NVDA: WMT to use GPUs to build massive cloud

More business for NVDA!


Walmart, which is building its own cloud-based data centers, will be using more–a lot more–Nvidia chips going forward, according to a research note from Global Equities Research.

In doing so, the retail giant joins a number of cloud computing providers like Web Services that are using more of what’s known as graphical processing units, or GPUs, to crunch data for machine learning. GPUs, the thinking goes, are better suited for analyzing huge data sets or controlling autonomous cars than more general-purpose chips that are used for more typical computing.

Over the next six months, Walmart will go “full steam” into deep neural networks, using clusters of Nvidia chips, Global Equities analyst Trip Chowdhry says in a note released Tuesday, which Barrons first spotted. Neural networks, a subset of artificial intelligence (AI), are sophisticated computing systems that mimic how the human brain learns.

Chowdhry says that Walmart is building a “GPU farm” that will be about a tenth of the size of rival Amazon Web Services “GPU” cloud.

Walmart is not the only retailer wary of putting its IT budget into AWS services. On Tuesday, the day after Amazon closed its $13.7 billion acquisition of grocery chain Whole Foods, CNBC reported that is scaling back its own use of AWS. What was most surprising to many about that report was the notion that Target used AWS to begin with. But then again, software developers at big companies often AWS to build test versions of their software without their bosses knowing.

A Target spokesman said that company has used and will continue to use multiple cloud providers–which he did not name. Early this year it evaluated that lineup and decided to make some changes, he added, without additional detail…