In decades past, a warehouse was simply a large building used for storing goods and materials you didn’t need to access anytime soon. The best place to build these facilities was in rural areas far from cities, where land was cheap and back roads were good enough to support the occasional truck making a delivery.

But today, those facilities are more often known as distribution centers, and they’re located near major highways close to the urban areas in order to support high inventory turnover and enable lightning-fast deliveries to homes and stores. And one more thing has changed: They increasingly need access to large amounts of electrical power to run the automated systems and robotics inside that allow them to operate at such high speeds.

In fact, commercial real estate firm Cushman & Wakefield has pointed to the cost and availability of electrical power as a growing concern for companies seeking sites for modern DCs. Meanwhile, logistics real estate giant Prologis recently boasted of building a DC in the Netherlands that was designed to solve that problem by making and storing much of its own energy on site through a large-scale solar network. In Prologis’ words, “Across the world, companies are running into the same problem: They want to grow—but the power grid can’t keep up. From the U.S. to Europe to Asia, grid congestion has become a barrier to progress, slowing projects that could bring jobs, innovation, and investment to local communities.”

However, even a widescale shift to solar won’t provide nearly enough power to support another type of technology that’s becoming increasingly crucial for warehouse operations—artificial intelligence (AI). Many forecasts for 2026 say that AI will begin to take over large swaths of logistics tasks currently done by company employees, whether it’s answering phones, forecasting inventory needs, or planning delivery routes. To meet the expected demand, major tech companies are rushing to construct the infrastructure that supports AI, which runs on graphical processing unit (GPU) chips inside powerful computer servers located in another kind of DC—the data center. And as we’ve all heard by now, data centers consume staggering amounts of electricity.

If AI expansion comes anywhere close to those projected levels, the estimates for those new power needs are stunning. At a recent trade show in New York held by the enterprise software vendor IFS, speakers said society will need a 50% increase in electrical power production by 2030 to support AI. Another speaker set the bar even higher, projecting that demand for electricity would triple by 2050 when those AI power demands are added to “the electrification of everything,” the term for global decarbonization efforts aimed at switching households, transportation operations, and industry from carbon-based fuels to electricity, according to Sabine Erlinghagen, CEO of Siemens Grid Software.

The limitations of electrical grids are already beginning to slow the development of one kind of DC—the distribution center—and will soon start to restrict the growth of the other type of DC—the data center. In coming years, both of those bottlenecks could have major effects on the entire supply chain industry.

The German technology provider Siemens today said it has completed a proof of concept (POC) program demonstrating the use of humanoid robots in industrial logistics, using bots from UK-based tech firm Humanoid.

Humanoid’s HMND 01 wheeled Alpha robot was deployed in real operations at a Siemens facility, marking the first step in a broader partnership between the two companies to test and validate how humanoid robots can be used in real-world environments, the firms said.

The pilot required the robot to perform a tote-to-conveyor destacking task within Siemens’ logistics process, wherein the robot autonomously picked totes from a storage stack, transported them to a conveyor, and placed them at the designated pickup point for human operators.

Specifically, the POC began with a first phase of building a physical twin to support testing, optimization, and rapid iteration. And the second phase involved a two-week on-site deployment at the Siemens Electronics Factory in Erlangen, where partners assessed the robots in a real-world production environment.

According to the firms, the robots met goals including a throughput of 60 tote moves per hour, operation with two different tote sizes, continuous autonomous task execution for more than 30 minutes, and uptime exceeding 8 hours. The project also achieved an overall pick and place success rate and an autonomous pick and place success rate above 90%.

“At Humanoid, we are a commercially driven company. Our focus is on creating robots that deliver measurable value in real-world settings. Working closely with industrial and technology partners allow us to validate our systems against real operational requirements and understand which use cases matter outside the lab,” Artem Sokolov, founder and CEO of Humanoid, said in a release.