Ep 651: Apple’s $1 billion bailout: Why Siri needs Gemini’s AI Brains

This is the Everyday AI Show, the everyday podcast where we simplify AI and bring its power to your fingertips. Listen daily for practical advice to boost your career, business, and everyday life.

Apple is about to write a $1 billion annual check to Google, its biggest rival, for the AI brain it couldn't build for Siri. So how did we get here? I mean, Apple was once the world's most prominent and powerful company, and they've been promising useful AI for years without a single hit, just swings and misses. Instead, they had to bring in a ringer from out of town in Google to fix it all. And Siri was reportedly so bad that it once failed 1 in every 3 requests in internal testing. Their smarter Siri was so catastrophically broken that Tim Cook had no choice but to rent Google's Gemini model to save face. And now the privacy-first company that is Apple is now dependent on the company that monetizes data in Google. And people are looking at this partnership like a business deal. It's not. This is Apple's public surrender, and we should hear it as that. But what comes next is the real question. Will it matter? And did Apple just lose the AI race permanently, or is this the smartest move that they could make? Well, let's find out today on Everyday AI. What's going on, y'all? Welcome to Everyday AI. My name is Jordan Wilson, and we do this, well, every day. It's your daily unedited, unscripted livestream podcast and newsletter helping everyday business leaders like you and me, not just keep up with what's happening in the world of AI, because it's nonstop, but how we can make sense of it to grow our companies and our careers. So today we're going to be talking about Apple's $1 billion bailout and why Siri needs Gemini's AI brains. But if you haven't already, make sure you go to our website at youreverydayai.com. We're going to be recapping the highlights from today's show in case you miss anything, as well as all of the other AI news that matters today. Yeah. Sometimes it feels if you take a day off, you are a week behind. So don't let that happen. Just read our newsletter. It takes like 7 minutes and that's all. But let's get straight into it. This is a bailout. Let's call it what it is. Also, this is Hot Take Tuesday, right? And you all voted for this. So usually put a poll in our newsletter or a poll on our Spotify in the show notes. So make sure you check every day to see if we have a new poll, but You wanted this hot take and here it is. Apple has completely failed when it comes to generative AI, which is crazy because when you think about Siri and the history of Siri, it's almost 15 years old now. And at the time it was revolutionary because it was artificial intelligence. It was natural language processing. But Apple hasn't improved it at all. In fact, they've promised all these improvements with a smarter AI Siri and Apple Intelligence, and they've only been hit with multiple class action lawsuits. So what did they do? They went out and struck a deal with one of their biggest competitors in Google to essentially rent Google Gemini to make Siri smarter. I wouldn't even say smarter, just smart, because Siri's dumb right now. So here's the bullet points of the news. What happened? So according to Bloomberg's Mark Gurman, Apple is finalizing a deal to license a customized version of Google's Gemini model that is 1.2 trillion parameters. So we're not sure if this is a version of Gemini 2.5 Pro, Gemini 2.5 Flash, or maybe an upcoming version of Gemini 3, Gemini 3 Flash. We're not sure. We should be getting Gemini 3 any week or month now. So we're not sure what it is, but it is a big model. A 1.2 trillion parameter model is much bigger than the 100-ish billion parameter models that Apple are using internally right now. And this is a $1 billion $1,000 a year licensing agreement. All right. We're going to get more into the privacy kind of battle because that's very interesting. And it's expected if you're like, all right, well, when's this going to happen? Apple has been crying AI wolf long enough. When is the AI wolf going to actually appear on the screen? Well, March. So, or early spring 2026, it's expected to be part of potentially the 26.4 iOS 26.4 software update rollout. So here's what we're going to go over on today's show. Some hot takes, but you're going to get the details on the partnership that has Apple paying its biggest competitor a billion dollars for AI intelligence because they couldn't get it figured out. You're going to know whether the world's richest company in Apple just admitted that it lost the AI race, and you're going to understand, understand the 3 main reasons why I think that Apple failed at AI. And yet. Past tense, because at least on the generative AI side, I don't think there's actually any catching up. I think they're way too far behind to even compete, right? This is almost— it's, it's going to be like one of those things. If you remember back to high school, you know, there's like the A team and the B team. Apple's going to compete in the B team tournament. No one's watching. No one cares, right? So anything that Apple does cook up internally, because reportedly they will be coming out with their own trillion parameter large language model. It's no one's gonna care. No one's gonna use it. It's gonna be a, let me just cut it straight. It's gonna be a waste of money. And I feel bad because I've talked to people at Apple, actually some of the original people on the original team leads on Siri. So I feel bad saying some of these things, but I'm also just saying the truth, right? I'm lucky enough through this show to get to talk to very smart people and, you know, the conversations that happen before and after we hit record are always the most telling. And I don't think anyone who's in a position in the enterprise has any hope that Apple will ever do anything when it comes to building a model that competes with what we have from Google, OpenAI, and Anthropic. So here's some more details. So this is a licensing agreement for a reportedly highly tuned version of Gemini. So it shouldn't feel or act or spit things out exactly how you would get something from Google Gemini. It will be highly tuned according to Siri's core intelligence. And this is a rental rather than a permanent expense. Apple did say that, yeah, this isn't for the long run, right? This is just a stopgap. It's a bridge. But their goal is to bring visible AI upgrades while the Siri's foundations are built rebuilt for reliability. This isn't the first time these two tech titans have shook hands because Google has been paying Apple tens of billions of dollars a year. So we've seen reports anywhere from, you know, $18 to $20 billion that Google has been paying Apple to be the default search engine on billions of Google— or sorry, billions of Apple devices. So now it's kind of returning the favor. It's almost like a small, like, gift, like, yes, thank you, you know, here's a billion dollars, please help us. All right, so now Apple is paying a billion dollars annually for at least a couple of years, but the Google branding will be completely invisible to users. And according to Bloomberg reports, Apple's really not going to talk about it, although they probably should, right? I think now the Google Gemini models are at the point where it might be good to kind of put that stamp of approval, right? Like, hey, don't worry, y'all. I know we've been telling you this Apple intelligence thing was really smart, but we're giving it to Google. So reportedly they're not going to do that. It's just more going to be a kind of like a white-labeled version of Google Gemini, you know, fine-tuned for Apple's purposes. But, you know, yeah. I don't know. Apple obviously promoted when they had a similar partnership with ChatGPT. So right now, if you have the setting enabled, if you have like a newer iPhone that Apple Intelligence runs on, there's a setting to send more complex queries straight to ChatGPT, right? So Apple had a partnership with OpenAI and they did announce that. And I think that was probably a good move for Apple, right? Even though I think hardly anyone uses it because, You know, if you want a real answer, you're not asking Siri, right? You're just going straight to the source. You know, you're going straight to Gemini, Claude, Copilot, ChatGPT, what, right? So yeah, this is going to be hands-off in the background. So here's my hot take. This is Hot Take Tuesday. Here's my 3 reasons why this happened. All right, so it's admirable that Apple is one of the few big companies that is privacy first, right? But guess what? Large language models need your data, and that is what Apple does not have. So the very premise that they built their empire on, especially since the advent of the iPhone, is, hey, we take your privacy seriously, right? We have a private cloud. Uh, you know, all your data is encrypted. We don't know it. We don't have it. We can't use it. Guess what? That's what you need to have good generative AI. That's what you need to have a frontier large language model, which Apple will never have. Yeah, sorry. I'm going to go check our newsletter and, you know, see all the, uh, apple.com email addresses unsubscribe after this one, but that's, that's the reality. I'm sorry. And one of the reasons is, well, they don't have the talent. Everyone left. Their head of foundation models left, right? Meta, Google, OpenAI, mainly Meta, right? Reportedly paying their former lead, I think, $200 million to join their team. Apple didn't want to play the game. So you are going to pay the consequences. Apple thought they were too big, and that's point 3. So number 1 is the data gap. Number 2 is the talent drain. And number 3 is Apple was arrogant, extremely arrogant. The fact that they legit tried to rebrand AI as Apple Intelligence. Literally go back and listen to, uh, I don't know, that was 2 years ago, their WWDC. Announcement, I think it was June 2024. So about a year and a half ago, I said at the time, I'm like, this is the most arrogant thing and it's not going to work. Right. And it didn't work. All Apple got pushing this Apple intelligence thing, like I said, was multiple class action lawsuits because they promised certain things and even literally ran commercials and it just never came to fruition. Right. But Apple was extremely arrogant in how they handled, um, how they handled this, right? Yeah. Also, Apple's not going to be reaching out to recruit me anytime soon. Um, but that's the reality. They thought they were bigger than the moment. They thought they were bigger than AI, quite literally, because they renamed it Apple Intelligence, and they're not. And they've proven that. And that's why, right, even if you look Right. I talk a lot about market cap on the show because I think, you know, that's a good combination between consumer and public sentiment on do people, does the public at large, do economists, do analysts believe in a company versus their revenue or in addition to the revenue. So market cap is a very good indicator of where a company's going, where it's been, and where it's at today, right? So pre-Gen AI, Apple had a legit stranglehold on the number one spot. It was the most valuable company in the world and it wasn't even close. It looked like something catastrophic would have to happen for anyone to ever catch Apple. Well, the, the catastrophic event was Apple's arrogance to not invest in large language models the way that their competitors did. You need the data, right? Uh, so you either had to do what everyone else did, scrape it, pay other, uh, third-party companies for the data, and Apple didn't do it. They didn't play the game. They didn't pay engineers hundreds of millions of dollars because they said, well, no, reportedly they said, all right, well, hey, you work at Apple. That's not how things work here. All right, you have stock in Apple, we don't pay you hundreds of millions of dollars. And that is why Apple, any month, any quarter, any year now, I would not be shocked if you look at the way things are trending, I would not be shocked to see Apple fall down to the number 5 spot for market cap in the US. It's not, I mean, it could literally happen in a couple of months. It is that close, right? Between number 2 and number 5, it's a couple hundred billion dollars. It's extremely close. Close. So Apple failed those 3 reasons. So that's where we're at today, and well, why we're asking the question, well, why did Apple even need this Google bailout? So according to earlier internal testing, Siri's failure rate hit an unusable 33%. Yeah, internal testing. Siri couldn't get the correct answer 33% of the time. And Can I be honest? I would be happy if Siri got 5% of things right. I don't, I've been an iPhone user my whole life. I don't remember the last time that I used Siri for anything other than like, what's the weather? And sometimes it even gets that wrong, right? It's bad when I have to use Alexa over Siri, which is just as dumb, right? Mainly I'm using ChatGPT voice mode or Gemini Live. So it was a failure. And Apple, well, how did it happen? Well, they tried to build on antiquated technology. So yes, you have to tip your hat to Apple. The original Siri was revolutionary, right? Back in 15 years ago, early 2010s, the fact that you could talk to this machine in a, you know, clunky phone and it could talk back to you, right? It was amazing. But they were building on antiquated technology and they just had tech debt that proved to be catastrophic. And the failure forced a ground-up rebuild while all of their competitors built from scratch. Everyone else built, um, on AI-native architecture and well, Apple try to just duct tape it when they really should have, you know, back in probably 2018, 2019, they should have been rebuilding it from scratch, knowing and understanding, uh, that the transformers were going to quite literally, uh, transform how technology worked. So it wasn't just a blind contract, right? This, this partnership between, um, Apple and Google reportedly wasn't just because like, oh yeah, well, Google, yeah, we've, uh, Google's been paying Apple $20 billion. So let's just, you know, skim a billion off the top and call it even. No, Apple reportedly ran an internal bake-off, uh, between Google, OpenAI, and Anthropic. And Google actually won on price. They undercut, it seemed like Anthropic's, uh, $1.5 billion offer, uh, was deemed too high. But I think if nothing else, This was maybe a defensive play for Google and a marketing play, right? And to block their rivals from having that level of partnership. Because let's be honest, this is a big deal. This is a big stamp of approval, right? That literally the world's eyes are on Apple and Siri, right? Siri has gone from, you know, 10 years ago as a cool party trick to now they're treating it like hardware because it is that important. I think most people realized, even though it seems kind of weird, right? The future of work is, well, it's agents, but it's voices or sorry, agents that you can control with your voice. And Siri needs to be a part of that if they want to be relevant in how people are working, right? Whether that's in 2 years, 3 years, or 5 years. But I do think for the most part, we are going to be talking to agents, and Siri is not able to do that without Google. So it's a big play for Google, even though they are one of the biggest players in the space. And I think it's just them and OpenAI really running away from the rest of the pack. But it is huge for Google, and I think a smart play to kind of block their rivals from having that level of partnership. All right, we got a little more, but before we do, Quick word from our sponsors. It's a problem I hear all the time. The gap between the AI champions and everyone else in your organization is sizable. You might have half a team that wants to fine-tune models by hand and the other half doesn't know what an API is. How do you get them working together on AI that moves the needle without creating a security nightmare? That's where Airia really shines. They built one platform with three ways to work. Your developers can go full pro code and build custom agents with Python. Your business analysts can use the low-code tools, or your domain experts who've never coded, they can use the drag-and-drop no-code builder. Everyone's building in the same secure, governed environment. No shadow IT, no security gaps. And because they're model agnostic, you're not locked into one vendor's ecosystem with Airia. You can even A/B test different agents against each other, try different models, and when you're ready, deploy to production on the same platform. Your AI strategy should unite your team, not divide them. Check out Airia in today's show notes or on our website for a free trial. Go to airia.ai. Ai.com. Get rid of the AI gap and move forward with a more resilient AI ecosystem. All right, so here is the big question when it comes to data, right? Like I said, Apple has been putting everyone's data on lockdown. Google makes billions of dollars in revenue every year from essentially collecting data. So how can Apple trust its biggest rival in knowing that it's going to be secure? Well, so this is all according to reports. The Gemini model will run exclusively on Apple's private cloud compute architecture. So Google will get no data, right? All of that real juicy stuff, right? Which we talked a couple of weeks ago on why I think actually OpenAI and ChatGPT are going to be a big player in the advertising game and a big competitor for Google. Well, the amount of data that you can get on a person from a large language model chatbot conversation, right? And we've seen that Siri is also going to be offering kind of more of a text interface as well. That is way more valuable than normal search engine data or, you know, a tracking pixel or something like that. But reportedly Google won't get any of that. It's essentially stateless compute. So it erases all logs. Right? So people are like, oh, okay, well, if I'm talking to Siri, it's all going to end up in Google's hand. Reportedly none. And there's layers of security that ensure Google is only a model provider, white label, hands-off, not getting any of the data. What I'm going to be looking at is the cost, right? Because one of the reasons why Apple has kind of clawed its way to the top, I'd say in the 2010s and early 2020s, their margins, right? Their margins are insanely good. But this is where it's going to get interesting, right? And some good engineering here. And we'll see if this is ultimately, you know, more Apple's doing or Google's. But this is a big model, right? So if you're running a 1.2 trillion parameter model, that is a significant cost, right? Going to the cloud and the extra layers of added protection, I'm sure are not cheap. right? Running the private cloud architecture. So reportedly this Gemini version of Siri is going to use a mixture of experts design that uses Spark, sparse activation to help manage this expense. So to oversimplify it, and, you know, I'll save the, the mixture of experts episode for another day. Essentially, you're not using the, the entire, you know, 1.2 trillion parameters every time you go because otherwise the cost would be astronomical and Apple's historically great profit margins would be shrunk down to nothing. So instead they're going to use sparse activation in this mixture of experts designed to essentially only activate just the parameters, right? Just the minimum amount of parameters that it needs to really hopefully provide a better experience. And Apple usually wins on margins, so that cost is going to be imperative for that to work. With Apple and for them to just not bleed. So what does this actually mean for the rest of us? If you're an iPhone user and you haven't touched Siri and you're like, this could make my life better. Why doesn't it work? Right? My wife is always asking this. She's like, why can't I just like talk to Siri and it knows what's in my text message and it'll go find something on the web and right? Yeah. That's what we all want. So what will the new Siri actually do? So according to Bloomberg, this Apple's current 150 billion parameter model was just stuck at simple pattern matching, right? So the new one, think of the difference between the old school transformer models versus the models that reason. So the new 1.2 trillion parameter Gemini version of Siri, it's more understanding and reasoning. You know, it is an 8-fold increase in complexity in what Siri will hopefully be more useful in doing. So Gemini is going to handle two distinct parts of kind of this under the hood, how Siri works, the planner and the summarizer. So Gemini will take over Siri's heavy lifting planner and summarizer functions. So that will hopefully, right, asterisk, hopefully enable complex multi-step multi-app commands for the first time. And this is honestly the holy grail of what people have wanted. And I think what a lot of the Google-powered phones are already starting to see, right? A truly useful and proactive mobile assistant, edge AI at its finest, right? That can use and understand and complex, or sorry, can understand complex data and plan ahead. So, you know, as an example, you know, maybe this is like one example I actually thought of because I do this a lot. So, you know, companies you know, will advertise with us and, you know, have our show go out to their conference, right? So as an example, we had a good partnership like this with IBM. So I was walking IBM's show floor and I was seeing all these, you know, booths out there, just kind of AI startups that I'd never heard of, right? I didn't have hardly any time, so I'd go through, snap pictures with my iPhone, and then I have to go back later and upload them all into ChatGPT or something like that and, you know, do some research. So in the future, right, Siri, you might be able to say something like, hey Siri, you know, go find the last 10 photos I took from the IBM conference. Look those companies up and put their name, website, and revenue in a spreadsheet or something like that, right? Maybe it won't be able to do the spreadsheet part, but you get the idea, right? Or you say, you know, put it in my notes, put it in my Apple Notes, right? The idea of something like image recognition, you know, computer vision, looking things up, planning, summarizing, right? So it could be actual useful AI that completes multi-step processes on your phone. So did this put Apple into the game, or is Apple out of the AI race? Well, I'll say this, because of this, Apple isn't technically out of the AI game yet, but it does view this Gemini Siri integration as a short-term fix or a bridge. But this is kind of classic Apple strategy, right? Like using Intel chips before they had their own Apple chips, right? The M series chips. And hey, on that side, they got it figured out. So who am I to say that maybe their next models won't be good. I'll say this though. I just don't think it's possible because here's the thing. Apple was paying for the engineers. They were paying for the team and they weren't being arrogant when they were working on their M chips, their M series chips, right? Their Apple Silicon. Uh, they were heads down grinding, investing the money and they're not doing that and they haven't been doing that with AI. Um, even though they're already reportedly developing their own 1 trillion parameter model. So yeah, Apple's plan is not to have this be a long-term fix, right? Maybe a year, 2 years max, but then they're hoping to roll out their own model. But my thing is this, it's not how AI works. This is not how large language models work, right? Unless you are distilling, let me put this out there because everyone's going to be like, oh, well, Jordan, look at the, you know, look at all these, you know, open source Chinese models. Um, guess what? None of them came out. None of those models came out. The impressive ones anyways, the ones that were quote unquote impressive on paper until the big players, Google and OpenAI started releasing reasoning models that you can go look at the reasoning trace, the chain of thought, right? So it's reportedly a lot of these new quote unquote open source models, these models that come from nowhere have been distilled, right? CEO Sam Altman. Said it without saying it. He's like, yeah, they're distilled, right? They essentially steal, take, borrow from our work, right? I don't think Apple is going to be able to catch up, right? You don't just say like, yeah, we're gonna, you know, even though they've been reportedly spending millions of dollars a day trying to develop large language models since 2023, according to other reports, you don't just say like, yeah, we're gonna have a trillion plus parameter model for 2026 or 2027 and it's gonna be a frontier. No, No, you, best case scenario is if they get something by 2027, it's where we were in December of 2024. Best, best case, right? You might get, uh, uh, an, an O3 level model by 2027, Apple. But guess what? By then we're going to have an O6 level, right? Whatever the GPT equivalent of the O3, O4 models from OpenAI, right? You're going to be multiple years behind. So good luck with that. So I will say, are they out of the game right now? Technically, in the long run, if this is their plan, yes, they are. They're not even in the game. They're in the B game, right? This is not going to be the same thing as the Intel to Apple Silicon M chips. It's not the same thing. It's not how it works. And ultimately, I went through the 3 reasons, right? The data, the arrogance. But all this is right now, Apple is just buying for a billion dollars. They're buying time to try to stay relevant, or they're hoping the AI hype is going to die down. Those things aren't going to happen. So I'll tell you this, Apple is either going to have to continue to pay the piper. They're going to have to continue to pay Google, OpenAI, Claude, um, or they're going to suffer. Their bottom line is going to suffer. They're going to fall to that number 5 spot or maybe below, right? Maybe in 2028, 2029, Apple is not even a top 5, uh, company by market cap in the US. It's extremely possible. All right, so there you have it. You have the details, you have the facts, stats, and my hot takes here on this Hot Take Tuesday on Apple's $1 billion bailout and why Siri needs Gemini's AI brains. All right. I hope today's show was helpful. If it is, let me know about it, but let other people know about it. All right. We do a lot to bring you guys the most up-to-date, unbiased, real information. Yeah, we do the Hot Take Tuesdays, but on Mondays we bring you the AI news that matters, uh, Wednesdays. So make sure to join us tomorrow. For Putting AI to Work on Wednesdays and then bringing— we actually got a ton of great interviews lined up for the rest of November and December to really finish off the year strong. So I hope this is helpful, but tell someone about it if it is. If you're listening on the podcast, I appreciate it. Make sure, especially if you're on Spotify, click that follow button, right? Well, no matter what platform you're using, click that follow button. But if you could leave us a rating on those platforms, that helps. All right. Thank you for tuning in. Make sure if you haven't already, go to youreverydayai.com, sign up for the free daily newsletter. Thank you for tuning in. We'll see you back tomorrow and every day for more Everyday AI. Thanks, y'all.

And that's a wrap for today's edition of Everyday AI. Thanks for joining us. If you enjoyed this episode, please subscribe and leave us a rating. It helps keep us going. For a little more AI magic, visit youreverydayai.com and sign up to our daily newsletter so you don't get left behind. Go break some barriers and we'll see you next time.