Everyone screwed up

Google and Apple and being the default.

Many years ago, Google made a silly mistake. If you bought an iPhone between 2008-2012 it came with Google Maps pre-installed.

Then, in 2009, Google introduce turn by turn directions to Google Maps on Androids, but did not introduce it on Google Maps on iPhones.

Apple recognised this as an essential feature, and immediately sprang to action, creating Apple Maps.

When Apple shipped Apple Maps in 2012, and changed it to be the default, Google Maps lost 60% of its mobile traffic. 2 months later, Google added turn by turn directions to Google Maps on iPhone.

Google realised that they are a horizontal services company built on advertising revenue, not a vertically integrated consumer product company. Being a horizontal services company means being the best choice everywhere, indiscriminate of device. They make their money from people seeing ads!

Then another thing happened. In 2014 Apple introduced web results into its “Spotlight” search feature (previously used for searching for files, apps and contacts). Spotlight used Bing as the default search engine.

This lead to a complete freakout at Google. Having just self-owned and lost a significant amount of market share on maps, they could now see a future where Apple builds their own search engine too, and of course makes it the default on iPhones.

Google had been paying Apple to be the default search engine on Safari since 2002, but it seems that in 2014 those payments really accelerated. In 2014 Google paid Apple US$1billion dollars to remain the default, and in 2017 Apple changed Spotlight’s default over to Google. Google paid Apple a cool US$9billion in 2018 for the privilege.

It has emerged that these payments are part of a revenue deal where Apple gets 36% of all Google’s search revenue from iPhones. All Apple has to do is leave Google as the default, and occasionally renegotiate the terms, so this is basically pure profit. Profit that has ballooned to $20billion in 2022.

So Apple never built a search engine or invested in the infrastructure to run one. Why would you when get 36% of search revenue anyway to do nothing?

But what if 10 years ago they did build one? What would search look like now? For one thing, with a bit of competition, Google searches probably wouldn’t be so overloaded with ads, and the ad buying market would also have greater diversity. Also, people might be more accustomed to a search experience integrated into the device rather than opening a web browser. And 25% of Apple’s free cash flow wouldn’t come from rent seeking on a technology they don’t own.

Had Apple built a search engine, they would have had to invest in the infrastructure to serve search results at scale. While they’re doing that, they’re building out data centres so they don’t have to outsource iCloud storage to Google, to the tune of at least $300m per year.

Dark

Famously in the late 90s in the run up to the dot com crash at a gargantuan cost companies laid fibre optic cables “dark fibre” all over the shop. When those companies went bust in the crash, Google bought it all up.

In the 2 decades since, Google has built out an international data centre network to support both their own products, and their GCP customers.

These servers include many with their own chips, Tensor Processing Units (TPUs). These TPUs are made for handling AI stuff in a similar way to nVidia’s more powerful and expensive GPUs, but with a far greater focus on energy efficiency. Energy efficiency is important when you are dealing with 5million search queries per minute.

And now we are here. Apple needs to do something about AI. On device chips aren’t efficient or powerful enough to do tasks more complex than summarising notifications or a few text messages. As soon as you want a chatbot like experience, or to write an email, or summarise a news article, you need to go to the cloud.

While Apple does have a few of its own small data centres, what it does not have are the fibre connected distributed data centres with low energy chips designed for the parallel processing of AI tasks for 1.4billion iPhone users.

There is really only one company that is capable of handling that scale, and that company is Google.

At a time when Google’s core business and revenue driver, search, is facing a potential paradigm shift, they have also found themselves in an incredibly strong position at the centre of cloud inference at scale.

And Apple finds itself increasingly having to rely on competitors to deliver features for their flagship product.

If only they had started a search engine.

Interestingly, Google also find themselves at the crossroads once again too. Given Apple will likely have to rely on Google and other third parties to deliver the AI features of the future, and Google have their Pixel phones on which they can tightly integrate these same AI features, what if once they develop, for example, technology to enable an AI personal assistant that offered personalised guidance…what if…Google simply did not allow this feature to be deployed on iPhones…

Listen to my podcast: