Inside BFI4: A Look at Amazon’s 1-Million-Item Fulfillment Center in Kent

Amazon allowed Supply Chain Dive into BFI4, its fulfillment center in Kent, Washington, which first opened in 2016 and is the first facility in the company’s network capable of processing more than one million items per day. The site pairs roughly 3,500 employees with robotics and automated packaging equipment to stow, pick and pack customer orders.

Inbound trucks arrive at the center, where staff unload parcels onto conveyors for stowage into inventory pods that span four floors. Workers at “Universal Stations” place items onto the pods, and instead of associates walking to shelves, Hercules robots bring pods to them. Picking happens at Amazon Robotics Semi-Automated Workstations, or ARSAWs, where projector lights highlight the correct item for each order.

Picked items then move by conveyor to packing areas, with single-item and multi-item orders routed separately. A CW1000 packaging machine measures each product and applies only the wrapping material needed, reducing the excess packaging common to manual boxing. Multi-item orders go to the Amazon Fulfillment Engine area, where associates pack items together with system guidance on box size and automated tape dispensers. Finished parcels travel to the ship dock and then to sorting centers or delivery stations.

What BFI4 Reveals About Amazon’s Automation Playbook

BFI4 is a working example of a “goods-to-person” fulfillment model. The tour shows pods transported by Hercules robots to ARSAWs and Universal Stations, meaning associates do not walk the warehouse floor to locate stock. The visible trade-off is a new set of support roles: Amnesty Floor Monitors are dedicated to clearing obstructions and restarting robots in the fenced robotics area. The verified picture is not a workerless warehouse; it is one where inventory movement is automated while human time shifts toward picking, packing and exception handling.

Advertisement

What the CW1000 Changes in Packing

The CW1000 packaging machine uses sensors to measure each item and size the wrapping accordingly. That directly addresses one of e-commerce’s hidden costs: oversized boxes and excess packing material. Using only the material required cuts packaging spend and waste, and it allows a high-volume facility to keep the packing line moving. The trade-off is the capital cost and maintenance behind such machinery, which the tour does not quantify.

Why BFI4 Matters as a Template

Opened in 2016, BFI4 was Amazon’s first fulfillment center capable of exceeding one million items per day. The company’s network has expanded since then, and the tour makes clear that the automation on display in Kent is not experimental: it is integrated into daily operations across stowage, picking and packing. For competitors and logistics providers, the site is a reference point for what robotics-supported throughput looks like at scale, and for how much human labor still operates alongside it.

What the Tour Does Not Tell Us

The article provides no data on productivity per employee, robot uptime, labor costs or the investment behind the machinery. Those numbers would be needed to judge whether BFI4’s model is more efficient than a traditional warehouse of comparable size. What the tour does establish is the direction of travel: pods moving to people, automated package sizing, and human workers focused on judgment tasks rather than walking.

What Logistics Operators Can Take From Kent’s Robotics Layout

  • Study BFI4’s “goods-to-person” flow before designing or retrofitting a fulfillment center: Hercules robots transport inventory pods spanning four floors to ARSAW stations, eliminating the walk-to-shelf step that typically slows picking.
  • Evaluate automatic packaging equipment on the CW1000 model: sensors measure each item and apply only the wrapping needed, reducing material waste and the labor time spent searching for the right box size.
  • Expect staffing to shift even where automation is deployed: BFI4 runs about 3,500 associates alongside robots, including Amnesty Floor Monitors who clear obstructions and restart machines, so training plans should cover exception handling and robot troubleshooting.
  • Use BFI4’s scale as a planning reference: the Kent facility handles more than one million items a day, a useful benchmark when modeling throughput for highly automated fulfillment sites.

Risk & Opportunity Assessment

Commercial RiskMediumBFI4’s operation depends on heavy robotics investment and about 3,500 associates; downtime in the fenced robotics area would stall picking and packing, and the tour provides no data on robot uptime or maintenance costs.
Competitive RiskMediumAmazon’s one-million-item-per-day capacity and integrated automation raise the throughput bar for third-party logistics providers and competitors still relying on walk-to-shelf picking.
Regulatory RiskLowThe tour raises no regulatory or compliance issues; labor-relations scrutiny is possible but not addressed in the article.
Reputation RiskMediumPublic discussion of warehouse automation often focuses on job replacement; the article’s emphasis on 3,500 associates working alongside robots could ease that concern, but any operational failure at a flagship automated site would attract attention.
Technology DisruptionHighHercules robots, ARSAW stations and CW1000 packaging machines replace the traditional model of employees walking to shelves and manually boxing items, signaling a broad shift toward goods-to-person automation across Amazon’s network.
Commercial OpportunityHighThe Kent site demonstrates how automation can enable more than one million items per day with reduced packaging waste, giving Amazon a scalable template for new fulfillment centers.