EN
Back to the archive

The encyclopedia · Software & IT · Technical decision · 2009–2020

Intel coupled chip design to supply-chain planning in one analytics stack

Intel optimized product architectures and supply chains as one problem, lifting annual revenue $1.9B and cutting annual costs $1.5B.

Intel

the move

In a capital-intensive business like semiconductors, a product architecture chosen badly is expensive to undo, and a supply chain planned badly wastes capacity that took years and billions to build.

Intel's answer was a single analytics capability spanning product architecture design through supply-chain planning, with the twin goals of maximizing revenue and minimizing cost. Product design alternatives are generated and scored with genetic algorithms and physics simulation; the winners are then solved into a large-scale production and distribution plan using mixed-integer programming.

Because the two halves share a model, Intel can examine far more scenarios in less time and still respond faster to customers. Over most of the product portfolio the system raised average annual revenue by $1.9 billion and cut annual costs by $1.5 billion, for a total benefit of $25.4 billion since 2009.

why it works

  • Tying design to capacity means a chip is only signed off if the plants can profitably build it.
  • Problem decomposition makes a huge optimization tractable, so the model runs fast enough for real decisions.
  • Genetic search and physics simulation let planners explore many architectures that people would never think to list.
  • More scenarios studied means smarter choices about which products to make and where to make them.
the payoffOptimize architecture and output as one probleminspired

what transfers

When design and production each look great alone, linking them beats improving either side by itself.

what came after

The capability became a corporate-wide system used across Intel's product portfolio and supply chain, and it won the 2020 INFORMS Franz Edelman Award. Intel reported the work also supported its sustainability efforts by making plants and logistics run more efficiently.

references

spotted an error? The archive wants to know.

same kind of clever