Microsoft let planners interrogate the optimizer that ships cloud servers

Microsoft's cloud supply chain had to place servers into more than 400 data centers across over 70 regions while AI demand was rising. Before 2020, much of the fulfillment process still depended on manual, spreadsheet-driven planning.

The clever move was not only to build a mixed-integer optimizer. Microsoft wrapped the optimizer in the Intelligent Fulfillment Service and added an LLM assistant, OptiGuide, so planners could explore scenarios and ask why a plan made each tradeoff.

INFORMS said IFS automated global shipment planning, cut fulfillment cycle times by more than half, reduced team workload by 23%, and generated tens to hundreds of millions of dollars in annual savings.

Make the solver explain itself

It turned tacit planner rules into explicit constraints the system could optimize.

The LLM interface solved the trust problem that often blocks operations research systems.

Scenario questions that took days of planner-engineer back-and-forth became minutes of guided reoptimization.

Optimization spreads faster when the people who must trust it can question it in their own language.

Microsoft won the 2026 INFORMS Franz Edelman Award for the work. The same program became part of Microsoft's broader INFORMS Prize recognition for embedding operations research across cloud infrastructure, from virtual-machine allocation to rack placement.