CapEx: What Is It Good For?

Before you break out into song and sing "absolutely nothing," let's put some data on the table.

CapEx, or capital expenditures, is exactly what it sounds like: companies spending money on property, plants, equipment, and infrastructure that, once deployed, should help increase future revenue and earnings. The challenge for investors is that the costs are immediate, while the benefits often take years to materialize.

For much of the last decade, the major hyperscalers—the companies building the infrastructure underpinning artificial intelligence—operated under an asset-light, high-margin business model. This allowed them to return extraordinary amounts of capital to shareholders through dividends and share repurchases. Today, however, the enormous build-out required to support AI is redirecting those dollars away from buybacks and into tangible assets such as data centers, semiconductors, networking equipment, and power infrastructure.

Fortunately, early indicators suggest these investments may already be bearing fruit. From 2007 through 2019, U.S. labor productivity growth averaged approximately 1.5% annually. Since the current business cycle began in the fourth quarter of 2019, that figure has increased to roughly 2.1% annually—a meaningful acceleration that, if sustained, could have significant implications for economic growth, corporate profitability, and living standards. 

Productivity is often described as doing more with less. However, history suggests that greater efficiency does not always result in less activity. In fact, it can lead to more.

This concept is known as Jevons Paradox. In 1865, economist William Stanley Jevons observed that improvements in steam engine efficiency did not reduce coal consumption as many expected. Instead, greater efficiency lowered costs, increased adoption, and ultimately caused total coal usage to rise.

Could artificial intelligence be producing a similar effect?

Despite concerns that AI could eliminate entry-level jobs, recent survey data suggests many organizations still view those positions as essential. Rather than eliminating roles, businesses appear to be reshaping them. As routine tasks become automated, employees may increasingly focus on oversight, client interaction, judgment, problem-solving, and managing AI-assisted workflows.

While it remains early, the initial evidence is encouraging. Companies are investing heavily in productive assets, productivity growth has improved, and businesses continue to discover new ways to deploy AI throughout their operations. The next several quarters should provide investors with a clearer picture of whether today's unprecedented capital spending will translate into a lasting productivity boom.

-The LVM Team