Whitechapel Thoughts

I haven't done a tech post in a while. And this seemed like as good a topic as any to add speculation too.

A lot of the articles are focusing on how "Whitechapel", Google's code name for their "in-house" SoC, will compare with Apple's custom chips. And many are, I think, rightly predicting that Google's efforts will not surpass Apple's.

Though, I also think that many of them are making some pretty poor assumptions.

Firstly, a lot of people point to Pixel's sales volumes as an indicator that if cost in an option that Google simply wouldn't be able to produce chipsets at an affordable price point. This is comically off base. It makes it is premised on the WHOLLY inaccurate assumption that Apple and other phone manufacturers enjoy similar margins.

Also, it kind of indirectly makes the assumption that Google NEEDS the phone to be a sales success right away. Keep in mind, Google is a hardware company second and an ad revenue company first. While some level of success is necessary, it isn't the same bar that a company like Apple or Samsung would have for their products.

But, I agree that Google's SoC is unlikely to surpass Apple's out of the gate. Apple has more experience and resources in this area. And if that is Google's expectations (which I supremely doubt) then someone should be fired. 

I also tend to think that it doesn't even need to surpass Apple's hardware for 2 reason; relative performance to the Pixel line and market adoption.

Market adoption is the easiest one to tackle. Not even in the Apple ecosystem does new hardware immediately take over the market share. Take a look at the pie chart here of the device breakdown. A VERY significant portion of iOS devices in use are SEVERAL years old. This becomes clear if you keep scrolling to the pie chart of the processors.

Why is this relevant? Well, if you make software you want as many people to use it as possible (within reason). So, unless there is a specific reason why your software needs all of the raw power of the latest SoC you're going to want to make sure it runs, and runs well on devices some number of years back.

So, let's just make up a number and say that the typical developer will be happy if their software runs well on between 65-75% of devices. Right now that puts people targeting Apple chips so old that even mid-range chips on newer devices outperform them. Don't believe me, here is a comparison between the Pixel 5's Snapdragon 765G and Apple's A10. The A10 loses out pretty considerably everywhere except GPU where it eeks a small win, but would likely lose in real life due to the overall CPU performance being lower.

BTW, I chose the A10 because if you took the SoC pie chart, that is around the point in time where roughly half the chips are the A10 or older. So, if you took all A10 and newer you would probably get pretty close to that 65-75% target. And, we can even go one generation up to the A11 Bionic and Apple is only winning by the slimmest of margins.

In short, you'll likely find that most apps in the Apple Store run perfectly on an A10 or A11 processor. Which would have been flagship SoCs at release. And their performance is the same or worse than a Snapdragon 765G which was only a mid-range processor at its release. So, we can also likely say that most apps in the Android ecosystem would run fine on that chipset. And reviews do seem to back this up.

The next thing Google's SoC needs to do is outperform its own historical lineup. And this is an area where having settled on the 765G helps them out a lot. Since their own fastest SoC is only a mid-range SoC and we've already established that there is no explicit need for SoCs much faster than that at present Google will likely have a well enough received SoC if it outperforms the 765G.

If they can nail that they are fine. And, if they aren't expecting Apple level margins that shouldn't be such a hard target to hit.

As far as unit sales go... I think we'll see a strategy similar Apple's. Their first gen chip will initially be released in their flagship phones. But, the next year, a new chipset will be made for flagships and the first gen SoC will end up in that years A-Line series of devices. Done right, the same chip may make an appearance in Pixel Book devices as well. 

One of the biggest problems with NOT using custom SoCs is that people are much more critical about your choices. Many critiqued the choice of a mid-range SoC for a flagship phone in the Pixel 5. When you have a custom SoC you can ensure that you are always using the most powerful chip from the lineup in your flagship devices. While people will benchmark Apple's chips against the competitors, when it comes to critiquing chip selection for a device those comparisons are to other Apple SoCs only because "obviously" Apple is going to use Apple SoCs in their devices.

Once Google has their own line of SoCs they will likely enjoy the same status. People will still benchmark Google chips against competitors. But, once the switch is made, in subsequent hardware generations there won't be any question about why Google chose a particular chip for their flagship devices. And this reduced criticism will also allow them to continue selling that chipset longer. 

That is the other less talked about benefit. Even when Google puts a mid-range SoC in a flagship phone, it would have gotten a LOT more criticism if it weren't a fairly recent chipset. But, Apple can get away with putting 2-3 year old chips into new devices because older chipsets is exactly how they distinguish flagship from mid-range to low end chipsets now. 

In fact the iPad lineup still sells iPads running the A12 chipset which is 3 years old now. You can buy a brand new iPad running a 3 year old chipset. And this is the 8th gen iPad which was released only about 6 months ago. If Apple were using 3rd party chips they would have been heavily criticized for such a move.

Put another way, without the constant stream of new 3rd party processors to make your chips seem as outdated as they are, you can continue to sell them for far longer and even in new devices which stretches the value proposition. 

And that may be another way to look at this move. This might actually be a longer term investment for Google. What might be a lower profit margin in the Pixel 6 could become a huge profit margin in the 6A later. And when Whitechapel's successor comes out, likely in the Pixel 7, you might later see a Pixel 7A variant running the first gen SoC and a more Premium (perhaps XL version) running the 2nd Gen SoC.

Given that some people are pointing to how long the chipsets can be supported as a potential reason, this makes sense. If Google can support Whitechapel for 5 years, they could sell it for 2 additional cycles and at the tail end of that they would still be able to offer the same 3 years they offer today. If they can support it longer, it will be even better.

And, there is one option Google has which Apple doesn't; Google could sell the chips to partners as well. This way they could either profit off the chips directly, or just drive up volume to drive down costs and increase margins.

So, there are my thoughts. Google only really needs to outperform the 765G at a reasonable price point and then be able to support it for a few extra years. If they can do that, and then iterate on it then they can sell those chipsets for longer and across more devices. If they can get partners to use the chips as well, then the math works even better and even faster.

While you can always compare Apple and Google directly, it is the differences between the companies where they each have the potential to shine. And the model Apple has used has benefits beyond just making fast chips. Ultimately, I don't know what Google's plans or hopes are for Whitechapel. But, as I said above, I doubt outperforming Apple is the primary factor for this move.