How we managed our global supply chain through a year of lockdown

A Year of Remote Supply Chain Management

When Dropbox sent employees home in March 2020, the Infrastructure team had already been watching COVID-19's spread in China and its potential effect on datacenter hardware suppliers. A year later, after building new datacenters, managing surging demand, and maintaining uptime SLAs—all without in-person meetings—the team has refined its approach to vendor relationships and capacity planning in ways that are likely to outlast the pandemic.

Rebuilding Supplier Relationships Without In-Person Contact

Dropbox's supply chain team had historically relied on regular visits to supplier headquarters and manufacturing sites in Taiwan, Korea, and Thailand, plus frequent local visits to rack manufacturing lines. The shift to remote work eliminated those touchpoints overnight, forcing the team to reinvent how it stayed connected with its supply base.

Virtual manufacturing line walks and quality inspections became the new standard. The early attempts were inefficient, requiring significant preparation with suppliers to get camera angles right and keep reviews focused. Suppliers, whose expertise lies in manufacturing rather than video production, needed time to learn how to effectively show hardware remotely. After refining the process together, the team completed numerous successful remote walk-throughs.

The key insight was that high-frequency, focused virtual connections worked better than less frequent in-person meetings had. Dropbox's smaller scale relative to other cloud customers makes supplier relationships particularly critical, and maintaining clear communication on hardware needs through remote channels has been essential.

Scaling Capacity for Remote Work Demand

As it became clear that remote work would persist, the capacity planning team began forecasting increased usage of Dropbox services. A global scenario planning process examined the potential impact of factory shutdowns, reduced workforces, and extended lead times across the supply chain.

The team's playbook centered on identifying and ranking levers to meet availability goals while balancing CapEx, OpEx, and engineering effort. They updated inventory policies and planning parameters to ensure critical infrastructure services like Magic Pocket and Edgestore maintained sufficient free capacity. Allocation strategies for new hardware deliveries were determined proactively with application owners, using approaches such as strategic distribution, fair share, and FIFO prioritization.

For short-term capacity risks, the team identified services that could migrate to public cloud infrastructure quickly and cost-efficiently. This helped stabilize forecasts to suppliers and avoid the bullwhip effect, where small disruptions cascade through the entire supply chain. These measures kept capacity SLA/SLO impacts minimal and ensured no customer-facing service disruptions.

Keeping Datacenters Running With Skeleton Crews

The datacenter team faced three priorities: maintaining physical infrastructure uptime SLAs, keeping essential onsite staff safe, and continuing planned datacenter expansion timelines. The first two were non-negotiable, given that users increasingly depended on Dropbox services from home and that employee health was paramount.

A rotating skeleton crew model with flexible hours proved as effective as the typical full-staffed 9-5 schedule. This model was adopted across all 10 datacenter sites. The team tracked government pandemic-safety ordinances across five counties in four states and worked closely with facility managers to align site policies with company values. Repair SLAs remained constant year over year, and no workplace COVID transmission was reported.

Building Datacenters Without Setting Foot On Site

Constructing two new datacenter sites during the pandemic presented an unprecedented challenge: managing construction completely remotely. Without a quality assurance team performing weekly onsite job walks, the team shifted strategy to require construction crews to upload daily or weekly photos to shared folders. This asynchronous visual communication, combined with virtual walk-throughs similar to those used with suppliers, allowed early identification of issues before they could affect schedules.

The approach worked. The projects met their deadlines with 100% virtual management, without compromising design or quality.

Adapting to a New Operating Reality

The past year reinforced the need for flexibility with partners while still delivering to customers. That meant changing not just where work happens but how it gets done—sometimes tracing disruptions deeper into the supply chain, sometimes rethinking meeting conventions, and often applying remote work practices from software development to hardware manufacturing and datacenter construction crews.

The infrastructure team's experience demonstrates that supply chain resilience depends as much on communication methods and relationship management as on inventory buffers and contingency plans. As the company moves toward its Virtual First model permanently, those practices are now baked into how the team operates.