S&P 500's four-year bull market leans on ten companies that make up 40% of the index

Treat the S&P 500's 117% gain ^[2] as a stand-in for the change in its total market value. That ignores buybacks, new shares and changes in membership, but it allows a rough split using J.P. Morgan Asset Management's figures: the ten largest companies held about 28% of the index in October 2022 and hold about 40% now ^[12]. On those numbers the top ten's combined value is about 3.1 times where it started (0.40 times 2.17, divided by 0.28), and the rest of the index is about 1.8 times (0.60 times 2.17, divided by 0.72) ^[19]. The ten rose roughly 210%. Everything else rose roughly 81% ^[19].

The gap supports the claim that the run depends on AI, or rather on the data-center budgets of a few large buyers ^[7]. It does not show that the run rests almost entirely on AI. The CNA report credits the profit growth to AI spending and a solid economic backdrop together ^[3]. Oxford Economics estimates that about one-third of recent US economic growth stems from AI, including some wealth effect from stock gains lifting consumer spending ^[8]. Part of that third runs through share prices, so a fall in AI stocks would take away some of the growth credited to AI ^[8].

"The AI theme is the defining feature of this bull market," said Anthony Saglimbene, chief market strategist at Ameriprise ^[4]. He said there will be "more pressure on especially technology companies to prove that the spending that they're doing today is actually going to translate into the profits" ^[5].

If the spending does turn into profit for the companies doing it, the concentration is what Angelo Kourkafas of Edward Jones called "a reflection of fundamental strength and earnings outperformance" ^[13]. A second path keeps the AI trade intact while the Fed's hikes slow the economy, perhaps severely ^[15]. In a third, AI disappoints, and in Kourkafas's words "portfolios might feel it in an outsized way" ^[14].

I think age is the smallest of the worries. By the count of Ryan Detrick of Carson Group, only seven bull markets since World War Two have lasted longer than this one ^[6]. "They don't end of old age; they end from disease," said Mark Hackett, chief market strategist for Nationwide ^[18]. The disease I would expect is a cut in hyperscaler data-center budgets, since those budgets lift this year's earnings forecast ^[7]. A cut would hit that forecast and the AI-exposed giants at the top of the index at once ^[11].

The view is wrong if the nine sectors that have trailed the index ^[21] start beating it while the Fed is still raising rates ^[15]. That would mean profit growth is coming from the wider economy ^[3] and not mainly from AI spending.