Amazon's 1 Million Robots: What Actually Changes at Work

What Amazon’s 1 Million Robots Actually Mean

On June 30, 2025, Amazon announced that the robots deployed across its logistics operations had reached 1 million units—spanning more than 300 facilities worldwide. In the same announcement, the company introduced DeepFleet, an AI model that coordinates robot movement. Amazon said DeepFleet could improve robot travel efficiency by 10%. Amazon’s announcement

1 million robots doesn’t mean 1 million workers’ worth of tasks have been fully replaced. Robots that move shelves and robots that pick items handle different tasks, and a single piece of a person’s job is sometimes split across multiple machines.

Still, the number of people a company needs to handle the same volume of work can change. Even as volume grows, the effect might show up as companies simply not hiring as many additional people as they once would have. That’s why we need to watch how hiring and job composition shift, not just how many robots are deployed.

How to Actually Compare Robot Costs to Human Wages

The hourly cost of a robot varies by model and operating conditions. The calculation also changes depending on whether you’re just dividing the purchase price, or including installation, system integration, power, maintenance, and the staff needed to manage it all.

Uptime matters too. Even if a robot is powered on all day, it can’t work while it’s charging, being repaired, or waiting for items to arrive. And if a human needs to step in to handle exceptions, that time and cost gets added in as well.

So when actually deciding whether to adopt a robot, you need to compare hourly price alongside how much work of equal quality got done. The right benchmark is outcome-based—total cost per item moved, or time spent correcting errors—not just the sticker price.

News of a cheaper robot can be a starting point for evaluating adoption. But that alone doesn’t let you conclude it will be cheaper and better than a human at any given worksite. How reliably it performs after installation is what actually determines profitability.

The “600,000” Figure Is a Future Hiring Forecast, Not a Layoff Number

In October 2025, CNN reported on internal Amazon documents obtained by The New York Times. According to the report, if automation proceeds as planned, Amazon could scale up sales volume while avoiding the hiring of roughly 600,000 additional workers it would otherwise have needed by 2033. The report also cited a goal of automating 75% of warehouse tasks.

Amazon told CNN that the document reflected one team’s perspective and did not represent the company’s overall hiring strategy. CNN’s report and Amazon’s response

This number shouldn’t be read as layoffs that have already happened, or as a confirmed layoff plan. How much of this internal forecast actually materializes needs to be verified separately. That said, a slowdown in new hiring still affects job seekers—so looking only at layoffs of current employees won’t tell you the full employment impact of automation.

Meanwhile, Amazon says its new fulfillment center in Shreveport needs 30% more staff for reliability, maintenance, and engineering roles. Amazon’s description of technical roles

This doesn’t mean overall employment at the facility grew by 30% either. Growth in technical roles, reductions in other roles, and whether existing employees can actually transition into the new positions are three separate questions. The existence of a training program alone doesn’t guarantee every employee can make that transition.

Fixating on Humanoids Means Missing How Adoption Actually Happens

Industrial robots often don’t need to resemble humans at all. Equipment that moves items along the floor, or robot arms doing repetitive tasks at a fixed station, can take whatever form suits the job. How human-like a machine looks isn’t the standard by which every task should be judged.

Conversely, it’s also hard to assume humanoids will only be used for housework or caregiving. In June 2024, GXO and Agility Robotics announced a multi-year contract to deploy Digit in logistics operations. At a SPANX warehouse, Digit was assigned the repetitive task of moving boxes—brought in by other mobile robots—onto conveyor belts. The companies’ contract announcement

The choice can differ depending on whether you deploy a robot into a space built for humans, or redesign the workplace itself around automation. Whether a robot delivers productivity on a specific task and whether it can scale broadly across many sites are also separate questions.

There’s no guarantee a newly introduced piece of equipment stays in use. In October 2025, Amazon introduced Blue Jay, a system coordinating multiple robot arms—but on February 25, 2026, announced it was no longer using that system in operations, explaining that the underlying technology would be applied elsewhere instead. It’s a case that shows why you need to look past the adoption announcement to the operational results that follow. Amazon’s Blue Jay announcement and update

Oswarld’s Lens

While building GTM strategy, I’ve repeatedly seen the hype and anxiety that surround new technology. Stories about jobs disappearing and stories about new jobs being created came up together, but the actual changes differed by industry and by task.

I don’t think robot adoption should be blocked outright. But at a moment when AI and robotics are advancing together, I don’t think we can assume there’ll be enough time to adapt just by watching from the sidelines.

The claim that opportunity opens up for people who can use the technology also needs to be made concretely. Not everyone needs to become a robot developer. You can start by breaking your own job into pieces and identifying which parts can be automated and which still require human judgment. Roles worth examining include responding when equipment stops, checking the quality of output, or redesigning how the work itself is done.

Companies, for their part, shouldn’t stop at announcing unit counts and cost savings. They also need to explain where the people whose work is shrinking can move to, and whether there’s actually time to learn and an actual position waiting for them.

What I want to know isn’t a single answer to “who’s cheaper, robots or people.” It’s which tasks are changing, and how people can prepare for the role they’ll play amid that change. In the next issue, I’ll continue this discussion by looking at how to judge good outcomes when using AI.

---

References & Further Reading

• Amazon announces 1 million robots and DeepFleet, 2025.06.30
• CNN’s report on Amazon’s automation and hiring forecast, plus company response, 2025.10.22
• GXO and Agility Robotics’ Digit logistics contract, 2024.06.27
• Amazon’s Blue Jay introduction and operational-discontinuation update, 2026.02.25

The author is Oswarld (Kwangseob Ahn). Current roles: Adjunct Professor at Sejong University, Strategy Consultant at INLEVEL9. Career, research, books, and recent work are kept current on the About page. Latest · July 2026: HEMA-2: A Consolidation-Aware Tri-Memory Architecture with Multi-Channel Scheduling for Lifelong Conversational AI.