Page 368 -

TECHNOLOGY HELPS NVIDIA ANTICIPATE THE FUTURE

I n 1999, NVIDIA made history when it invented the graphics processing unit (GPU). Today,
Nvidia’s chips can be found in a broad range of products, including video game consoles,
smartphones, tablets, auto infotainment systems, and supercomputers. Headquartered
in Santa Clara, California, the company has 7,000 employees across 20 countries, and
earned $3.5 billion in revenue in 2011.
Because so many Nvidia chips are made for the consumer electronics industry, one of the
company’s toughest challenges is to accurately forecast customer demand and to adjust its
inventory levels accordingly. Consumer trends can be fickle and subject to sudden shifts one
way or the other. If, for example, the demand for a video game console drops unexpectedly,
Nvidia might be stuck with thousands of excess chips for those systems, which represents a
significant loss for the company.
Nvidia's chips are created long before they are sold to customers, requiring production
planners to make estimates of how much material the company will need and how much
production time to schedule at Nvidia’s foundries, which are located primarily in Asia. When
Nvidia’s customers estimated how many Nvidia chips they would need, Nvidia’s planners made
their own independent estimates. Using these estimates, Nvidia would buy enough material
(primarily silicon wafers) in advance and schedule enough capacity at the company’s found-
ries (which are primarily in Asia) to meet what it thought would be the right level of demand.
Business units would meet with Nvidia’s finance unit to discuss the number of chips to be
produced, based on high-level estimates. Nvidia’s chip operations group, which was respon-
sible for the actual production, never received the forecasts and could only see existing inven-
tory. Nvidia’s production department used spreadsheets to create rough inventory forecasts,
but those spreadsheets did not allow planners to drill down, sort data by product, compare
different types of inventory, or view data by business segment, and the data for these spread-
sheets had to be gathered from a number of systems.
Management received a wake-up call when Nvidia switched its old manufacturing process
to a 40 nanometer process. The company was forced to carry inventory created by the old
manufacturing process as well as for customers who were not ready to change. Management
discovered that the current system lacked the ability to handle the complexity of two separate
sets of inventory and was unable to balance supply and demand for its new products and its
existing products, as well as predict how long it would take for its customers to transition to the
40 nanometer method. Nvidia
wound up with way too much
inventory, and when it started
cutting back, its suppliers were
caught off guard.
To address these problems,
Nvidia set up a supply chain
steering committee to review
its supply chain processes. The
steering committee recom-
mended that Nvidia replace its
spreadsheet-based inventory
forecasting system with some-
thing more current. SAP soft-
ware proved to be the logical
choice. Most of Nvidia's data
were already located within
© JohnKwan / Shutterstock
367

MIS_13_Ch_09 Global.indd 367 1/17/2013 2:28:52 PM