The freight market is behaving in unusual ways, and AI data center construction is a major reason why. Commodity shippers hauling chemicals, plastics, building materials, and metals are paying premium rates into a soft-volume market — and the AI data center boom is a primary driver, according to Blake Azell, Vice President of Customer Success and Support at IntelliTrans, a Roper Technologies company and sister company to DAT.

Azell calls the phenomenon a "capacity tax": rates are elevated not because demand is broadly strong, but because flatbed and specialized capacity is being absorbed by hyperscaler construction projects at margins traditional commodity shippers cannot match. The dynamic is particularly striking because it comes amid a broader freight recession that began in 2023, following the post-pandemic capacity crunch of 2021–2022. What is different this time is that a single, capital-intensive end market is distorting specific equipment segments — flatbed and specialized — even as overall tonnage weakens.

"Our commodity shippers are now almost getting into a bidding war with the tech giants — and that's really where the concern is," Azell said during an appearance on FreightWaves Today.

The Scale of Data Center Investment

The scale of hyperscaler spending underscores the pressure. These companies are spending the equivalent of what the interstate highway system cost annually — roughly $20 billion a year over 35 years — every two weeks, totaling approximately $700 billion in projected outlays. By comparison, the interstate highway system's total cost over more than three decades is being matched by tech giants in a matter of days.

Microsoft, Google, Meta, and Amazon have each publicly disclosed multi-year capital expenditure plans running into the tens of billions annually, with AI infrastructure cited as a leading use of funds. The buildout is further amplified by federal legislation, including the CHIPS and Science Act, which is driving concurrent construction of semiconductor fabrication plants that also require massive flatbed and specialized freight for equipment, structural steel, and process materials.

A single 500-megawatt data center requires an estimated 30,000 truckloads of concrete, steel, copper, fiber optics, PVC pipe, generators, and transmission equipment. A recently announced 10-gigawatt facility in Utah — 20 times that size, associated with Kevin O'Leary — would multiply those freight needs accordingly.

Only 8% of announced and contracted data center projects are currently under any level of construction, pointing to years of sustained flatbed demand ahead. Current estimates put truckload volumes at approximately 3 million this year, with annual increases projected to continue. Between 50% and 70% of those truckloads move in flatbed or open-dimensional markets, with additional volume in bulk for materials like concrete and gravel.

A Structural Mismatch, Not a Cyclical Recovery

Against that backdrop, overall freight volumes have weakened. Azell cited roughly 4% volume erosion since 2023, with chemicals, forest products, non-metallic minerals, and motor vehicles either flat or declining on the rail side.

Yet the flatbed load-to-truck ratio has reached 73-to-1, and tender rejections are running at 16%. Azell described this combination as a structural mismatch rather than a cyclical recovery — a paradox where shippers are paying premium pricing into softened volumes.

For years, the assumption has been that a tight market with high prices signals a healthy market. Azell argues that is not the case today, based on conversations with shippers across the bulk and breakbulk segments. The divergence also complicates freight procurement strategies that were built around the assumption that all equipment types move together with the broader economic cycle.

Driver Supply and Carrier Economics

Driver supply has compounded the squeeze. Azell estimated the industry has lost roughly 250,000 drivers since 2020 through a combination of retirement, regulatory enforcement, and other attrition. Fuel surcharge economics continue to pressure the small independent carriers that dominate the flatbed and specialized segments.

Because hyperscalers are largely price-insensitive to freight cost increases — their capital spending is backstopped by strong cash flows and rising equity valuations — they can outbid traditional commodity shippers for available capacity without meaningful financial pain. These companies are among the most profitable, operationally cash-flowing enterprises in history, making it unlikely that freight cost inflation will deter their buildout plans.

Azell noted the irony: AI and cloud computing, designed to improve efficiency and reduce costs, are simultaneously driving up costs across other commodity markets that support food, housing, and other essentials.

Global Competition and Permanent Structural Change

The investment is not limited to U.S.-based tech companies. Saudi Arabia is investing hundreds of billions of dollars in data center infrastructure, having converted a massive desert installation into a data center facility. The U.S. Department of Defense has allocated a significant portion of its budget toward data centers. Competition with China, Europe, and Middle Eastern sovereign funds is accelerating investment globally.

Construction labor is also being affected, with workers shifting to AI data center projects — including from homebuilding — because of the wage premiums these projects can offer. That labor competition feeds back into housing affordability, a dynamic already under pressure from persistent construction-cost inflation.

Power demand from data centers is growing exponentially after decades of relatively linear increases in power consumption. Utilities and grid operators in several U.S. regions, including PJM Interconnection and ERCOT, have publicly flagged data center load growth as a factor in long-term resource planning. Azell also noted a resurgence in coal transport to power these facilities, reflecting the massive energy requirements.

The data center market itself is projected to grow from roughly $83 billion currently to approximately $150 billion over the next five years, according to figures Azell cited.

"For the next foreseeable future, unless there are structural changes within the data center environments and regulation," Azell said, "those markets will continue to stay very tight."

Four Recommendations for Shippers

Azell outlined four steps shippers should take to navigate this environment:

1. Abandon national rate averages. Shippers should completely discard national indices and instead benchmark by specific lane, region, and freight type. National averages no longer reflect the reality of divergent local markets.

2. Renegotiate or validate contract rates. Rates and contracts negotiated a year ago may no longer support capacity commitments. Small independent carriers are "bleeding OpEx" under fuel surcharge pressures and are becoming selective about which loads they accept. Shippers need to confirm that their current and upcoming rates still secure reliable capacity, and should build tiered carrier backup plans.

3. Pursue modal arbitrage through rail. Rail capacity utilization is sitting near 70%, and rail rates are up only about 2%, making transload moves a cost-advantaged alternative where lane geography allows. Both intermodal and bulk rail present compelling opportunities for shippers able to restructure their freight networks.

4. Invest in data strategy before chasing AI tools. Azell argued that the competitive edge lies in clean, accessible underlying data rather than in the AI algorithms built on top of it. "It's the race to the underlying data to help with this problem," he said, "and it's not the race to the AI algorithm."

Looking Ahead

With only 8% of announced data center projects currently under construction, and with hyperscalers, sovereign wealth funds, and defense agencies all accelerating investment, shippers should treat the capacity pressure as a permanent structural feature of the freight market rather than a temporary disruption. Even where states and municipalities have imposed moratoriums on data center construction, the pipeline of contracted and announced projects ensures sustained flatbed demand for years to come. Grid interconnection delays, permitting timelines, and power generation constraints may slow individual project timelines, but they extend rather than eliminate the construction window — meaning flatbed and specialized demand could persist even longer than current schedules suggest.

Source: FreightWaves