A robust industrial rebound is underpinning the economy, and it has powered through an energy shock from the US-Iran war; elevated inflation that’s kept interest rates high and homebuilding low; and a bout of new, perhaps more permanent tariffs.
The rebound has deepened its roots beyond the artificial-intelligence boom to lift demand for construction, transportation and energy production. But make no mistake, the main driver is still the mad rush to build data centers and install all the equipment needed to power and cool the latest chips.
Doomsayers have been looking for cracks in the foundation of this recovery, but there are no signs of a letup in the massive capital spending to chase the AI dream. In fact, there are indications that this AI infrastructure buildout could take years. A couple of little-known industrials that are tied closely to data centers — Amphenol Corp. and Vertiv Holdings Co. — are good thermometers for the boom. Right now, these two data-center suppliers are on fire.
That would normally be a recipe for a boom-and-bust cycle, similar to the dot-com misadventure in the late 1990s and early 2000s. That period wasn’t just about money-losing web retailers that imploded. It was also when companies rushed to lay fiber-optic cables to unleash the promise of the internet. The overcapacity resulted in so-called dark fiber, and the drop in bandwidth prices caused several network companies and hardware suppliers to go bust. In hindsight, they didn’t lay too much cable, they just did it too quickly. That capacity was absorbed over time, and the need for new fiber-optic cable continues today.
So what will spare the AI boom from a similar fate? Capital spending budgets are still rising as so-called hyperscalers fear being left behind. But one external factor is applying the brakes, even if slightly, that could smooth out the buildup over time and help avoid a drastic bust: the political pushback against data centers in some communities.
The not-in-my-backyard movement has already slowed the increase of computing capacity. Local governments, officials and candidates have picked up on voters’ worries about increased traffic, noise and water use. Added to the pushback is the dearth of US electricity capacity and concern that consumer rates will rise. A Politico poll taken in July showed 41% oppose a data center built 3 miles from their home, rising from 28% in January.
See more:
Some of the resistance is fueled by fear that AI will eventually take everybody’s job. The internet and all the software that it enabled, such as email, spreadsheets and digital calendars, didn’t decimate the workforce, although it eliminated specific jobs. AI has already undercut jobs for transcript and translation services. Call-center workers and order processors should probably seek other opportunities.
It’s much harder to envision the new jobs that will be created by AI, but they will come. The productivity gains from AI, even it these early stages, are already showing up in company earnings. The AI revolution will be another example of a technology that allows humans higher wages and more leisure time. These technologies usually generate fear at first.
Whatever the motivation — employment fears, electricity costs or environmental concerns — opposition to projects that bring jobs and tax revenue to communities won’t make them go away. It will only slow down the construction of data centers, lengthening the buildout period and pushing them to areas that are more business friendly.
Some companies have taken the hint and are being proactive. For example, BlackRock Inc. signed an agreement with North America’s Building Trades Unions to provide labor for data center and energy projects. Anthropic partnered with Macquarie Asset Management to build data centers and pledged to cover any electricity price increases to consumers.
Technology will help mitigate water use, save on energy and even reduce noise. Vertiv offers a closed-loop cooling system that reuses all the water it needs. The industry is moving to higher electricity voltage that’s more efficient when powering computer chips. Amphenol, which makes fiber-optic connectors among other equipment, is embedding sensors in its products to monitor system health. The reality is that data centers have a light environmental footprint compared with a chemical plant, steel mill, glass furnace or pretty much any traditional manufacturing activity.
Make no mistake, the demand for data centers remains red hot. Consider the industrial thermometers, Amphenol and Vertiv. In the last three years, shares of Amphenol have jumped 281% and those of Vertiz have soared 663%, dwarfing the S&P 500 Index’s 72% gain. Vertiv makes products like its cooling system that provide power and heat management for data centers, which account for about 80% of its revenue. Amphenol gets almost 45% of its sales from data centers.
Both companies are adding capacity and acquiring companies to keep up with surging demand. In its data center business, Amphenol’s organic sales, which adjust for impacts from acquisitions and currency translation, jumped 30% in the second quarter from a year ago. Vertiv’s organic sales rose 18%. These are huge moves for large industrial companies.
Amphenol’s $210 billion market capitalization is now bigger than Boeing Co.’s. Vertiv rose from a Goldman Sachs-backed $5.3 billion SPAC deal in 2019 to a company with a market cap of $104 billion. Both are leaders in their industries and are driving new power-management and cooling technologies.
At some point, the breakneck pace driving Vertiv and Amphenol will cool. This will begin to show up in reduced orders and indicate the possibility that the buildout is starting to overshoot demand. The opposition to data centers — or AI factories, as Nvidia’s Jensen Huang likes to call them — may throttle the pace at which hyperscalers can add capacity, flattening and lengthening the curve for AI infrastructure construction and helping curb the excesses seen in the dot-com era.
It may be harder to find an acceptable backyard, but that could turn out to be a good thing for the AI industry.
A message from Advisor Perspectives and VettaFi: Discover something new! Click here to register for our upcoming webcasts.
Bloomberg News provided this article. For more articles like this please visit bloomberg.com.
