Robot Charging: The Distribution Load Nobody’s Modeling

Somewhere in your service territory, a warehouse that has drawn 500 kW for 20 years is about to draw 10 times that. The feeder and the substation serving it were never sized for this. Nobody has filed an amended interconnection agreement because the load increase isn't coming from a new tenant or a known expansion. It's coming from robots replacing the human workers inside.

This is not a hypothetical. Amazon's 4.5-million-square-foot facility in Ontario, California already runs more than 7,000 robots alongside 2,000 human workers. The company’s next-generation fulfillment centers, the first of which opened in Shreveport, Louisiana in late 2024, are designed for 10x this robotic density.

The energy industry has mobilized around data center load growth and for good reason. AI-driven compute demand is projected to add 100 to 200 TWh to U.S. electricity consumption by 2030. Data center load is increasingly finding its way into Integrated Resource Plans, rate case filings and distribution capacity studies. But a contrarian observation should keep utility planners awake at night: data centers are the easy problem.

Data centers are the rich demand tourists. They can afford to choose where they locate. They look for cheap power, minimal latency and a fast interconnect. They negotiate bespoke supply contracts and build in places like rural Iowa, in ERCOT and next to legacy nuclear facilities in Pennsylvania. When a hyperscaler needs 500 MW, it picks the path of least resistance.

Robots are different. An army of them is quietly taking up residence in America’s warehouses and factories. They are not in the exurban greenfields where data centers roam, but in dense industrial corridors where the grid is already loaded. These are locations where substations were sized decades ago and where the local distribution feeders were never designed for what’s coming. These robots go to where the warehouses already are. They go to where the people they are supplementing (and eventually replacing) currently work.

These robots are demand hostages. And they are multiplying.

The Math that Should Concern Every Planner

Last year Amazon deployed its one millionth warehouse robot and over 1,000 North American warehouses (including Amazon) already have integrated robotic fleets. The projections from market research firm NextMSC are staggering: North American warehouses alone will deploy over 500,000 robots per year by 2030.

The energy draw of any single warehouse robot today is relatively modest, but the aggregate draw from a warehouse full of robots is not. Each robot consumes approximately 0.5-3 kW during operation and charges at 1-7 kW. A single facility running a few thousand of them sees a step change in demand that dwarfs its pre-automation baseline. According to Mobius Risk Group, at 50% labor-replacement penetration in manufacturing and warehousing alone, humanoid robot charging would consume 27 TWh, more than 3.5 times the electricity of all U.S. electric vehicles on the road today. That figure does not include the hundreds of thousands of conventional warehouse robots already deployed.

What makes this load different from data centers or EVs is not just its scale but its character. Today’s warehouse robots, like Amazon’s Kiva fleet, operate on continuous “opportunity charging” cycles: roughly 60 minutes of work followed by five minutes at a charging station. The result is a sustained electrical baseline; not dramatic spikes, but a facility with thousands of robots drawing continuous power still represents a step change in demand that most legacy interconnection agreements never contemplated. And the next generation of warehouse automation, including heavier autonomous forklifts, robotic picking arms, and humanoid robots with larger batteries and longer charge cycles, will layer genuinely peaky demand profiles on top of that elevated baseline.

The Inland Empire: A Case Study in What’s Coming

The place to watch is Southern California’s Inland Empire. Riverside and San Bernardino Counties form the nation’s primary West Coast distribution hub out of the ports of Los Angeles and Long Beach. The region contains more than 4,000 warehouses occupying roughly a billion square feet. Amazon alone operates more than 40 local facilities and employs approximately 30,000 workers.

The associated electrical grid infrastructure, mostly served by Southern California Edison, was sized for a prior generation of warehouse operations. SCE is projecting $8 billion in annual capital deployment through 2028, and over 85% of this is directed towards distribution grid investment. But these plans do not yet account for robotic load growth at the scale the corridor is heading toward.

Today the grid is absorbing these fleets without distress, but today’s fleets represent a fraction of the workforce they are designed to replace. At meaningful labor-replacement penetration, a single large facility’s robot fleet will add several megawatts of continuous demand on top of the existing baseline. Scale that across dozens of facilities along the corridor (many served by feeders rated for 10-15 MW that were never expected to carry this kind of incremental load) and the infrastructure gap becomes concrete.

The standard behind-the-meter answer, backup generation, is not available here. This section of Riverside and San Bernardino Counties ranks as one of the most ozone-polluted regions in the nation. The South Coast Air Quality Management District’s Warehouse Indirect Source Rule requires local warehouses over 100,000 square feet to reduce their associated emissions. So traditional combustion generators are not a viable standing solution in the region.

But the typical warehouse envelope in the Inland Empire presents a real integrated energy opportunity. A 500,000-square-foot warehouse roof is a multi-megawatt solar asset. Short-term battery buffer storage, sized to the facility’s charging profile, can absorb the spikes that would otherwise push a feeder past its rating. And low-cost natural gas access makes clean fuel cells and linear generators viable sources of dispatchable generation where traditional reciprocating engines would fail air quality permitting. The combination of rooftop solar, buffer storage and clean dispatchable generation can materially reduce both the facility’s grid impact and its demand charges. This turns what would otherwise be a distribution constraint into a manageable integration hurdle.

The 5-to-7-Year Problem

Even if this load growth were perfectly predictable, the normal infrastructure response timeline would not keep pace. From identifying a new load trend to energizing upgraded infrastructure typically takes ~5 years under optimistic assumptions (incorporating the forecast into a distribution plan takes 6-12 months; filing and adjudicating a rate case takes 10-24 months; ordering and delivering transformers and substation equipment takes 2-4 years; construction takes 6-12 months).

Today’s infrastructure requirements and supply chain timelines make this particularly challenging. More than half of the nation’s in-service distribution transformers are past their expected service life and Wood Mackenzie’s August 2025 analysis “Untangling the US Transformer Supply Chain Crisis” projects supply deficits of 10-30% for distribution and power transformers persisting well into the 2030s, with power transformer lead times averaging 128 weeks. If robot charging loads scale meaningfully by 2030 (a timeline consistent with warehouse and manufacturing industry projections), the utility planning conversations should have started 2–3 years ago.

Four Questions Every Planner Should be Asking Now

Does the service territory have logistics corridor exposure? The Inland Empire, the I-78/I-81 corridor in eastern Pennsylvania and New Jersey, Metro Atlanta, North Fort Worth, and Chicago’s I-55/I-80 corridor are the most exposed clusters. If a service territory includes dense warehouse districts like these, it has the problem.

Do interconnection agreements reflect robotic load? Most legacy agreements do not contemplate a 5x-10x increase in facility demand. Utility account managers need to dig deep into their C&I customer long-term plans and Commissions should require warehouse operators to file amended agreements well in advance of deploying robotic fleets that push peak demand beyond original design thresholds.

Are feeder, transformer and substation plans modeling the right load categories? Every IRP has line items for data centers, EVs and electrification. If there is nothing for automation in warehousing and manufacturing, there is a blind spot at the distribution level precisely where robot charging loads will materialize first. Warehouse and manufacturing automation should be a distinct demand category in Integrated Resource Plans and distribution capacity studies.

Are robots being treated as pure loads or potential grid assets? A 500-robot facility would likely have up to 1 MWh of centrally managed battery capacity available at various points in time. Unlike passenger EVs, where individual schedules and range anxiety make vehicle-to-grid programs complex, warehouse robots run predictable, centrally managed charge cycles. This is a demand response and ancillary services resource; tariffs and interconnection rules should leverage it.

The first commissions and utilities to incorporate robot charging into their planning frameworks will turn a reliability risk into a revenue opportunity. Their service territories will become the preferred locations for the next generation of automated logistics, precisely because the power will be there. For everyone else, the load is coming anyway. If the planners don't decide, the transf