What we’re watching next in humanoids

FedEx just outsourced its warehouse future to Berkshire Gray.

FedEx announced on March 31, 2026, that it will lean into partnerships rather than build proprietary automation tech in-house, naming Berkshire Gray as a key collaborator to accelerate its logistics automation across the parcel network. The move signals a pragmatic pivot: ship more capability faster by stitching together proven components from experienced operators rather than betting everything on a single internal stack. In plain terms, FedEx is betting that a vendor-enabled automation ecosystem will outpace a first-principles build in both speed and reliability.

What’s on the table, technically, is non-humanoid automation: autonomous mobile robots, sortation modules, and integration layers that tie robotic fleets to FedEx’s warehouse management and operations planning tools. Berkshire Gray is positioned to supply end-to-end components—robotic fleets, software for routing and task assignment, and the necessary integration heft to keep conveyors and docks synchronized with shipment data. The partnership aligns with a broader industry pattern: large shippers opting for scalable, vendor-backed solutions that can be rolled out facility-by-facility rather than rearchitecting IT and material handling from the ground up.

From a technology-readiness perspective, this looks like field deployments in real facilities rather than purely lab demonstrations. FedEx has long traded long and costly internal development cycles for proven external platforms that can plug into existing warehouses, and Berkshire Gray’s offerings are designed to operate in live environments with real-time payloads, variable item sizes, and fluctuating parcel volumes. The practical takeaway for R&D teams watching is that the interface layer—the software that harmonizes WMS, yard management, and dock scheduling with robotic tasks—will be the determinative constraint, not the performance of the robots themselves.

Two honest limitations to flag, grounded in the domain’s realities: first, integration complexity. Enterprise warehouses are stitched together from decades of bespoke equipment, custom conveyors, and legacy software. Even with a well-vetted partner, aligning Berkshire Gray’s workflows with FedEx’s order profiles, exception handling, and real-time visibility dashboards requires careful data modeling and change management. Second, vendor reliance. Pushing automation maturity through an external partner accelerates deployment but can increase dependency on the partner’s software update cadence, service levels, and engineering staffing. In other words: it buys speed, but it also concentrates risk around a single ecosystem.

Compared with FedEx’s prior posture—where some automation bets resided in-house or via smaller pilots—this arrangement should improve deployment cadence and scale across the network. In practice, the improvement is not just hardware; it’s the ability to replicate a proven operating model across dozens of facilities with standardized KPIs for throughput, accuracy, and uptime. The tradeoff is a dependency on Berkshire Gray’s roadmap and service model for ongoing optimization and fault resolution, rather than rapid, bespoke, self-contained iterations.

Power, runtime, and charging specifics remain under wraps in the public briefing. Typical enterprise AMRs and sortation modules operate on electric power with lithium-ion or solid-state chemistries and docking stations for overnight or shift-long charging; however, FedEx’s exact runtime targets and charging architectures for this program aren’t disclosed. What matters for operators is predictable uptime and predictable maintenance windows, not a single “hero” spec.

What we’re watching next in humanoids

How FedEx’s WMS and yard controls will ingest Berkshire Gray tasking without latency or data skew.

The scalability signal: whether the rollout maintains the same per-site throughput gains as facilities expand.

The contract and service structure: cadence of software updates, remote diagnostics, and on-site maintenance commitments.

Any movement toward future humanoid-assisted tasks in loading/unloading or inspection workflows, and how those would be evaluated alongside non-humanoid automation.

Real-world reliability metrics (throughput, pick accuracy, downtime) over the next several quarters.

Sources

FedEx chooses partnerships over proprietary tech for its automation strategy