Walmart's automated grocery warehouses run peak power bills above $800,000 a month

Power is the one new operating cost in the Journal's account that comes with a before and an after figure. People familiar with the matter told the Journal that Walmart's mostly manual grocery warehouses ran utility bills of about $250,000 a month, while automated ones can top $800,000 in peak periods ^[1]. The gap is at least $550,000 a month ^[21].

The rest of the new overhead comes without figures. Engineers who keep the equipment running are paid far more than conventional warehouse staff ^[11]. Walmart's cardboard boxes, used for decades, were too large for the automated systems. It switched to shorter boxes that must be taped shut, at added cost ^[12].

On the other side sits a workforce of more than 140,000 warehouse employees ^[6], about 700 per building if spread evenly across the roughly 200 sites being automated ^[22]. The reporting does not say how many of those jobs the robots are meant to replace or what they cost Walmart, so the savings cannot be worked out from it. Some of the work is staying put. Jumbo bags of dog food still need people ^[13], and the small robots that move merchandise through the buildings fail often enough to disrupt operations ^[2].

Breakpack, pulling items out of cardboard cases to ship them one at a time, is the clearest case of a task that resisted robots, or rather, one that resisted them for years and then gave way ^[8]. Walmart and its partners searched for a long time before settling on small wheeled robots with motion and obstacle sensors ^[8]. Those robots are now going into more than a dozen warehouses, a little over 6% of the program's roughly 200 buildings ^[25].

Rob Montgomery, head of supply-chain operations for Walmart U.S., told the Journal that a key challenge is re-engineering warehouses without taking them offline ^[10]. "We are kind of in this peak complexity," he said ^[9]. Walmart expects its automation spending to peak this year and next ^[18].

Three readings fit the record. The overhead may be a conversion cost, with power and engineering bills levelling off once rebuilding ends and labour costs then falling. Or those bills are permanent, and labour costs fall by less than they rise because oversized goods and failing robots keep people on the floor. A third possibility is that the warehouses work out while in-store automation keeps absorbing money ^[15].

I think the second reading fits the evidence best. Anyone expecting fast labour savings from the 200-building program should lower that expectation. The case against is that the hardest task named in the reporting now has a working answer being installed ^[8]. A peak-period power bill also overstates a typical month ^[1]. The view is wrong if warehouse headcount falls as conversions finish while power and engineering costs stop climbing.

Walmart is pressing on. Its leadership and board treat automation as a long-term priority ^[19]. Amazon's lead in warehouse automation, and its passing Walmart as the largest U.S. company by annual revenue, have added pressure, according to the Journal ^[17]. "We know that if we wait till perfect," Montgomery said, "it'll take too long." ^[20]