Amazon’s Robot-Making Shift: The Build-or-Buy Question

A procurement decision becomes a strategic question

Amazon’s shift from purchasing robots towards making them, highlighted in the original report from automate.org, raises a question that reaches beyond robotics procurement: when should a business own the technology behind its operations?

The headline establishes that change in direction, but does not explain its scope, economics or results. It would therefore be premature to call it a blueprint for other organisations. The useful management question is not whether to imitate Amazon, but how to decide which parts of automation deserve investment, control and internal expertise.

The robots are coming. For business owners and operators, the practical issue is where they can remove cost first — and whether building, buying or commissioning a solution is the most credible route.

Control has a price

Developing robots internally could give a business greater influence over design, integration and future changes. Where a task is unusually specific, that control might be valuable. It could allow equipment to be shaped around an operational requirement rather than requiring the operation to accommodate an available product.

But control also brings responsibility. A build decision should account for engineering, testing, maintenance, documentation and support throughout the system’s working life. Ownership is not simply a different way to acquire hardware; it means accepting responsibility for keeping a capability useful.

Buying should face equally disciplined scrutiny. Decision-makers should examine integration requirements, servicing arrangements, supplier dependence and the cost of adapting equipment when workflows change. Neither route should win on purchase price alone.

Find the economic problem before the robot

For an operator considering automation, the starting point should be a clearly defined task and a measurable business problem. Is the objective to reduce repetitive handling, improve consistency, relieve a bottleneck or limit exposure to a hazardous activity?

Each objective needs its own baseline. Record the labour time, delays, rework and support effort associated with the current process. Then compare possible changes against the same measures, including a simpler process redesign rather than automation.

Labour capacity released is not automatically cash saved. A proposal should explain whether time would be redeployed, overtime reduced or future recruitment avoided. Without that distinction, a productivity improvement can become an overstated financial case.

Make staffing and safety part of the design

A credible plan should specify who will operate the equipment, resolve exceptions and take responsibility when it stops. Training and maintenance capacity belong in the investment appraisal, not in a later implementation discussion.

Safety also needs explicit assessment. Removing someone from one task should not be assumed to make the whole operation safer. Installation, cleaning, recovery and servicing must be considered alongside normal operation. Staff input can help identify practical requirements that a procurement specification might otherwise miss.

Choose the smallest defensible next step

Amazon’s change invites scrutiny of the build-or-buy decision, not a universal preference for either option. Business leaders should choose one bounded process, establish its current cost and set clear conditions for a trial.

The robots are coming, but competitive discipline means resisting technology-led spending. The first investment should address a demonstrable operating problem, with accountable ownership and a credible path from technical performance to financial value.