Why Google, Not Apple, Risks Losing the AI Platform Battle

I know how crazy the title of this post sounds. It’s hard to think of another company better positioned than Google and their decades-long business of machine learning to succeed when it comes to AI. However, there are some underappreciated elements of this debate I’d like to throw into the discussion.

Baseline Assumptions and Market Fundamentals

We need to clarify what the baseline assumptions are. First, the next platform will be born from the mobile computing foundation established with smartphones. There has been no larger democratization of computing than the super computer in our pockets. This single movement has brought a computer to over two billion people and will eventually connect over four billion globally. In retrospect, the PC, in its desk or lap-based form factor, was actually a barrier to bringing a computer to every person on the planet.

Second, iOS and Android are the dominant mobile operating systems and therefore both will play a key role in ushering in the next platform whatever it will be. Apple’s iOS has an active installed base of around 850-900 (iPhone + iPad) million people and Android has an active installed base of around 1.8-2 billion. But it is in this part of the discussion we have to recognize that Android running on nearly two billion devices does not equate to Google’s version of Android or a version of Android that ties to Google services like Maps, Play Store, Search, etc. If we exclude China, we eliminate somewhere between 600-700 million devices, meaning a version of Google’s loosely controlled Android OS, with some of their core services, runs on about 1.4 billion devices, mostly smartphones.

Third, we must recognize Google’s somewhat loose control of Android with their partners. They require their major partners like Samsung, HTC, LG, Huawei (outside of China), Xiaomi (outside of China), Micromax, and many others, to pass official certification and agree to specific terms in order to sell a device that includes essential items to compete in the global market — Google Play store, Maps, search, and other core Google services. However, they also allow these same partners to add some of their own custom services, some which compete with Google’s core services, in order to add an element of differentiation.

Enter Google Pixel

Without understanding my third point, it is hard to appreciate how strategic the Pixel is to Google’s future and specifically, their future where an artificially intelligent smart assistant becomes something a consumer interacts with many times throughout the day.

The Pixel is positioned specifically as the only smartphone on the market that exemplifies the best of everything Google. This statement alone is a recognition of the diverse landscape that is Android where Google’s most important partners are NOT shipping devices which represent the best of the Google ecosystem as they try to add value on top of Google services and, as a result, share the customer with Google. The Pixel is purely and simply Google’s attempt to provide a solution to the market where they do not have to share the customer with anyone else.

This is a critically important point, as it is tactically important for Google to own and not share the customer when it comes to AI if they are to truly have a role in the platform of the future. Here is where the hard numbers make my point.

Do the math on the major partners Google shares the customer with and we see the issue Google is faced with when it comes to an artificial intelligent personal assistant. These core partners are likely to add their own AI agent to their Android devices, and will leave Google’s off, so long as Google gives them the option to:

  1. Samsung
  2. Huawei
  3. Xiaomi

Those three are the ones who are already confirmed to be working on their own set of AI personal agents. It is possible a number of other growing Android OEMs like Vivo, Oppo, LeEco, etc., will also integrate their own agents but let’s just assume it’s only Samsung, Huawei, and Xiaomi. The issue for Google is those three vendors alone make up on average between 34-38% (roughly 130 million per quarter, north of 500m annually) of all quarterly Android smartphone shipments. Perhaps most importantly, they make up the most important 34-38% because those three make up the vast majority of smartphones that cost more than $300 — a more valuable customer to Google. Right there, Google’s AI assistant is likely locked out of 34-38% of the Android smartphones sold each quarter. Add Apple, since no Google AI assistant will be tightly integrated to iOS because Apple has Siri, and Google is locked out of an average 47% (nearly 200m per quarter and ~800 million annually) of smartphones sold each quarter, but greater than 90% of smartphones sold are more premium tiers — the most valuable customers. Assuming Google does let their assistant be shipped on other Android smartphones, you are talking the bottom of the barrel vendors who have no brand and sell most of their phones for under $300 to customers who will provide little to no value to Google on the AI front. This is the real issue Google has in front of them in getting their AI assistant into the hands of billions of customers Google can monetize.

The Pixel for Google is an offensive play against this dynamic to try and gain share in a critical part of the market where they can control the AI element and bring the purest version of their AI to a set of customers they don’t have to share. Of course, Google will offer their assistant to their partners but, in reality, few will take them up on the offer. The only other play, which is possible but would add a gallon of fuel to an already extremely hot fire of anxiety of Google’s partners toward them, would be to force their partners like Samsung, Huawei, Xiaomi, etc., to ship the Google Assistant on their devices as part of their certification to ship other services like Google Play store. Google, however, has to closely walk the tightrope of monopoly-like behavior in what they force on their partners when it comes to core services. My sense is they will leave Assistant as an option. Which, in turn, means they have an exceptionally narrow path to try to get their AI assistant into the hands of the masses. To put it simply, Google either delivers it to consumers themselves through their own hardware or it is unlikely to go anywhere.

Android may have the vast majority of smartphone market share but that reality will not translate into Google’s AI having the vast majority of share in the scenario I lay out above.

From a smartphone and tablet hardware standpoint, Apple has the largest installed base of customers followed by Samsung, then Huawei. Arguably, these three have the greatest potential to share the market for AI personal assistants since these three brands have the majority of consumers on the planet owning one of their smartphones.

Published by

Ben Bajarin

Ben Bajarin is a Principal Analyst and the head of primary research at Creative Strategies, Inc - An industry analysis, market intelligence and research firm located in Silicon Valley. His primary focus is consumer technology and market trend research and he is responsible for studying over 30 countries. Full Bio

45 thoughts on “Why Google, Not Apple, Risks Losing the AI Platform Battle”

  1. Okay, I will try to use analyst language (poorly).

    So Google looses the “Assistant’s War” to other’s within Android. Could the user not decide which assistant to, you know, actually use? Ostensibly these assistants can be provided as a separate application. This more than anything is the “democratization” (versus prevalence) of computing.

    Remember Siri was an App before Apple bought it. When Apple purchased Siri, everyone’s copy stopped working. Some “democratization”…

    1. “Remember Siri was an App before Apple bought it. When Apple purchased Siri, everyone’s copy stopped working. Some “democratization”…”

      Now that Apple has made a strong and secure framework for Siri, it has been able to open it back up with APIs for all app developer’s to take advantage of it (Like HomeKit, Healthkit, etc.).

      So, that is one type of democratization that Apple has moved toward — all app developers can now integrate Siri into their apps and go on to do new and interesting things to make their apps ever more useful.

      1. I think you misunderstood. Maybe not. People that had purchased Siri up until that point had their purchased software rendered useless.

          1. I must have got in early then, because I didn’t pay anything for it. Regardless, they took it away but then brought it back, giving it away as part of the OS. I don’t see how there is much to complain about.

          2. There is profoundly and fundamentally much to complain about.
            a) They stopped a working product from working. A product at least some paid for.
            b) The removed a product from an owners machine.
            c) In many cases a new machine was required to run the re-introduced product. Since Siri runs server side, there was no good technical reason to do so.
            So as you see, it’s a matter of principle.

  2. 1-The flaw in your argument is that you make an equivalence of an AI platform to the assistant in your Smartphone, when in fact an AI platform is Cloud base computer engine that will be on almost everything in the future, actually the assistant in your Smartphone is nothing compare to all the other services and new product that an AI platform will enable.

    2- Just because other OEM want to create their own Assistant, does not necessarily mean that it will provide a good enough services or even be useful enough to discourage users from switching to the one provide by Google.

    2-An AI platform is not an App nor a service, it is a Cloud Based intelligent engine, that will interconnect every Hardware, services, location, identity and knowledge of almost everything around you through the cloud and will be available everywhere. On almost anything How do you expect Apple to wins when currently they cannot even provide a basic ICloud services that work on the iPhone or have any useful services that much user will use to provide good data set?

    This type of analysis is the reason I sometimes wonder whether you guys truly understand what an AI platform really is

    1. Go back and read my article on the role on device AI and training will work vs the cloud stuff. It is simply not a one or the other approach. So there has to be a on device role which integrates both the inferencing as well as the training of the AI agent with personal secure data. I spent the last week with everyone in the semiconductor industry focusing on AI/ML and not a single expert disagrees that there is both on device for personal stuff and then cloud for communal stuff. So you simply can’t ignore the part that will require tight integration or it will go no where. This is where the arc of my article emphasizes the weakness Google has becuase they don’t actually control or own the customer, their partners do.

      1. This is where the arc of my article emphasizes the weakness Google has because they don’t actually control or own the customer, their partners do.benbajarin

        That’s the flaw in your argument, Ben, we’re talking about different thing, you argument are centered around the phone assistant, while mine is around AI platform,

        It Not necessarily about who make the hardware, it’s about who can provide better services and solution that most user can access and want to use.

        Google today has more control over the experience of an IPhone user than you think because majority of those users use Google services which provide them enormous information and Data for Future Services and Solution.

        Why do you think that all the Google cloud services perform better in the IPhone than that from Apple despite having total control of the software and Hardware integration?

          1. I think you’re the one under-estimating how much users are engage in Google services through YouTube, Gmail, Google Map, Chrome, Google Search, and so on. ?

            How much of their services do you think are in the top 10 most used apps?

            Where do you think their 25% / year revenue growth comes from?

            Again, you center your argument about AI around the assistant in your Smartphone the same way that much pundit used to centered their argument about computer around desktop computer, what I’m telling you an AI engine is a lot more than that, that your Smartphone is just one form of output, and that there will be a lot More in the future than you think.

            The vision behind having your own personal Google isn’t only about Google search per se, you need to use your imagine to understand where their heading with this kind of vision

            Google knows a lot more about you than you know about yourself, they remember more information about you than you brain can even comprehend

            Having your own personal Google in 5 to 10 years from now could also mean having your own personal Google robots.

          2. One thing I would add is that Google does not have the financial incentive to turn their AI assistant beyond a search bar.

            The search bar is where Google makes money. If their AI assistant was really good and people used it instead of search, Google would lose money.

            Contrast this with Amazon which can make money when people use the Echo to purchase stuff, or by locking customers into Prime. Or you can contrast it with Apple with makes money by providing the best experience they can.

            If Google launched their Assistant on all Android phones and customers used it instead of search, Google would have a huge problem on their hands. Simply put, Google Assistant is detrimental to Google’s business model.

            Google continues to see great revenue growth, but Internet Ad spending is running out of space to grow. Competing Internet Ad networks are seeing decreasing revenue. Digital Ad spending is already 1/3 of the total Ad market, which means the ceiling is getting closer and growth is likely to slow down significantly in the next few years.

            Google’s biggest challenge will be how to maintain momentum when the Internet Ad market saturates and inevitably slows down. The cautious roll out of Google Assistant should be seen also from this view.

    2. If argue that at this point, nobody really knows what an AI platform really is. At the very least, nobody knows what AI platform consumers will really appreciate.

      1. Just because you do not know, or many experts who write about it does not seem to know either does not mean that those who are building it, test it, that sees a lot of progress when using it In-house to solve complex problems that were once said impossible, and who are testing in-house solutions to bring to market soon do not know either.

        sundar pichai wasn’t dreaming when he said we were going to move from a Mobile First world to AI first one, it was a reflection of what they have in Lab that will would probably see soon and the progress they have seen internally solving complex issue.

          1. Most respectfully, and my the same token:

            “nobody knows what AI platform consumers will really appreciate.”-Naofumi

            That’s the secret? Customer appreciation?
            Not market leverage, lock-in, anti-competitive practices, but customer appreciation?

          2. Depends on how important AI will ultimately be. If it turns out to be AI first, then appreciation will be the driving factor. Of AI is just one feature, then the other parts will be more important.

          3. I tend to think a good voice UI is what people will find useful, and if a product delivers value/benefit it can succeed in a consumer market (the user is the buyer). Other factors come into play of course, but if the user has a poor experience and doesn’t receive much in the way of value/benefit, the product will not succeed.

          4. “That’s the secret? Customer appreciation?”

            Yes. In consumer markets products and services must first deliver some real value/benefit to the customer, something that matters to the customer, the product takes care of a job-to-be-done, and the customer ‘appreciates’ this. If the AirPods aren’t good and don’t do anything useful for customers, they will not succeed. Marketing, lock-in, et al won’t save them.

          5. On the topic at hand, a backoffice technology, that position is naive, at best.

            When Stac offered the bear disk compression, MS used Doubledisk, bundled it with DOS, and killed Stac. Stacy won the subsequent court case. So much for customer satisfaction.

            I personally believe Apple is far more manipulative than MS ever was, or even could be.

          6. Technology is now consumer-facing, in the context of people making use of AI or Voice UI products, so how you sell to actual consumers is the topic at hand (quite a lot changes when the user is also the buyer). You continue to misunderstand why Apple is succeeding (hint, your personal beliefs don’t come into play).

          7. The interface is but one aspect. It’s like selling a “blue one”. Until Ford wins and everyone gets a “black one”.

          8. Not even close, I didn’t say anything about “the interface”. Apple sells the experience, the jobs-to-be-done. That either delivers value/benefit and sells, or it doesn’t and it fails. There’s no actual magic or trickery that makes people buy Apple products.

          9. Okay. As you proclaim.
            Backstage shananigans never have anything to do with anything and I can only assume Windows in the ’90s was consumer facing and correct, because, you know, far more consumers chose it.

          10. Windows was never consumer-facing. That shift only happened recently and as we’ve seen, Windows lost, going from a dominant position in computing devices to what Microsoft admitted was a 14 percent share in 2014. I would assume that’s even less now with the growth of Android and iOS (as well as Mac).

          11. Fascinating, you really don’t understand how computing devices have only recently become consumer-facing, and why that makes all the difference in buying patterns. So you seek to explain Apple’s success in other ways, such as marketing, lock-in, anti-competitive practices, and so on.

          12. No sir, I don’t understand how your sweeping proclamations only apply as you choose to apply them.

            When MS ruled the roost, they truly solved jobs as well. Jobs done by consumers as well as professionals. They delivered real value/benefit. They still do. There was also VERY high customer disapproval. Didn’t matter. They were at the top in spite of this due to other factors and practices.

            BTW, you presented BS when contrasting all computers. Did that count cars, lathe machines, and scientific instruments too?
            MS’s drop (according to you) was an admitted 14% of all computers. If it was all computers, then I’m certain it was far less. I’m also certain that under the same data set, OSX computers were one tenth of that, or less.

            Now, you just can’t have it both ways. I clearly demonstrated that despite customer dissatisfaction. MS was (and still is) at the top in PC’s. Your thesis doesn’t hold water, customer satisfaction as the sole criterion of success. There are many criteria to be fulfilled for success. Period!

          13. When we talk about jobs-to-be-done we mean the buyer is hiring a product or service. You are correct that this did happen with Windows, but the dominance of Windows was spurred by adoption in business where the end user was not making the buying decision. Windows became a de facto standard which led to purchases among early adopters in the consumer market, helping to grow the dominant position of Windows. But, Windows was decidedly not consumer-facing. No computing device was at this time. The shift to the consumer market happened much later. Don’t get confused by consumers simply being able to purchase a Windows PC (effectively their only practical choice in most cases), that doesn’t mean PCs were consumer-facing.

            The 14 percent share data comes directly from Microsoft: “Microsoft Chief Operating Officer Kevin Turner said that Windows’ overall device market share is at 14 percent. Windows still has a 90 percent market share in the PC category, according to Turner, but the majority of computing devices are to be found in the mobile space.”

            Microsoft knows mobile devices are computers.

            Now, we all know Windows rules the PC world despite customer dissatisfaction, I never said anything to the contrary. But even Microsoft admits that they no longer rule computing devices, and that has everything to do with how computing has shifted to be truly consumer-facing, how abstraction and simplification (and mobility) has taken computing mainstream. Now the end user is almost always the buyer, and that changes how buying decisions are made.

            You are also correct that the inner workings of AI are not “consumer visible”. This is not relevant since we have shifted to abstraction. The end user will make the buying decision based on the experience of the product/service and jobs-to-be-done. The product/service will succeed or fail on its own merit. I have never said no other factors come into play, but when you’re dealing with a true consumer market it is the buyer’s experience with the product or service that is paramount.

          14. This is the closest we have come on agreeing.

            Except for this to a degree…
            “shifted to abstraction”.

            Yes, it’s a convenient way for users to use something. Fair enough. It also masks the truth. There is no real “magic” in that box, it’s cosmetic. You will use what your ecosystem offers. Period. That’s all.

            The comment about MS having only 14% of all computing devices is actually irrelevant to me. I truly don’t care. For the purposes of this discussion, that also means OSX machines are one tenth of that, or less. Both statements must be true if there is to be an accurate picture.


            There are some things general consumers aren’t really qualified to judge. They are entitled to their opinions, to be certain, but there’s only room for opinion in the absence of actual knowledge (by definition).

            Pharmaceutical companies advertising directly to patients comes to mind. Is the general patient really informed enough, and rational enough, to making the right decision on the best medicine for something? In a weird way, that influence amounts to the very patient practicing medicine without a license. Where it is useful is in having an intelligent conversation with their physician in determining the proper course of action. I grant you, this is also an extreme example to stress test an idea.

          15. You should care about Microsoft’s 14 percent share, it demonstrates the shift to consumer-facing computing devices. Now that computing has shifted to consumers OSX seems to be doing quite well compared to the rest of the PC industry, which is what we should expect.

            Abstraction only masks complexity, incorporating that into the overall experience which is made simpler in the process. This is a good thing and it is the direction in which progress moves. It is not cosmetic, not at all, it is in fact mostly functional. You seem to be confused by the basic concept of abstraction. It is not Apple making things look pretty.

            Pharmaceuticals are not truly consumer-facing for the most part. There are exceptions with simpler drugs, which proves my point about abstraction actually.

            Here’s an exercise which might be useful. Explain to me why Apple is succeeding. I have some notion of what your answer will be, but I’d like to see if you can articulate why Apple is succeeding. Of course I am assuming you understand that Apple is in fact succeeding. That might seem silly, but some still argue Apple is not succeeding today. So let’s start with the premise that Apple is indeed succeeding. Why is that?

          16. I really have no need to care for MS’s share. I’ve said it before, PCs were oversold. If all you need is email, web browsing, and light WP, mobile is enough. My wants are greater. Consumer facing is fine as long as it doesn’t get in my way. For instance, I don’t think you mind OSX retaining the command line (or do you….? ;-)).

            I could dig it up, I’m sure you have it somewhere too. If the PC industry is tanking (it’s not), OSX is too. It’s on average (generously) 1/10 of the Windows market. What is tanking is buying a $1000 PC just to do what a phone can do, without the calling features. What is tanking is the overselling of the PC.

            Explaining to you, or anyone, why Apple is successful is something I care even less about. Any company really. It’s not something I spend any time thinking about. But since you ask, here’s my topmost thought….

            Apple is succeeding because they are developing for the average. This comes at a cost to the rest.

          17. That’s not far off from what I thought you’d say. To sum up, you don’t understand why Apple is succeeding, but you are trying to explain it in terms of your personal beliefs and narrative.

          18. You clearly should have understood that I answered you out of courtesy, after having politely told you that I don’t really care how any company is doing.

          19. For someone who claims they don’t care about how any company is doing, you sure comment on a lot of sites and articles focused on Apple, going as far as creating the alias applecynic on MacDailyNews. That sounds a lot like someone who cares a great deal. Care to try again?

          20. That only goes to show how shortsighted you are. Let me make it simple for you, I care for the sport, not a team.

            Feel free to spin that as you see fit, but let me go on record for myself in saying, I root for no company, with Apple having the benefit of my contempt on matters on user freedom, not business, grounds.

            My alias on Macdailynews is accurate, and transparent as to my intentions. Now please go back to your hallucination inducing space bananas.

          21. If you care about the sport then you’d naturally want to have all the data you can get about the sport, and that includes all the team stats, which gives you a deeper understanding of the sport, how the game might be evolving, how the players are changing, what rules need updating, and so on. Your pose of ‘not caring’ is what is transparent.

          22. That would be a reasoned approach… If. I . Cared. I don’t.
            I’m not an investor, employee, or supplier. As a customer, real and potential, I am at the opposite side of the table from ALL vendors, including Apple.

            Meanwhile, I place no faith in the clowns Apple competes with, so if the future you envision does come true, thank goodness for Linux.

          23. Actions speak louder than words, and your actions reveal that you care about Apple very much.

          24. I normally associate caring with positive things, but you’re absolutely correct. That must also mean I care about you too.
            Last word is yours. See ‘ya.

Leave a Reply

Your email address will not be published. Required fields are marked *