The artificial-intelligence boom is raging, fueled by a mad dash to add computing capacity. Tech giants are funneling billions of dollars to construction companies and industrial suppliers of equipment and power to build the vast data centers that the technology requires.
It feels frothy and ripe for a correction because of the hype about AI and the historical similarities to the dot-com bubble of the mid-1990s, which was propelled by unrealistic valuations and risky bets on flimsy business models chasing the dream of online riches until it burst early in the new century. And while the internet’s promise to increase efficiency and transform society certainly came true, the Nasdaq Composite Index didn’t match its March 2000 peak until 15 years later.
This history has made some investors wary of an overshoot on this AI infrastructure build-out, even though it’s in early stages. The only way to justify the expenditure in their eyes is for AI to provide the return on investment that companies and individuals are willing to pay for. The market wants to see use cases — beyond summarizing a missed meeting or drafting a slick memo — that increase worker productivity and make the technology indispensable and ubiquitous.
Evidence of these cases is starting to pop up in earnings reports as companies embrace AI tools.
Trane Technologies is deploying AI agents to manage building systems, such as air conditioning and lighting. Nu Skin has an in-house large language model that will give customers beauty tips while collecting data to sell more personalized products. Agentic AI is the “next frontier” at Allstate Corp., Chief Executive Officer Tom Wilson said. These digital agents will speed claim settlements and offer tailored services. Colgate-Palmolive is using AI to improve inventory management and to personalize marketing to customers.
Two companies in the trucking industry, though, especially exemplify how AI tools are bolstering profits even as competitors struggle amid a three-year freight downturn.
C.H. Robinson Worldwide Inc., a freight forwarder that matches cargo with carriers in all modes of transportation, touted its use of AI for increasing market share and posting the best quarterly operating margin since 2022. The company, which employs 450 engineers and data scientists, handles 40% more shipments per employee at its trucking unit than three years ago as AI helps automate the “quote-to-cash life cycle of a load,” CEO Dave Bozeman said on a third-quarter earnings call.
The freight broker is still in the early phases of its AI rollout, Bozeman said, and the technology has worked well enough to boost the expected AI-related gains on adjusted operating income in 2026 to $336 million from $220 million in 2024.
The freight forwarder “is the first company in our coverage seeing material leverage in AI investments that should continue to scale in ’26,” Jason Seidl, an analyst with TD Cowen, wrote in a note after third-quarter earnings. C.H. Robinson’s shares have gained 45% this year, far outpacing competitors and the S&P 500 Index.
XPO Inc., which consolidates small loads into one truck, said AI was instrumental for increasing adjusted operating income by 10% in the third quarter and expanding margins by 150 basis points in a historically long cargo slump. The carrier is using AI to increase sales by sifting through reams of data to pinpoint potential new customers and to cut costs by cutting the miles driven to deliver loads and reducing hauls of empty trailers.
“Notably, productivity was the largest contributor to our strong cost performance in the quarter, enabled by our AI-driven optimization tools, which are generating measurable returns,” CEO Mario Harik said on a conference call with analysts. XPO’s shares have climbed about 5% this year compared with large declines of 20% or more for peers such as Old Dominion Freight Line Inc., Saia Inc. and ArcBest Corp.
Those measurable returns are why the AI boom has legs. Both C.H. Robinson and XPO have large internal teams of software engineers who have been deploying automation for years on proprietary systems. The step-change in productivity is more recent as AI tools have improved, and the computer power needed to run them is only increasing.
Smaller competitors that don’t have an army of coders will have to adopt the technology or face being left behind. The US economy — powered by entrepreneurs — is good at filling a need, and third-party software-as-a-service companies will help small companies with the transition.
This adoption will take time and will likely be pervasive just as Microsoft Corp.’s Excel spreadsheets and Adobe Inc.’s Portable Document Format swept through offices large and small in the 1990s and profoundly changed how work was done.
In other words, the demand will be there for this AI build-out as more companies begin their “AI journey,” a term that has become popular among executives. AI infrastructure and power supply may require investment of more than $5 trillion, according to a JPMorgan report. The pace of data center growth will be throttled by the availability of power as the backlog for gas turbines stretches to three to four years and nuclear energy has an even longer timeline, the report said. The availability of capital and even real estate may become a constraint at some point. These pinch points will help keep overcapacity in check. The investment may be more recurrent because the computer chip life cycle is much shorter than fiber-optic cable, which can last two decades or more. In a decade or less, quantum computing may become a meaningful contributor to AI capacity.
For now, the AI boom has swelled profits of companies that directly supply data centers, such as Vertiv Holdings Co. and Amphenol Corp. It’s also lifted companies like GE Vernova Inc., which makes gas turbines, and Prologis Inc., an industrial real estate firm. Now, the productivity benefits are beginning to crop up in the profit margins of AI customers. Those efficiency gains are the sturdy legs that are needed for the AI boom to continue.
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.
Read more articles by Thomas Black
