Real Men Have CAD

“Real men have fabs.” That famous quote from the early 1990s, is usually attributed to Jerry Sanders, Chairman of Advanced Micro Devices, and was used to underscore the belief that owning a semiconductor manufacturing facility, or fab, was the only way a semiconductor company could be successful.

Many innovative chip companies saw things differently, however, and by the mid-1990s there were many successful “fabless” semiconductor companies. These pioneers realized that the $1-2B it cost for the construction of a new fab every couple of years was not only prohibitive from a cost perspective, but also not necessary at all.

These chip makers turned to TSMC, UMC, Seiko-Epson and even IBM for the manufacturing of the wafers and chips, while focusing their efforts on design innovation, testing and marketing.

By the mid-1990s, many of these fabless chip makers were among the most profitable companies in the semiconductor industry.

Fast forward to today.

The continuing economic downturn has forced all companies to take a hard look at cutting, or at least managing, costs. The semiconductor industry is no different from other markets, and every chip maker is looking for a way to turn fixed costs into variable costs.

In his post on Xuropa last year James Colgan said that one way semiconductor companies plan to turn fixed costs into variable costs is by moving parts of the EDA (electronic design automation) flow to a SaaS model, where components of the semiconductor design software reside in the cloud.

Clearly there is a trend in the semiconductor industry to take advantage of cloud computing as a way to reduce costs. The chart below shows the adoption curve for many of the EDA modules migrating to the cloud.

This will have ramifications for how EDA companies such as Cadence, Synopsys and Mentor Graphics market their products to customers. No longer will customers want to purchase costly software design tools, or even licenses to those tools, if they can use them via the cloud on a pay-per-use model. This trend is just now in its infancy, as is the way EDA companies talk about their relationship to cloud computing.

But as rapid as other software applications have moved the cloud, we're probably not too far away from hearing some stubborn semiconductor executive stand up and proclaim that "real men have CAD."

Cloud Chips

It seems as if every technology company in the industry has announced a cloud computing strategy. IBM and HP. Oracle and Microsoft. CA and Saleforce.com. Amazon.com and Google. All major software and computing companies are moving to market with cloud computing solutions. It’s no wonder that a report from McKinsey & Company claimed that there are 22 different definitions of cloud computing in the marketplace.

So I guess I shouldn’t have been too surprised when I spoke to a friend of mine at Nvidia Inc. recently and he told me his company was “bullish” on cloud computing. For those of you unfamiliar with Nvidia, they are, in their own words, “the world leader in visual computing technologies and the inventor of the GPU, a high-performance processor that generates breathtaking, interactive graphics on workstations, personal computers, game consoles, and mobile devices.”

In short, Nvidia makes really powerful semiconductors for graphic applications. But why would a chip company be so bullish about cloud computing?

Let’s start with one of those 22 definitions of cloud computing. At it’s simplest, it is a way of computing, via the Internet, which broadly shares computer resources instead of using software or storage on a local PC.

Now some of those “computer resources” that live in the cloud are graphically intensive, 3D-modeling software applications that are used for everything from designing jet aircraft to modeling photo-realistic images of a patient’s heart for a doctor to view before surgery.

If you’re an aeronautical engineer working for Boeing, for instance, you might use a cloud computing application to run some 3D models for wind resistance over one of the plane’s wings. That type of graphically intensive application isn’t going to run on just any old computer server in the cloud. No, complex 3D modeling requires a server that is run by powerful GPUs, the type made by Nvidia.

Make no mistake, Nvidia and other semiconductor companies aren’t going to start building server farms and begin hosting cloud-computing applications. But companies such as IBM, HP, and others do. And those companies either buy or build specialized servers to run these graphically intensive applications.

In turn, the firms providing cloud-computing resources need to offer a variety of applications to meet the widely diverse needs of their customers. As a result, the demand for servers that can handle the enormous processing load placed on them by complex 3D graphic software is going to increase.

This interconnected food chain of cloud computing is creating opportunities for technology companies across the spectrum – from enterprise software to semiconductors. That’s why Nvidia, and any other forward-looking chip company, is bullish on the cloud computing opportunity.