America’s biggest tech firms are forecasted to collectively spend $670 billion on AI infrastructure capex in 2026 alone.

While funding these massive infrastructure projects has proceeded at breakneck speeds, the deployment of AI-related infrastructure has yet to keep pace with this growth.

According to a JPMorgan analysis, 60% of data center infrastructure that’s earmarked for completion in 2027 has yet to reach construction phase while another 7% has been hit with delays.

In short, construction, just like energy and advanced chips, is a bottleneck for the AI boom.

Due to this supply-demand imbalance, Sterling Infrastructure (STRL), a construction firm, has become one of the unheralded winners of the AI-driven bull market thus far.

Investor Essentials Daily:
Friday News-based Update
Powered by Valens Research

America’s biggest tech companies are undertaking spending vast amounts of capital for their respective AI infrastructure build-outs.

Alphabet (GOOGL), Microsoft (MSFT), Meta (META), and Amazon (AMZN) are forecasted to spend over $670 billion data center-related capex in this year alone, a massive surge compared to the $410 billion these firms collectively spent in 2025.

While capex for AI infrastructure has surged dramatically, the deployment of AI data centers and other related infrastructure hasn’t kept pace with the rise in funding.

According to JPMorgan’s findings, 60% of the data center infrastructure that’s scheduled for completion in 2027 has yet to reach the construction phase while another 7% has been impacted by delays.

Constructing and successfully deploying AI-related infrastructure is a tedious process. Even if a hyperscaler has earmarked billions of dollars, they would have to secure land and building permits, survey the land they’re building infrastructure on, and secure energy supply, among others.

Combined together, these issues can easily push construction timelines way beyond schedule.

In short, construction, just like advanced chips and energy, is a bottleneck AI firms have to address. As a result of this supply-demand imbalance, chipmakers and power companies have emerged as winners of the AI-driven bull market.

And in the same vein, companies that specialize in data center infrastructure construction have likewise benefitted, becoming the unheralded winners of the AI boom.

One of the winners is Sterling Infrastructure (STRL).

Sterling Infrastructure is a construction company that used to exclusively work on transportation projects. However, in 2015, it decided to rethink its business, leading to its expansion into other end markets.

It currently operates three major segments, namely E-infrastructure Solutions, Transportation Solutions, and Building Solutions.

E-infrastructure Solutions is the unit involved in preconstruction, site development, electrical, and maintenance retrofits and upgrades for data centers, semiconductor factories, e-commerce distribution centers, and warehouses. Customers include Amazon, Meta, and Walmart (WMT), to name a few.

Meanwhile, Transportation Solutions specializes in infrastructure and rehabilitation projects for highways, roads, bridges, airports, ports, and drainage systems. On the other hand, Building Solutions focuses on residential and commercial construction and plumbing.

Sterling Infrastructure’s transformation has improved its economic quality. Annual revenue has grown from roughly over $600 billion in 2016 to $2.4 billion in 2025. Its Uniform return on assets (“ROA”) over the same period saw massive improvement as well, going from -1.8% to 60%.

E-infrastructure has become the fastest growing and biggest segment of the three business units in recent years.

After delivering a revenue that fell below the $500 million mark in 2020, the segment’s sales has grown to nearly $1.5 billion as of last year.

Source: Sterling Infrastructure

The rapid growth of Sterling Infrastructure’s E-infrastructure segment can be attributed to the AI boom.

With infrastructure capex rising to record highs, demand for the company’s offerings have likewise risen accordingly. As of Q1 2026, Sterling is working on roughly 360 projects.

Moreover, Sterling Infrastructure’s combined backlog currently amounts to over $5 billion, with roughly $3.8 billion categorized as signed backlog.

Hyperscalers will do everything they can to fast-track the construction of their data centers. As a result, Sterling Infrastructure is well-positioned to capitalize, having established itself as a critical partner for the construction of AI-related infrastructure.

As long as spending trends hold, the company should be able to maintain and grow its returns in the next few years.

Best regards,

Joel Litman & Rob Spivey
Chief Investment Officer &
Director of Research
at Valens Research