Welcome to the Apple (NASDAQ:AAPL) edition of my "under the hood" series (part I: Google's AI breakthroughs, part II: IBM Watson). This one is a little different than the others because Apple is massively lagging behind in capitalizing on AI (Artificial Intelligence) and analytics. Further, Apple is so secretive about everything it does that we have to look at results instead of explanations. In this article, I make some fundamental observations about how Apple's management deals with a new world of data-driven applications.
What about the data?
Apple is a consumer electronics company, period. This is in stark contrast to being a data company. Tim Cook and other Apple management have repeatedly commented on this. The gist of Apple's stance on data and cloud services is that Apple does not care about your data and, unlike Alphabet (NASDAQ:GOOG) (NASDAQ:GOOGL) or Facebook (NASDAQ:FB), does not monetize it. I am not challenging that assumption. Apple does not need to monetize its user data. What I am challenging is how little Apple cares about the business opportunities around data infrastructure, AI and cloud services.
The prevailing wisdom on Apple is that it needs to diversify away from the iPhone, as sales are expected to fall going forward into Q1 2016. Another contributor recently broke down revenue categories. The key argument from that article is that Apple will be able to increase revenue from services such as Apple Pay, Apple Music or iCloud storage and productivity applications significantly over the next few years. These are all consumer services and they do not help with building applications.
A few years ago, competitor Microsoft (NASDAQ:MSFT) began a transformation from a PC-software company to a cloud services company. This made sense because Microsoft relied on third-party vendors selling enough commodity PCs with Windows and Office to generate one-off revenue from software licenses. With the shrinking PC market, Microsoft was driven towards changing itself into a service business. Then again, Microsoft was never a device company the way Apple was. Microsoft's cloud business is thriving after this successful transformation.
Meanwhile, Apple's strategy is still the same one that has worked for it these past decades: create incredibly well-designed devices that integrate into the Apple ecosystem and sell them at fantastic premiums (Apple makes almost all of the profit in the phone market). As iPhone sales make up about 67% of total revenue (with iPads and Macs both <10%), new revenue leaders are what Wall Street wants and Apple needs. Since Tim Cook (and, arguably, Jony Ive) took over, the recipe of identifying new device categories has primarily brought us the Apple watch. It might bring us a car, although that does not seem so sure any more. More recently, rumor has it Apple might be looking at virtual reality (VR).
The question is whether this recipe will really continue to work for Apple long term. Here is why I think it will not. The 20th century was arguably the century of the PC, and with it an explosion of other microchip-driven devices. Apple has been right in the center of this explosion, and especially in the first decade of this century, has transformed mobile devices as an electronics category. It is temping to think that Apple can just go on to produce more devices in new categories and dominate them as well. Humans are really, really bad at anticipating how much things can change.
Here is how I think things are about to change: The world is changing from a device-driven world to a data-driven world. Take a moment to reflect on how much devices have changed between the 1990s and now. Now run your imagination on how much they will change in the upcoming decades. My prediction is that devices will become less important than interaction, services and algorithms. We are already seeing the first steps of this in the massive surge of Software-as-a-Service businesses disintermediating traditional business models. This has become possible ever since computing power and data analysis services have become commoditized through cloud infrastructure providers such Amazon (NASDAQ:AMZN) Web Services.
So far, Apple has done little to capitalize on this shift. The iCloud has almost 800 million users now, but those are retail-centered storage and message services. Apple hosts apps and provides development environments and programming languages for them (although Objective C is an abomination). Apple does not provide infrastructure as a service or any cloud services that would allow developers to easily integrate natural language processing, image recognition or predictive analytics into their applications the way IBM (NYSE:IBM), Microsoft, Amazon or Alphabet do. This means that in a world where applications become more data-driven, developers have to leave the Apple ecosystem for third-party providers. An overview of machine learning-related resources for iOS can be found here.
I find it weird how little Apple seems to care about this.
Siri and other secrets
So what is Apple thinking about all this? I said this would be an article of my "under the hood" series, but Apple is annoyingly secretive about everything it does. When Apple researchers attend conferences, they take everything in, but don't give anything away, while Alphabet and Facebook are publishing papers almost weekly and releasing various open source tools. Their efforts and funding are fueling an explosion in the development of more intelligent services. These companies have become natural destinations for AI and software talent, while Apple is a destination for design and hardware engineering talent.
Apple's flagship machine learning product, Siri, was an acquisition. The Siri back-end was originally built by Nuance (NASDAQ:NUAN), a leading speech processing company. I could describe how Siri roughly works, but it is very similar to Watson in many ways. However, there is no key publication I could refer to. Even though the Siri founders themselves were known figures in the AI scene (e.g. Adam Cheyer), they have left Apple since. Whatever was known about their pre-Apple work is probably not very indicative of today's Siri.
Siri's components are speech recognition, sentence parsing, intent analysis and, centrally, connection to lots of third-party web services like OpenTable, WolframAlpha or Bing to collect whatever data is relevant to the query. Finally, text-to-speech translates the output into Siri's voice, the mechanism of which is described here (although, not in a very technical manner).
The devil is, of course, in the details, and unlike IBM, Google or Facebook, Apple has no interest in letting us now what is under the hood. This is a shame, because Apple could, with some effort, provide individual modules of Siri as web services that could be integrated into iOS apps. This would be a great alternative to Watson that would allow Apple to break into the cloud analytics segment.
Now, Apple is, of course, not just sitting on its hands. Instead, it is quietly acquiring small AI startups here and there. Also, it is reported to actively hire in its machine learning division. Indeed, there are quite a few jobs like this offered on LinkedIn (NYSE:LNKD):
The problem is that this is not enough. What Apple, in my opinion, is lacking is someone driving its vision in this area. Just acquiring a few startups here and there cannot replace someone with a long-term vision of the future that reaches beyond the next few devices. Facebook understands this and has hired leading machine learning figures to run its AI research lab in New York. Alphabet is strongly driven by Larry Page's "moonshot" vision towards a general artificial intelligence that is permeating all of its services. Even Twitter (NYSE:TWTR) has acquired leading expertise through the spinoff of Harvard machine learning star Ryan Adams.
See, the reason these other companies are so attractive for leading figures in the field is because they a) have lots of interesting data, b) lots of money and c) encourage publications. Apple has money, but as soon as you enter its gates you have to stop publishing your work, which researchers dislike.
The takeaway from all of this is that I think Apple's secrecy is hurting itself. With regard to its devices, secrecy is part of the Apple marketing myth and this has worked incredibly well. In contrast, I see no reason why Apple is not more open about its efforts with projects like Siri.
The difference between Apple's past and Apple's future is that devices like an iPad or a Macbook (on which I am writing this article) could stand on their own due to superior hardware design and user interfaces. Future projects like a car will need to rely on data services, algorithms and learning capabilities to a much larger degree. As of right now, Apple is not fostering the right environment to ensure that such projects will be successful.
What Apple needs
I think Apple needs to rethink its strategy around data services, AI and the cloud. The company may very well continue to dominate mobile device markets even without doing so. However, if management does not take action soon, applications will increasingly shift towards using the data and analytics services of Apple's competitors. If algorithms eat the world, devices will become marginalized compared to the intelligent services running on them. Tim Cook needs to make a key hire to lead Apple into the data-driven future and ensure the company's ecosystem continues to thrive.
This article is part of my ongoing 'Under the hood' series in the technology space. If there is anything in the software/machine learning/analytics space you would like me to cover, please comment or send me a message. Future installments of this will include Amazon (Web Services), Facebook and Baidu.
Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours.
I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.