Earlier this week Google announced its new Pixel phones, and as part of that they also unveiled their own application processor (AP) chip for the phone, which they call Tensor. Here is a link to the Twitter thread introducing Tensor. At this stage, we know very little about the chip. A few details surfaced in the press, but the whole thing leaves us with more questions than answers.
We just published a look at the AP industry yesterday, so to keep it simple – the AP is the chip that makes a smartphone “smart”, it runs the operating system and all the apps and other software on the phone. This is a crucial part of the device, and one of the only ways left for smartphone makers to differentiate their devices. Apple appears to be crushing the market on the back of its industry-leading A Series APs, and many others are trying to build their own solution. But why is Google doing this?
Google’s Pixel has always felt like an aferthought. It is a way for them to show off the full capabilities of Android, a benchmark against which we can measure other Android phones, but it has never been a truly commercial endeavor. IDC reported that they sold 7 million Pixels in 2019, and 2020’s model was probably lower. This is rounding error in the 2 billion units a year mobile market.
So maybe Google is going to get serious about promoting Pixel this year. This strikes us as unlikely as that would put them in competition with all their partners, the smartphone makers, who drive a lot of traffic to Google sites (which is the whole point of Android after all). It seems unlikely that Google wants to anger companies like this while it is in the middle of multiple anti-trust fights all looking at their practices around Android.
Maybe Google will instead sell Tensor to other handset makers. This could be huge. The other vendors badly need a good AP, especially one that can handle all the AI goodness that a phone called Tensor from an AI-first company like Google can offer. If true, we would be very wary. Google does not have a great track record when it comes to sustaining products. To make this work, any handset maker who wants to use this for their phones would have to believe that Google has a multi-year roadmap, one which they will truly commit to delivering. Running a chip business is a lot of tedious work, little of which can be entirely automated. Serving these customers would require a lot of work that is very far from Google’s core competencies.
On Twitter someone pointed out to us that the Tensor is most likely to be used in Google Chromebooks, which sell in much larger quantities than Pixel phones. This makes sense, a phone’s AP should be powerful enough to drive a Chromebook. On the other hand, the same logic as above for phones applies. Google only sells a small portion of Chromebooks itself, the others are sold via the brands that design them like HP and Samsung. Those companies may want find that using a Tensor chip has advantages, but they are going to weigh all the standard commercial considerations before going down that path, and high up on that list of questions has to be the longevity of the Tensor product line.
Which leave us with why we really think Google built Tensor – because they could. Google has a pattern here. They have incredible engineers with time on their hands. This is not the first, not even the 100th time they have rolled something out with incredible features that had no clear business case. Google has built a very strong chip design team and they have ironed out the production back-end to facilitate this. This means the cost for them to design a new chip is very low. We wrote about how this can go well and how they likely have some important tools in place to build more chips. So we should not be surprised that they are deploying these now in mobile.
This does not mean a revolution is coming to smartphones. As interesting as Tensor may be (and let’s not forget we have no idea how well it performs), as capable as Google’s engineers are, we think it is unlikely that Tensor is going to shake up the industry. We think you cannot rely on Google to stay in this business.
Photo by Matt Walsh on Unsplash
God save us from Qualcomm! They produce 400+ zero day or other bugs in their modem processors straight off the line, enabling Pegasus type shenanigans. Hateful! Scam! No standards for big corps.
Hmmm, I believe DD Advisors has been covering Qualcomm for decades, yet you still don’t appear to fully understand Qualcomm’s business model. Qualcomm’s royalty is a low single digit percentage of the wholesale price of the device capped at $400 and not the total sales price of the phone as your article implies.
“…recall that Qualcomm’s much-contested, highly-profitable licensing business charges royalties based on the price of phones sold….”
There’s a lot more nuance to the licensing terms than that. And what happens when the ASP of licensed devices falls below $400? The point os that falling prices for Android phones hurts both QCT and QTL.
Falling phone prices can hurt Qualcomm , but leaving out the context of the $400 price cap misleads most of your readership as flagship/ premium devices priced upwards of $1500 or more would have to fall significantly (75%) before “hurting Qualcomm”.
Note, this comment was in reply to your referenced article- 2021 STATE OF THE MOBILE BASEBAND
Further, I find it interesting that in the above referenced article you fail to discuss Qualcomm’s recent acquisition of Nuvia when you write “……but right now it (Qualcomm) seems to be an engine with the gears not fully engaged.
From the Qualcomm PR- Qualcomm to Acquire NUVIA | Qualcomm
“…The acquisition of NUVIA builds on Qualcomm Technologies’ Snapdragon technology leadership, delivering step-function improvements in CPU performance and power efficiency to meet the demands of next-generation 5G computing.
NUVIA comprises a proven world-class CPU and technology design team, with industry-leading expertise in high performance processors, Systems on a Chip (SoC) and power management for compute-intensive devices and applications. The addition of NUVIA CPUs to Qualcomm Technologies’ already leading mobile graphics processing unit (GPU), AI engine, DSP and dedicated multimedia accelerators will further extend the leadership of Qualcomm Snapdragon platforms, and positions Snapdragon as the preferred platform for the future of connected computing.
NUVIA CPUs are expected to be integrated across Qualcomm Technologies’ broad portfolio of products, powering flagship smartphones, next-generation laptops, and digital cockpits, as well as Advanced Driver Assistance Systems, extended reality and infrastructure networking solutions….”
First, I’m not saying falling prices is a major factor for QTL, but it absolutely is a factor. The ASP of licensed devices swings QTL revenue around materially each quarter. I think the mobile market is bifurcating between Apple at the high end and then lots of low -end phones. That cannot be good for QTL or the rest of the company.
As for Nuvia, see my reply to the next comment. It is a step in the right direction but it will not solve all their problems.