Intel’s Bet: All in on Process Technology

It is hard to watch everything going on at Mobile World Congress and not sense the missing force in computing that is Intel. The charts I posted yesterday showing the growing mobile landscape and the stagnant PC one is causing a great deal of pain for Intel. While their server business is strong and yet to be threatened by any serious ARM competitor, and while their modem business looks promising for 5G, we saw many other competitors show off new processors and new design wins with their mobile processors. If this were a PC trade show, Intel would be all over it. At a mobile trade show, their impact is limited.

That being said, I’m quite intrigued by Intel and reading between the lines I’d like to explain what I believe their bet on the future is and why I’m not writing them off.

To really understand Intel, we need to separate their x86 chipset architecture from their manufacturing capabilities. Intel may be tempted to tell you these two things are related but they are not. As Intel pursues Moore’s Law, they push the size and shape (geometry) of the silicon so you can get more out of each advancement in process technology. When we refer to chipsets in nanometers we are talking about their geometry. Going off memory, in 2005 we were at 90nm. Today, Intel is shipping at 14nm, as is Samsung. That is the most advanced process technology mass producing chipsets today and, at 14nm, you can fit 5,000 transistors on the width of a human hair.

I believe Intel’s bet is on process technology, not their x86 architecture. The reality is, it is getting harder and harder to progress process technology to the smaller size. 10nm is next and Intel and Samsung will get there but it is getting harder and more expensive. TSMC, who manufactures for Apple, may get there but it will only be because of Apple that they do. I’ll explain that another time.

Plain and simple, I believe Intel’s bet is they are the last man standing at the end of the day in process technology leadership. After 10nm comes 7nm after 7nm comes 5nm. There is huge debate as to whether the industry gets to 7nm and there is even more doubt we get to 5nm. Intel’s bet is they will get to 7nm or to 5nm, or even beyond, and no one else does. Which means they end up making the chips for everyone who needs them and that is a huge business.

Should this happen, Intel will have to make ARM chips for Qualcomm, Mediatek, TI, Nvidia, Apple, etc. One could argue they should make this transition already but, for whatever the reason, they are staying the course. Is this plausible? It actually is. Intel’s engineers have proven they could beat the odds before to keep Moore’s Law going. Which is why I’m not writing them off.

At the same time, if I was running Intel, I’d throw down the gauntlet to my engineers. Intel believes they have some of the best chip designers in the world. Right now, those designers are architecting x86 designs which largely go into PCs and servers. If I ran Intel, I’d assemble a team and tell them to show me if they could develop some of the best ARM designs in the world. If they could, I’d use this as the foundation for their future.

If their ARM designs were competitive and they could bundle them onto the SoC with their modem, then all of a sudden they have a competitive product to Qualcomm. Thanks to their ability to manufacture, they can actually sell at an aggressive price, perhaps even undercut Qualcomm and start getting customers for the Intel ARM architecture suite of products. Intel is unique in this case because they design and manufacture and, as long as Intel does have some of the best chipset designers in the world and they create a competitive ARM product on their leading edge process technology which no one else will have, then this thesis could work out.

There are a lot of assumptions I make in this, as does Intel. It all feels very risky but Intel has a great deal of confidence in their engineers and they have a proven track record of overcoming obstacles that stand in the way of Moore’s Law.

Published by

Ben Bajarin

Ben Bajarin is a Principal Analyst and the head of primary research at Creative Strategies, Inc - An industry analysis, market intelligence and research firm located in Silicon Valley. His primary focus is consumer technology and market trend research and he is responsible for studying over 30 countries. Full Bio

3 thoughts on “Intel’s Bet: All in on Process Technology”

  1. What’s kind of sad is that Intel could have had a big jump on their ARM competitors if they had wanted to continue development of the StrongARM line that they inherited from DEC. In the days of the Newton, it was by far the most power efficient (mips/watt) CPU of its day.

    1. Agree. They bet on x86 and went all in on the architecture and there is still sentiment around that today. But I’m more optimistic on process than their architecture.

Leave a Reply

Your email address will not be published. Required fields are marked *