Are We Reaching the Limits of Homegrown Silicon?

Last week, the big news in China semiconductors was Oppo’s announcement that it was largely disbanding its Zeku chip division. This story did not get much attention in the US, but we think it is significant and worth mentioning in the context of the broader trend of non-chip companies designing their own chips. We have been writing about this topic for a long time as more and more companies tried their hand at this. Does Oppos’ move signify the crest of that tide? Are we going to see more companies walk away from these projects?

First, some background. For a whole bunch of reasons, over the past decade the economics of chip development have allowed large chip customers to enter the market of designing their own chips. Apple’s M-Series CPU and Google’s TPU AI accelerator are the best known examples of this, but there are many more. This extends beyond the biggest tech companies to John Deere and Wi Fi Access Point maker TP Link. Our thesis is that companies only want to do this when the chip conveys some form of strategic advantage, desiging a chip just to save a little bit of money is not a breakeven proposition. So Google saves billions of dollars in capex because of its chips, and Apple’s chip drive better end-device performance resulting in tens of billions of incremental sales.

A list of companies designing their own chips

AmazonNTT DoCoMo
Byte DanceSeagate
China MobileSK Telecom
FordTP Link
GoogleWestern Digital
John DeereZTE
Korea Telecom
Source: Digits to Dollars

All of which brings us to Oppo. Oppo is in the very difficult position of competing against Apple. As noted, Apple has their own Applications Processor (AP) for their phones, which convey a pretty significant competitive advantage. An advantage which comes at the expense of Oppo and its peers. So the rationale for designing their own chip is sound, but something went wrong along the way.

According to reports in the Chinese press, Oppo found that it was not getting the results it had hoped for, despite incurring a huge expense in the attempt. From what we can tell, Oppo was not designing a complete AP, instead focusing on some sub-system, likely around imaging, which they felt was particularly important. Despite spending all that money on their chip, they saw essentially no response in the market, and in fact lost further share to Apple and its domestic competitors. We suspect that they have not given up entirely, but have definitely scaled back their ambitions. Moreover, Oppo has an insurance policy built into their plans. Oppo is owned by the BBK Group which also owns Vivo, RealMe and a handful of other brands. While these companies ostensibly operate independently, on matters of engineering the various group entities do share engineering resources. We can only guess what the internal politics in the group are like, but we suspect if Vivo successfully designs their own chip, Oppo could probably source it from them at some cost.

Putting this all together, we think Oppo’s situation points to a fairly specific set of circumstances that do not necessarily mean others are going to give up rolling their own chips. That being said, we do think Oppo’s case points to the limits of this model. Designing a chip is still expensive and there are probably only a few hundred companies in the world that can afford to do it. Moreover, Oppo shows the critical importance of software in all of this. To put it simply, Apple controls its software, Oppo does not, it runs Android. As a result, many of the gains Apple enjoys from its A series chips are not available to Oppo. This, probably more than anything else is what stood in Oppo’s way. This is going to be a problem for other handset makers as well, but is less relevant to many of the other companies on that list above.

So while we think there are limits to who will design their own chips, there is likely still room for others to enter the mix.

One response to “Are We Reaching the Limits of Homegrown Silicon?

  1. Pingback: 我们是否正在接近自主生产硅的极限? - 偏执的码农·

Leave a Reply