202. AI/ChatGPT and the Race Between Microsoft and Google to Win Samsung Mobile Search
AI might sound like the reason, but with Big Companies it's all about money and control, not features.
A twitter thread originally appearing April 17, 2023 with a few tweaks.
Google Devising Radical Search Changes to Beat Back A.I. Rivals // What an amazing story of marketplace competition. New ideas from underdog to challenge much-disliked incumbent with > 80% share. Except this has nothing to do with technology at all. It’s all money and control. 1/
2/ Super easy to spin this up as a story about a “better technology”. Everything is better with AI so of course search should be too (though Google has always used AI in search). While AI might have been discussed at a first meeting over dinner, all that follows is money/control.
3/ At the root of this is a relationship where Google pays the major phone makers BILLIONS of dollars each for default search across the phone. These are enormously complex deals. They are also the root of one major tension between Google and Samsung.
4/ Amazing thing about phone business (1.2B phones shipped 2022, Samsung is ~27% of those) is how so much of phone economics flow to two companies: Apple and Google. That’s the first tension: money.
Samsung did $245B/$35B rev/inc v. Google of $283B/$60B. Samsung includes phones, components, everything.
5/ Samsung phones (MX division, or mobile experience) $80B/5.2B of that roughly. Samsung is a huge and complicated company with complicated relationship between divisions and mostly inscrutable to the rest of the world. It is also 20% of Korea GDP.
As an example of the complicated nature of Samsung. The components group sells hardware (screens, memory, chips) to both Samsung Mobile MX and to their competitors (including Apple) while Samsung MX buys chips and components from other manufacturers. The flow of money between them is impossible to track from the outside.
6/ tl;dr Google makes most of the money in the mobile phone market, except for Apple. (Leaving out carriers because they are not global providers of their service) OTOH this seems to make sense given they make the phone software. Except that is supposed to be open source. What happened? That’s the other tension point: control.
7/ Google won the battle but is (rapidly) losing the war. The idea was to provide software and OEMs would just stamp out commodity phones. Google knew value of software to it. Using the platform they gained more information flow which opened more sockets for ads (payments, maps, mail, video, and plain search).
Open source was in a sense a head fake. Early on with phone sales exploding it all made sense. In fact, the phone makers got exactly what PC makers had always wanted from Windows—the ability to change whatever they wanted. Then they realized just how difficult that is to do.
8/ Now here we are with phone unit sales down/slowing as market is saturated. Phone makers at first were happy just complaining but now see that not only do they not get a “fair share” of search but without them there would be no search market. They’ve seen Pixel remain tiny.
9/ First thing phone makers did was try to gain control of all entry points—obvious ones—in phones. Photos, chat, maps, speakers, home IoT, music, books, video. Plus they tried to add features to Android they felt were “missing” which was everything from 5G to Pen (S-Pen anyone). This turned out to be much more difficult than anticipated. Software is difficult too, especially when doing battle with Google.
10/ This made for a kind of redundant and suboptimal user experience and a lot of tension with Google where its strategy seemed to get undermined. That led to the splitting of Android into the core part and the app part. This line is the source of endless tension and shenanigans on all sides.
11/ While anyone can presumably build a photo app or cobble a book store together, building search and maps is enormously difficult (though perhaps so is building phones). Phone makers feel they should benefit from the economics of search. They too want an annuity stream and profit. They deliver it to customers directly in the phone, after all.
12/ This leads to the negotiation over defaults—frankly something created by Microsoft and then litigated. These billions in payments from Google are a crazy complicated deal that arrives at some definition of fair value for the search default. Tied up in all this is to the degree possible trying to avoid mucking with Android too much.
13/ This is not some simple calculus. There is a vastness of give-take that comes with a deal like this. In a famous moment in the Microsoft trial, relationship between PC makers and Microsoft was described. It was an unpleasant description arising from very tense negotiations:
14/ Two huge companies negotiating is like observing two bears standing off in the woods. From a distance you can’t tell if they are playing, fighting, or mating. In the end the fisticuffs stops and the bears sort of just walk away. When two $250B companies deal it’s like that.
15/ Samsung does not just “want good search” in practice they could genuinely care less about quality of search results. Google does not just “want search on Samsung’s flagship phones”—no phone/maker matters, just all Android endpoints and entry points within an endpoint. Honest. No features will be discussed unless they are about customizing the experience to be unique to Samsung or maintaining Google’s control.
16/ Among a zillion things, Samsung wants to control an experience unique to Samsung phones—the group is not called “phones” but mobile experience. Samsung wants to have the same vertically integrated experience that Apple creates. They hate that Apple is viewed as making so much better phone experiences than Samsung.
17/ So a deal like this is as much as about revenue as it is about an experience. Apple commands much more about the experience with Google and at least Samsung wants that too but doesn't control nearly as much given Android. Privacy, data collection, etc. are Apple variables. Samsung wants much more/different I am sure. They want to do everything iPhone+iPad+Watch+Mac do.
18/ Meanwhile Google wants Samsung to carry the whole of Android and to stop picking it apart feature by feature. Good luck with that. Why? Because regulators are now watching. To them *everything* looks like monopoly maintenance. Google is really constrained. Google can’t just say “here’s the software for free don’t change it” or “if you want this you have to also want this other thing.” Super tricky.
19/ If google pays too much or puts too many constraints then the whole of search draws regulator ire. Bing has many advantages in this regard—it has less surface area, nothing really to control, and can offer Samsung connections on Windows to complete the whole “MX” they desire. All of that without competing with Samsung at all.
20/ Quickly one can see why this whole negotiation has nothing to do with AI. AI might even be negative because it adds to the negotiation a need for globalization, “alignment”, and even indemnity. Plus Samsung has recalled phones too often and really doesn’t want an AI recall.
Consider for example if there is a broad copyright lawsuit over AI models. If Samsung carries the GPT-derived search then they would be the party delivering the infringing technology and given their deep pockets they would be a target. The potential damages would be a per-device claim and enormous. There would be no way for them to not be a defendant. Indemnification was a big part of the way Windows licensing worked and a huge issue for being carried. But no one wants the liability until the law is settled.
The last thing Samsung wants is for their choice of Search default to be viewed as providing bad answers or hallucinating. That would not be innovation. As previously said, I believe search + LLM is not the right combination at this time.
21/ This is why Google is squeezed. If they add too much AI and it is as “crazy” as some say, then Apple—who does not like unaligned stuff—will be the next to question using Google. Status quo works best for phone makers especially as their fees keep going up. (Note: no one has any idea if Apple will enter search on its own).
22/ Samsung might want differentiated search but not at the cost of giving up billions of dollars of zero-effort profit. So Microsoft needs to pay non-economic (for it) in order to “win” the deal. And Samsung can keep using that with Google for…dollars and control. Leverage.
23/ In all this, if/when Google wins the deal, they will end up paying more and ceding a bit more control. And the kicker…Bing losing the deal just looks like the new innovative technology was held back by the monopolist, Regulators hate that. And who talks to them all the time…?
24/ I get why there is a big story on the role of AI in search as a lynchpin of the deal, but it isn’t because AI is cool or AI solves problems. Deals are insanely complicated and the interests competing. It makes a good story. https://www.nytimes.com/2023/04/16/technology/google-search-engine-ai.html
25/ Patterns are not unlike PC/Windows ecosystem. Profits flowing to one party while others trying too hard to differentiate from each other. Toss in regulatory oversight. Conflicting goals. And nowhere does technology enter the picture. Money & control are what BigCos understand. // END
PS/ My view is that this is all “yesterday’s battle”. The future of search is not “adding AI to todays search experience and business model”. The next of any huge business is not just the same thing a little different.
PS/ The next google will have rethought search end to end. A new model for privacy, accuracy, monetization, etc. The next Google will be AI based (too) but wildly differently. It is being built today. No one knows what it is.
PPS/ The calculation that enters into this are the estimates of the annuity value of a user. Samsung seems to be about $10 per user per year depending on how to calculate. Google > $100 per user. If they use a phone for everything then how much is that worth?