State of the Roll Your Own Semis Market

By our count there are now 28 non-semi conductor companies developing their own chips, and we know we are missing a few names. This is clearly a big market and it is growing rapidly. We plan to explore the space more deeply over a few upcoming posts. Here, we want to lay out the landscape, and then we will look at some of its intricacies and then map out where we think things are going.

Here is the list we have so far:

Alibaba
Amazon
Apple
Baidu
Byte Dance
China Mobile
Cisco
Facebook
Ford
Fujitsu
Google
Hitachi
John Deere
Korea Telecom
Microsoft
NTT DoCoMo
Oppo
Samsung
Seagate
SK Telecom
Tencent
Tesla
TP Link
Vivo
Western Digital
Xiaomi

These are all companies who have either publicly announced chips or have had credible press reports claiming they are. Notably absent from this list are a lot of Chinese companies. Huawei should probably be on this list, but it is very hard to tell what, if anything, they are still building – probably something, but a far cry from what they were designing before they ran afoul of the US Government. Automotive is another tricky category. Ford is on this list, but General Motors is not, they seem to still be working with third parties (large and start-up), but do not seem to have the internal design team that Ford does. As far as we know, John Deere has not exactly said publicly they are designing their own chips, but they acquired an AI company for $300 million early this year, and we have heard from multiple sources they are rolling their own. And once again, many Chinese auto companies are reported to be building their own chips as well, but the data is sparse. Next update, we will have more of them on the list. And while we are on the subject, we should probably add Foxconn to the list, we are overdue to write a post on all their semis activity, but those are growing so fast it is hard to track. So we know there are many more names for this list, please feel free to e-mail us your additions.

Of course, the growing question is how much of this is sustainable. Ask any of these companies and they will say they are in it for the long-haul, but ask any semiconductor executive and they will say the opposite and likely add an anecdote about some employee who left for one of those companies only to come back a year later.

One way to judge these companies’ commitments is to count the number of different chips they have developed. By that count, these 28 companies have 43 chips under development. However, this is a bit misleading. Apple and Google both have a half dozen chips announced, and only about a dozen of these companies can be connected to more than a single type of chip.

Another good metric is to look at how many of these companies have produced more than one generation of a chip. As anyone who buys chips knows, a roadmap is an important indicator. Alibaba for instance has announced four or five chips, but most of them never seem to have entered production, let alone multiple generations. By contrast, Apple is into double digit generations for some of its chips. So we looked at which companies have developed multiple generations of their chips and found 17 that definitely have gone through multiple generations, well over half the list.

For us the stand-out feature of this list, and the biggest change from the last time we compiled the data, is the growth beyond the Internet companies. The automotive sector is the obvious one, but there are likely many other industrial companies working on this as well. The trend is breaking out of “technology” companies, which is not that surprising given what we are seeing across the industry. In our next post, we will look at why this may not be a path everyone can take, but we are still going to see a lot more companies make the attempt.

Leave a Reply