The Number That Moves Markets
When mega-cap technology companies report their quarterly results, one figure has come to overshadow nearly every other line item: capital expenditures. In a week dense with reports from the largest companies on the planet, the question that traders, portfolio managers, and long-term investors all ask in unison is whether the capex numbers will continue to climb. The reason this single line carries such weight is that it ties directly to the durability of the entire AI investment thesis. Each dollar these companies pour into compute, chips, data centers, and power infrastructure is, in effect, a wager on future cash flows. The market wants to see that the spending is justified by the revenue it produces.
For long-term investors, the emphasis on capex is not a fashion. It is the foundation of the analysis. Capex feeds directly into the return-on-capital calculation that determines whether these businesses are creating or destroying value. In the short term, however, the same figure becomes a trigger for volatility. Markets tend to anchor on the most recent print and extrapolate it forward. If revenue accelerates, the implied return on the enormous capital base looks attractive, and the stocks rerate higher. If revenue softens in any subsegment, investors immediately question whether the capex is being deployed responsibly. That mechanical extrapolation is what makes earnings days for the largest names feel so monumental for index movements.
Long-Term Discipline Over Index Mimicry
A defensible approach to managing money in this concentrated environment begins with refusing to invest in deference to a benchmark. There is little intellectual rigor in deciding to be fifty or two hundred basis points overweight a name simply because the index is heavy in it. The better discipline is to populate a portfolio with what looks like the best risk-adjusted return regardless of what the comparator says. That mindset matters more than ever in a market where a handful of stocks dominate the cap-weighted indices, because the temptation to chase index exposure can dress up momentum following as research-driven conviction.
Volatility is simply the price of admission for long-term investors. Earnings days will swing hard in both directions, but those swings are noise around a longer arc that is determined by whether capital is producing returns.
The Broadening Beneath the Surface
A subtle but important development over the past six or seven months is the broadening of earnings growth beyond the largest names. Soft data, including consumer sentiment surveys like the University of Michigan reading, has continued to look weak. Hard data tells a different story. Purchasing manager indexes have improved, and earnings growth has spread into mid-cap names, industrials, and other corners of the economy that were previously left behind. Mega-cap tech remains extraordinarily robust, but it is no longer the only place where growth is showing up.
Some of this broadening is the direct result of the capex buildout itself. The dollars being spent on AI infrastructure do not stay inside a few hyperscaler balance sheets. They diffuse outward into the supply chain, lifting companies that, on the surface, look like old-economy semiconductor analogs. Texas Instruments is a striking illustration. Long thought of as a legacy analog name, its data center business now accounts for more than ten percent of its revenue and is growing at more than ninety percent year over year. That single data point shows how capex spending is rippling well beyond leading-edge GPU makers and into businesses that the market had not associated with the AI cycle at all.
Hardware Euphoria, Software Disregard
Inside the technology complex itself, a striking divergence has opened up between hard-asset companies and asset-light software businesses. The semiconductor index has broken out to the upside, while many of the software names that will ultimately distribute this technology have done the exact opposite. The spread between companies that produce physical things and companies that license or host software has widened materially.
This divergence is uncomfortable to call a top in, because timing the hardware trade has historically been treacherous. Demand at the leading edge is genuinely off the charts, and the only honest question is how long that pace can persist. Yet the asset-light side of the market is increasingly where opportunity is hiding in plain sight. Many software names have been sold off because they are perceived to be threatened by AI, or because they have been swept out alongside their peers in indiscriminate ETF and basket selling. In a meaningful number of cases, the underlying businesses are unlikely to be impaired in the way the price action suggests. The baby has been thrown out with the bathwater.
Where the Better Hunting Ground Lies
Spending research time chasing the next leading-edge AI semiconductor name is, at this stage, a low-probability exercise. The crowd is already there, valuations price in stunning growth, and the central debate is duration rather than direction. By contrast, mining beaten-down software names for businesses whose fundamentals do not match their stock charts offers a more fruitful use of analytical effort. The pattern of looking for opportunity in correlated trades that have been sold off, rather than in the names that have already gone parabolic, is a discipline well suited to this moment.
The week ahead will deliver enormous amounts of new information. The capex line will dominate the headlines and drive the market-cap swings, as it should, since it is the cleanest signal of how the AI cycle is unfolding. But the more interesting insight may not be in the headline figure itself. It will be in how that capex is diffusing through the rest of the economy, and in which businesses, far from the leading edge, are quietly compounding returns while the spotlight is fixed elsewhere.