If Designing Semiconductors Were Easy, Everyone Would Do It

Some news came out last week that Google may be working on making their own processors. Reports like this have been around since 2013. For those of us close to the semiconductor industry, this is not a new piece of information. Speculation was that Google may be working on their own silicon designs for smartphones. However, the industry scuttlebutt I’ve been hearing for a few years was Google was exploring designing semiconductors for their own servers. I’ve heard the same about Facebook for what it’s worth.

On paper, this seems somewhat logical. Backend server side computing is becoming increasingly important to the world of computers today and will to an even greater degree in the future. All one needs to do is look at the advantage Apple is creating for itself in designing its own SoC for the iPhone and the iPad to understand the temptation to design silicon for proprietary advantage and, in this case, for Google on the server side of the equation. Google has ambitions to be one of the most powerful cloud computing companies in the world and designing their own silicon makes sense.

The challenge, however, in designing semiconductors is not easy and most companies lack the internal resources and talent to execute. Here is my short list of companies who have tried and true designers and core expertise to create proprietary semiconductor architectures. Intel, Qualcomm, Apple, AMD, Nvidia, and Broadcom. Intel and AMD largely make x86 cores and the others make ARM cores. While there are certainly many names who produce and ship ARM cores, there are only a handful of companies who truly design their own using the ARM instruction set. The names I mentioned are among that list and are the best at it, in my opinion.

When acquiring an ARM architectural license, the kind you need to design your own chips, it is not mandatory to announce to the public a license has been acquired. Google could certainly have one and we would not know about it – yet.

Google may certainly have an ARM core license or an ARM architectural license, but what they do with it is the main question. If designing semiconductors were easy everyone would do it. However, not everyone can.

I’ve long maintained that those who design silicon, and design it well, will play a major role in the future of computing. Thinking about Google following the trend of companies designing their own semiconductors in the hopes it will bring them a proprietary advantage is an interesting thesis. But Google, and many others who do this, would need to acquire the necessary talent to design best-in-class semiconductors. With the consolidation we are seeing, it makes sense Google or someone else could acquire a company like AMD and gain the chip design talent needed. Microsoft also has an architectural license for ARM and I could make a case Microsoft could start making their own chips as well at some time in the future. The main point is, it will have to be acquired. As I think about any number of these companies like Microsoft, Google, Facebook, etc., if we seen an acquisition of semiconductor IP and talent, then it will be quite clear where they are headed and why.

The other area to watch here is China. Last year, the government of China committed to spending 100 billion dollars in 2015, advancing their efforts in proprietary chipset architectures. Huawei, which owns Hi-Silicon, has an ARM architectural license and does make a great deal of its own silicon. They are one of many companies the government would like to see designing silicon for things like PCs, smartphones, tablets, servers, and network equipment both for China and the rest of the world.

Watch what happens here. Trends in the semiconductor space are quite telling directionally about where the industry and particular players are headed.

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

10 thoughts on “If Designing Semiconductors Were Easy, Everyone Would Do It”

  1. So the moral of the story is, “If you want to design semiconductors, shop well”.
    That is certainly the fastest, and possibly cheapest route to market. There is a flip side.

    There are indeed those versed in the skill of semiconductor design, which is hard. There are also those versed in semiconductor manufacture, which is also very hard. There is an upside to this, the first chip costs a billion dollars, all the rest cost a quarter each.

    Since very few OEMs would have incentive to manufacture their own chips, since ostensibly they won’t be supplying others, a middle ground can be: hire the talent in chip design and buy a plant. Apple may have the volumes, they certainly have the money, to fund their own plant. At their margins, if they wanted a PR coup, they can even afford to do it in the US.

    Google can do the same, and presumably supply Android OEMs.

    1. Good additional thoughts to mine. Also, places with plants (cough Intel) can look to help use their design prowess and their manufacturing capabilities to do custom design. They are actually trying this and very few are biting, but that is because they don’t seem to be so serious about it. AMD’s semi-custom business is the same as they did that for Microsoft and Sony’s game consoles.

      The plan option is valid also since there will be many plants for sale at some point as well.

  2. @benbajarin:disqus Actually reading the details from arstechnica[1] it seems Google might have a reasonable plan – co-develop the processor , while optimizing towards and contributing IP towards their unique domain knowledge in virtual and augmented reality.

    Another point – VR/AR could be a big platform shift, and those tend to stir industries around, so who knows how everybody will come out of the other side ?


  3. Superb post however , I was wanting to know if
    you could write a litte more on this subject?
    I’d be very grateful if you could elaborate a little bit more.
    Bless you!

  4. Hello, Neat post. There is an issue together with your web
    site in web explorer, would test this? IE still is the marketplace leader and a
    big component to people will omit your magnificent writing because of
    this problem.

  5. Terrific article! This is the type of information that should be shared
    around the internet. Shame on Google for no longer positioning this publish upper!
    Come on over and consult with my web site . Thanks =)

Leave a Reply

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