Intel failing fast - but in a good way

It’s not every day that a company extols the benefits of failing, but that’s exactly what Justin Rattner – CTO Intel – did last week at the company’s European Research and Innovation conference in Barcelona.

Rattner was talking about the benefits of research, and how it’s not all about success and that failure is just as important. He was alluding to the fact that it’s better to find out that an idea isn’t viable early on, rather than way down the line when you’ve entered the manufacturing phase, or distribution cycle, or worst case scenario, after your product has entered the retail channel.

Rattner called this process failing fast, and it potentially saves Intel a significant amount of money. That’s not to say that research is cheap, far from it in fact, but while failure at the research stage could cost Intel millions, failure at the production stage could run into hundreds of millions. And if failure arrives at the retail stage, you can add the cost of customer confidence too – remember the floating point problem with the first generation Pentium chips?

Of course the ultimate goal is for research to be successful, but even that doesn’t necessarily result in a successful product. Rattner pointed to some of the most iconic research labs of the last century – Bell Labs, Xerox PARC, IBM Labs etc. – and commented that although these research centres came up with no end of innovative ideas, very few of them were turned into actual products, at least not by the companies running the labs.

Rattner said that inventions don’t amount to anything if you can’t make commercially viable products based on those inventions. As such, Intel tries not to separate research from manufacturing, in an effort to ensure that viable ideas can always make the transition into real products.

But even with Intel’s attitude towards research and manufacturing, it’s not always easy to make it to the finish line. Rattner confessed that sometimes when the research is done, and the idea has been proved viable, there’s no product development team available to run with it. In fact he said that virtualisation was just such a concept, but when it became clear how important it was, Intel “doubled down” and made it happen anyway.

There’s no doubt that research is paramount for a company like Intel, and its close ties with the academic community both drives that research forward, and grows new talent. And as long as Intel’s product development teams continue to work hand in hand with the research labs, all that innovation will eventually find its way to consumers.