Strange AI Bedfellows
In which Jack uses magic pills, climbs a beanstalk, and gets two giants to kill the other giant and steals his golden eggs.
If you came up like me in the 90s and 00s — the "Aughts", ugh — then you probably view Microsoft and Apple as archenemies. Microsoft: the embodiment of corporate monotony, with beige desktop towers. Apple: the free-spirited innovator, with sleek laptops transported by birkenstock through sun-dappled parks. These were the giants of computing we all touched every day.
This idea is decades out of date at this point, right? It's a vestigial memory of a simpler, early-internet era. Since then, more giants have risen from the Earth. Google came first. Then Amazon. Then Meta. And now Nvidia is flexing its GPUs.
These are all multi-trillion dollar companies. But Apple and Microsoft are still the kings of the hill and it turns out those 90s stereotypes aren't totally dead. MS Office products still make up of 20+% of MSFT revenue. And the hippie artist has an iPhone to go with her laptop. Laptop sales are 8% of APPL revenue.. but iPhone sales are 50%.
Today there's a new darling in Silicon Valley and its name is OpenAI. This wunderkind is fueled by Nvidia chips but has sweet-talked Microsoft and others into making these titanic capex expenditures. It leverages LLM architectures invented at Google. One model of OpenAI is that it's operating like a brilliant parasite, harvesting resources from larger hosts and growing stronger without looking like a problem to any of them.. yet.
I'm going to argue that OpenAI's real target is Google. And instead of ever looking like a direct competitor, it's going to use two other (larger) giants to cannibalize most of the business away. Google is still operating very comfortably, with a sense of complacency and expansiveness, anchored by an insane money printing machine in ads and search. But search is climbing the steps to the chopping block. And the guillotine is an LLM.
Hey, I might use Siri
Most people are yawning because the iPhone 16 is out. New camera, new color, everything else the same. But there were some AI demos that forecast a whole new experience.
Apple is teaming up with OpenAI to effectively embed OpenAI's models into the iPhone experience. Most people are still only familiar with chatGPT and so they're just imagining a chatbot on their phone. Think bigger. The 4o (multi-modal) and O1 (chain of thought) models are already well beyond a base model like GPT4. And Apple has always been exceptional in delivering delightful user experiences and their demo here shows the same. AI won't be separate from the rest of your phone, it will be in every part of it from the camera to the home screen to your calendar to Siri.
Yes really, Siri might not suck in the future.
I remain dubious until I'm right. But I can see a world a year or two from now when I'm transcribing my notes in Siri. And getting a recipe. And adding a calendar appointment. And looking up directions.
That's a very different world actually.
Here's what I mean... Google is great at giving answers to all sorts of questions, but these questions often lack context. When you look up "running shoes", what do you want? Do you want the shoes Usain Bolt used to set the world record? Or a comparison of trail running shoes? Or just a link to the last pair you bought? Context is king, and it's worth its weight in bitcoin. (As Tyler Cowen is fond of saying, "Context is that which is scarce.") It's so important, in fact, that Google has built or bought a mobile OS and a browser and an email app and any number of other things just so they could gather as much of that information as possible. With my browsing history in hand, they know which Zappos page I want. Good for Zappos. Good for Google.
Much of my life (happily) is still unGoogleable but I have my phone almost all day, every day. Apple has the potential to gain far more personal context than Google. Crucially, Apple also has something of a track record on privacy. Out of all the behemoth tech companies, they have the strongest trust record. They prioritize on-device encryption and are consistent on end-to-end encryption for things like iMessage.
Apple is a conduit for OpenAI to get its product integrated into our daily lives. It will find and tweak our photos, build us slideshows, listen to our calls and FaceTimes and take notes or action items, and remind us of things we said we'd do in text messages. It will send us reminders to look at that new grill we were thinking about buying. It will ask us if we want our usual grocery list delivered automatically, along with a special ingredient set for Friday night dinner when the Smiths are coming over. It will also have access to our apps. Want some new book suggestions? Let's take a look at your Audible listening history. If the FTC ever forces Apple to have a competing App Store or to lower their 30% app fees, AI integration would be a killer product to either keep users on the Apple App Store or to give a tiered system based on integration.
OpenAI gets exorbitant licensing and usage fees out of this, of course. Google currently pays Apple $20 billion a year just to be the default iPhone browser. This sort of integration is highly valuable and right now OpenAI is getting it for free. The unGoogleable aspect of our lives is important for OpenAI too. There's a finite (though very large) amount of publicly available training data available for models. What happens when it's trained on our text message history and photos? OpenAI is going to get a lot of data. I'm going to be very curious about the security and data sharing agreements between Apple and OpenAI.
Bada..bing?
Microsoft has been with OpenAI for awhile now. They own 49%. They will succeed if OpenAI products succeed. But they'll do a bunch more than THAT.
Microsoft is going to build a whole brand around the Copilot experience. The first was Github Copilot but expect plenty more Copilots baked into Office and other products. The idea is the same as for Apple: context is that which is scarce. The closer you can get to user data - whether it's wrapped up in the Camera Roll on your phone or in the third tab of a spreadsheet - the more valuable these services become. If Apple is OpenAI's vector into the home, Microsoft is the wedge into the office.
And then there's Bing. The oft maligned, ridiculed, and very occasionally used — you know, just for fun - search engine. Microsoft has been trying to make fetch happen with Bing for years. No surprise there — search-based ad revenue is still an obscene 57% of Google revenue. That's worth at least a trillion in market cap. A billion dollars isn't cool, you know what's cool? A trilly. Amiright?
OpenAI gives Microsoft a way to tackle search obliquely. They don't need to consume market share, they need to subsume the market itself. People won't search near as much anymore if you can just get the answer to your question directly. LLMs will do that. Suddenly, Google's cash cow might be looking a bit malnourished.
New Old Battlegrounds
OpenAI is brilliant. Hell, you have to be to execute on products as complex as LLMs. So it shouldn't be surprising that their business acumen is just as brilliant. They're enabling two giants (MSFT, APPL) to guillotine another giant (GOOG) while simultaneously accessing two very large datasets (home and office) with strong moats.
So what now?
Well first of all, Meta will thrive. Their advertising is special because it's so targeted, and social media remains a fundamentally human business. You can sprinkle AI fairy dust on both the experience and the backend execution, but an LLM isn't going to take over Instagram because you care about seeing actual people during your infinite scroll.. just like you want to watch two humans play chess (IF you want to watch chess) and not infinitely-better robot machines.
Google, though? They've got a fight on their hands, and they are severely handicapped by their history of success. Google has been #1 for so long that they have become a freaking VERB. You Google stuff. If they lose enough market share to shake up that reputation, the Street will be very unhappy with them.
Lastly, I think we're going to see some old products get some AI-powered makeovers very soon. Google still has some very important products with very special data. Gmail controls the communication of over a billion people. Chrome is the browser of choice for over 3 billion. And I don't know about you, but I use Google Workspace, not Microsoft Office.
The most compelling places I want an LLM or an AI copilot are in those three products. I want a personalized, highly secure assistant trained on my comms, my thoughts and my browsing history.
I think these are the three most important products Google has. And I expect to see a run on them soon.
Now I have no idea if any of this is right. Google may be a slumbering beast, like Vhagar, but they’ll be fearsome once they take wing. It's all wild speculation. But in an AI world, today's speculation is tomorrow's newly minted trillion dollar value transfer. I'm intrigued enough to believe there's some sort of strategy here.
And no matter what, NVDA is going to keep going up.