (AAPL.O), Microsoft (MSFT.O), Alphabet (GOOGL.O), Meta Platforms (META.O) and Amazon.com (AMZN.O) amid the ongoing artificial intelligence fervor.
The narrative about which Big Tech firm is ahead in AI seemingly changes monthly, with Alphabet currently in the ascendancy after the recent launch of its Gemini 3. The model, trained on internally developed chips, has wowed observers on benchmark tests. A few months earlier, Mark Zuckerberg’s Meta was catching eyeballs by snapping up talent aggressively. Arguably the only constant, in the informal AI hype league-table at least, is that Apple lags behind. The $4.2 trillion iPhone maker run by Tim Cook isn’t a cutting-edge developer of large language models (LLMs). It has also largely steered clear of the industry’s capital-intensive data center buildout.
With the market in the throes of AI euphoria, it might logically follow that Cook’s standoffish stance could incur a valuation penalty. In fact, the opposite is true. The Cupertino, California-based group trades at 34 times 2026 earnings, according to Visible Alpha data. That’s a 15% premium to the next highest-rated of the five behemoths: Amazon, whose multiple is 29.
The e-commerce group founded by Jeff Bezos is hardly an AI vibe magnet itself. Its stake in LLM developer Anthropic is smaller than Microsoft’s holding in OpenAI. Analysts also reckon that Amazon’s capital expenditure, which covers data center investments, will only rise 15% next year, compared with almost 60% for Meta and about 30% for Microsoft and Alphabet. In other words, the highest Big Tech valuations arguably belong to the two companies with the least direct exposure to LLMs, which seems odd given all the talk of an AI stock-market bubble.
There’s a simple explanation, and it’s straight out of the old-school finance textbook beloved by Buffett. Over the long run, the ability to churn out hard cash is all that matters. A key difference between earnings and free cash flow is the way that capital expenditures are handled: these lumpy investments are smoothed out over time through depreciation when calculating net income but deducted in full immediately on the cash flow statement. That’s why some Wall Streeters like to say that free cash flow is a fact, while earnings are merely an opinion.
That belief seems to be alive and well in technology investing circles today. Investors appear to be paying more for Apple’s earnings precisely because they convert entirely into real money. Cook’s annual free cash flow from 2026 to 2029 will equate to 108% of earnings on average, according to Visible Alpha data. Put simply, analysts think all of Apple’s net income will be available for dividends and buybacks. That makes sense since the company is partly just a toll collector. The group makes most of its money from selling iPhones and services that run on them, and users have shown little willingness to switch. While Cook has ramped up AI investment, it barely puts a dent in cash flow. The upshot is that Cook’s earnings are as good as money in the bank, justifying a high multiple.
At the other extreme lies Meta, whose free cash flow will average just half of its earnings in the coming years, analysts reckon. That’s down partly to Zuckerberg’s massive data center capital-expenditure splurge. The company’s price-earnings multiple, perhaps unsurprisingly, is relatively low at 22. Alphabet, Amazon and Microsoft are in between – and so are their valuations. For these five companies, the correlation between valuation multiples and the cash-flow percentage of earnings is statistically strong at about 90%, according to a Breakingviews calculation. Lower valuations for bigger-spending companies, and vice-versa, makes sense. Investors appear to be skeptical that all the data center investments will pay off, preferring cash now rather than the promise of it later.
Nvidia (NVDA.O) and Oracle (ORCL.N), meanwhile, are the oddities in this framework. Jensen Huang’s chip designer looks a lot like Apple in terms of the amount of net income that counts as free cash flow. Yet its stock is valued at just 24 times 2026 earnings – closer to Meta than the rest. One explanation is that investors have little confidence in the good times lasting for Nvidia. Hardware demand is often cyclical, and if AI hype fades, new orders for processors will dry up and margins will fall.
Oracle is a more curious outlier, at 27 times next year’s earnings. The company run by Larry Ellison is almost Apple’s mirror image. It is investing so heavily in AI data centers that the company’s free cash flow will be negative for years, according to Visible Alpha data. That makes its relatively healthy valuation a puzzle. One solution is that Oracle’s earnings, if not its cash flows, are rising fast, making the 2026 multiple seem deceptively high. Using 2028 valuation multiples instead, Ellison’s group trades below Nvidia and Meta.
Where AI goes from here is anybody’s guess. But the lesson for now seems to be that investors are paying more attention to fundamentals than all of the bubble talk would suggest. It’s a vindication, for now, of Cook’s relatively cautious stance at Apple. While rivals throw more and more capital in search of AI supremacy, the iPhone maker will probably keep its focus on returning oodles of cash back to shareholders. In the long run, that’s what investors care about most.
Follow Robert Cyran on Bluesky.
