Backlog orders for gas turbines drive AI data centers to shift to boiler and steam alternatives

A backlog of gas turbine orders forces AI data centers to switch to boiler and steam technology

The long waiting period for new gas turbines is pushing data center developers to switch to older industrial power technologies. Developers building data centers for artificial intelligence (AI) workloads are exploring the use of industrial boilers and steam turbines as sources of electricity. According to CryptoBriefing and Yahoo Finance, the shift is driven by a growing backlog of orders for gas turbines, which had previously been the preferred power technology for large computing facilities.

Demand for gas turbines has remained high as AI training and reasoning workloads push power demand to levels that many local power grids cannot carry. To avoid long queues for grid connection, data center operators often turn to on-site power generation strategies. However, the strategy now faces new bottlenecks. According to reports, turbine manufacturers are unable to keep up with orders, causing buyers to face long waiting times before equipment is delivered and installed.

Industrial boilers equipped with steam turbines represent a more traditional way of generating electricity. The technology has long been used in heavy industry and utility-scale power plants. Compared to modern combined-cycle gas turbines, their efficiency is generally considered low. At present, its appeal seems to stem more from the availability of supply than from performance advantages, as the procurement cycle for such equipment may be shorter than for new turbine models.

This trend highlights the broader tensions in AI infrastructure construction. Companies racing to deploy computing power face physical constraints that software timetables typically don't encounter. Power generation equipment involves manufacturing lead times, supply chains for specialized components, and professional labor required for installation. These factors are growing far less quickly than chip production or software deployment.

In the past two years, AI-related energy needs have attracted widespread attention from utility companies, regulators and investors. The power required by a large data center park is comparable to that of a small city. This has put pressure on transmission infrastructure in multiple regions and prompted some operators to consider nuclear energy, natural gas and today's traditional steam technology as alternatives.

Reported interest in boilers and steam turbines suggests that operators are willing to accept older and less efficient equipment to avoid delays. Whether this becomes a broad trend or just a stopgap measure will likely depend on how quickly gas turbine manufacturers expand production capacity.

Market Impact

If data center operators are widespread in adopting industrial boilers and steam turbines, equipment manufacturers in this segment may see a rebound in demand after years of recession. Gas turbine suppliers may face continued pressure to expand manufacturing capacity or risk losing business to alternative power solutions.

For the broader AI infrastructure industry, persistent power bottlenecks may slow the bringing of new data centers online. This could affect the timetables of cloud providers and AI companies that have publicly committed to rapidly expanding capacity. Energy supply capabilities, not just chip supply, are becoming a constraining factor worthy of attention for investors tracking AI infrastructure construction.

The reported shift to industrial boilers and steam turbines highlights the fact that while demand for computing power continues to accelerate, physical power constraints are still shaping the growth pace of AI infrastructure.

FAQs

Why do AI data centers consider industrial boilers and steam turbines?

The report shows that a significant backlog of gas turbines as the power source of choice for large data centers is forcing some operators to consider older boiler and steam turbine technologies that may be available more quickly.

Are steam turbines less efficient than gas turbines?

Steam turbines paired with industrial boilers are generally considered to be less efficient than modern combined-cycle gas turbines, but they represent a long-established power generation technology.

Why do AI data centers need so much power?

AI training and reasoning workloads consume a lot of power, and the power required in some large data center campuses is comparable to that of a small city.

Will this trend affect the growth rate of AI infrastructure?

A continuing shortage of power equipment may slow the launch of new AI data centers, as power generation schedules grow far faster than software or chip deployments.