（预览）视频持续战胜文字、Google与YouTube的机会盛宴、关于AI泡沫的6个问题

Ben, how are you doing?

I'm doing okay, Andrew. How are you?

Do you have an iPhone review ready for me today?

I do not. Mine is actually arriving today. I am not one to sleep through my alarm. The first time I slept through my alarm in years was literally on iPhone order day. I was 30 minutes late to order, and thus a week late to receive it.

A full week late here. I mean, look, you had no smart glasses last week—no iPhone this week. Ben, I'm concerned. Are you sure that you're the tech guy on the podcast here?

Hey, I'm here to examine the space between Wall Street results and products. The products are nice, but not necessary.

Not a product reviewer.

Okay, fair enough. Well, I have the new iPhone, and just a 30-second review on my end.

Yeah, no one cares about my review. We need yours.

1. The iPhone 17 Upgrade

I love the new iPhone. It has tremendous battery life. I'm charging my phone once every day and a half at this point, as opposed to once every 12 hours with my rickety 15 Pro.

Wait, wait, wait, wait.

So that's a massive upgrade.

So you only charge it when you need to?

Correct.

So, like, you charge it one night, then the next night you don't charge it. You just wait till, like, halfway through the f— This is insane behavior. This is insane behavior.

When you lay it out like that, no. When you lay it out like that, I do sound like a psychopath. That's not going to be my standard operating procedure going forward. But mainly in this first-week, 10-day period, I'm just curious to see how long the battery lasts if I don't charge it at night.

I'm sure I'll settle into a rhythm where I just plug it in every night and forget about it, and that's going to work great. It wasn't working great with my old phone. I'd have to recharge during the day. No longer a problem now that I'm living the 17 Pro life, so thank you, Apple.

I would also note that the camera is much better than the previous cameras. Noticeable improvements.

What are you upgrading from?

The 15 Pro.

Okay.

So it's not like I was on a 13 or 12.

That's reasonable.

I mean, they were really good cameras, but the 17 is a noticeable improvement, and so I've really enjoyed that. My 7-month-old daughter, Rose, is a big fan of Apple's new selfie technology, so that's working great.

Wait, she's operating the phone?

Well, no, but she loves sitting there while I operate it for her because the selfies work better. I don't know exactly what Apple did with the selfies. You could probably tell me better.

Yeah, do you want me to?

Yeah, please let me know.

And it's actually funny because Apple was late to the camera in general. The first iPhone didn't even do video. It was sort of an afterthought. It's funny because it became such a driver for the phone in general.

But then the selfie camera itself sort of languished for ages, and this year they're like, “Oh yeah, this is actually probably a really important thing.” Your selfie camera was aligned with the phone. And this—

Mm-hmm.

—transformed media, right? Suddenly we were getting vertical videos. That's what you got naturally with the front camera, and that's fine. That's a thing in culture now. We see vertical videos all the time. They're actually better to consume on the phone.

Yeah.

But sometimes you want to get a broad shot, right? You want to get things across. If you use a selfie camera, then you would normally have to crop it, or flip the phone and hold it awkwardly. It's just easier to hold a phone up and down.

So what they did was basically use a square sensor. You can hold the phone any way you want, and you just press a button—

And it adjusts—

—to flip between portrait and landscape, and it's going to be the highest possible quality every time, no matter what you do. So they—

I'm here to testify. It works great in practice, even as you're juggling a baby in your arms. So great job, Apple.

The only other note I have is that our listeners were exactly right. Liquid Glass is not only not horrible, but pretty elegant and pleasant to use. I've got my guard up for some sort of bugs or, I don't know, maybe the experience starts to get crappy a couple of weeks from now, but I've really enjoyed my early interactions with Liquid Glass. So across the board—

Here's a question.

Some wins.

No, this is the big question. Have you picked up your old phone at all since you used the new one?

Not once.

Well, you should, because if a user interface change is working, then you pick up the old one and you're like, “This feels so ancient. I can't believe this is how it used to work.”

So you have to do that test after this podcast. Let us know.

2. Apple Returns To Its Strengths

Does the Tim Cook apology need to be as loud as the disrespect was? What do you think? A lot of people like these new iPhones.

Well, I mean, this is what I picked up on from Apple at WWDC when they announced Liquid Glass, which is—

Mm-hmm.

—this is a company—I think I called it Apple Retreats. Sometimes retreats are good and necessary and appropriate. And they—

Yep.

—this year was clearly a year when they said, “We're going to do what we do best. We're going to focus on the operating system and the user interface, and we're going to focus on really great new phones.”

The AI questions are all still wide open, all still TBD, but it's good to sort of reset and be like, “Yeah, we're still good at what we do.” And to the extent that—

That's the key question: Are they still good at what they do? That's what I was asking back at WWDC. In the wake of that, have the core competencies eroded?

I don't want to give Apple too much credit in the midst of my iPhone 17 euphoria, because part of the reason I'm so happy is because I had come to hate the 15 Pro I had. But I really do feel like they nailed these upgrades.

So I'm back in, a satisfied Apple customer, and I can't wait to hear what you think of the iPhone 17 when your—

Oh, I already know mine's better than yours.

—Max arrives.

Well, yeah, mine is better than yours. I can already tell you that right now. Number 1, it's a Max.

Mm-hmm.

If you're worried about battery life, you gotta be rocking the Max.

You know what I'm worried about with the Max? I already spend too much time on my phone. I actually don't want a bigger screen that's going to be an even better experience, because I don't think that would be healthy for my touch-grass habits. That's my concern.

Yeah, it's funny. I held onto the regular Pro for ages and ages and ages. I finally switched last year, and then last night I was picking up my phone, and I'm like, “Wait, is this my old phone? It's so small.” I'm like, “Oh, no, this is the Max.”

I'm so fully acclimated. I'm like, “I want more. Give me a bigger one.” I can't believe I used to live that way. I do still—

The Max of Max, yeah.

—have problems typing. I don't know if you ever noticed my texting; there's random line breaks because my thumb is still not fully acclimated, even after a year.

But also, my phone will be better because it will be orange. So, big L by you, but what are you going to do?

Well, I just hope it gets there eventually, you know? Already a week late. Who knows when the iPhone Max will eventually arrive on this continent.

Today between 1:00 and 5:00.

Okay. Well, speaking of today, we've got some bubble talk on deck in the middle, but we will begin today with the blob and the giant hiding in plain sight: YouTube and Google.

You had some interesting follow-ups earlier this week to your interview last week and the conversation we had on the show. So I'm going to read from your article on Tuesday. The headline was “The YouTube Tip of the Google Spear,” and in that piece you write, “Here's the trillion-dollar question.”

No, that's not what I wrote.

What is—

I said, “What is the million—nope, strike out—billion—nope, strike out—trillion-dollar question?” Let's get some accurate copy-and-pasting here.

It's true. It's true.

When I recorded it—

Well, I thought to myself, I don't know if reading strikethrough text is really going to come through on the podcast medium. But I'm glad that you jumped in—

I can do a reproduction of the podcast recording of the article. Or—oh, I did so much better the first time. Here's the million—

Billion—no, trillion-dollar question. You know, you just gotta catch yourself and then say it again. That is the strikethrough in real life.

Well, you could tell. I consume these via text, not podcast.

As they are meant to be consumed—

That’s right.

Anyhow, continue.

I’m a proud wordcel here on the pod. Here’s the million-, billion-, trillion-dollar question.

No, no. You have to catch yourself and say the next one. You’re saying the wrong thing. Anyhow, continue.

3. Video Beats Text

Look, we’ll workshop it. The million-, trillion-, billion-dollar question is: What is going to matter more in the long run, text or video?

The history of social networking suggests that video is, in the long run, much more compelling to many more people. To put it another way, the things that people in tech and media are interested in have not historically been aligned with what actually makes for the largest service or makes the most money. People like me, or those reading me, care about text and ideas. The services that matter specialize in video and entertainment.

So, Ben, let’s start with that idea, because I don’t think we talk about it enough on this show. Do you want to expound on the dynamics you’re describing there and why video might matter more than text in the long run?

Well, what did you just say? How do you consume Stratechery?

Via newsletter, text.

The written word, yes.

That’s right.

It’s kind of an interesting dynamic that I picked up on years ago, in the context of Facebook versus Twitter. The media loves Twitter, and people in tech love Twitter. The whole lure of Twitter—and I remember being at business school and trying to recruit people to use Twitter. I’m like, “You have to be on here. How can you be interested in technology and not be on here?” I did a whole thing as part of the tech club, teaching people how to set up their feed and follow XYZ.

I’m like, “Yeah, it’s really hard to get started, but once you get it figured out, then it’s super great.” That was always my critique of Twitter: It’s too hard to get started, too hard to get your feed optimized. I think there’s a bit of that that certainly is true. The feed is much more algorithmic now, which I think is more approachable for people, generally speaking, even if some people who like the old Twitter and their perfectly curated feed find it annoying or frustrating.

Mm-hmm.

But I think that’s only part of it. I think the reality is that the reason why I love Twitter so much, and I presume the reason why you love it so much, is that it is so dense and packed with—

Information, yeah.

Yeah, it’s like the information from all sorts of things all together, and you can read it, and you can feel super plugged in. Even today, I still rationalize the amount of time I spend on Twitter because, on a daily basis, I probably see at least 1 tweet a day, and probably more, that actually is beneficial, that affects my thinking, that makes me think about something different or triggers a thought process that leads to an idea. There’s nothing better for doing that, even today.

Yeah.

I think the reality is—and this is kind of hard to talk about because, of course, us being wordcels, to your point, highly value this trait, so this sounds insulting—most people just aren’t that interested in doing that.

This goes back to the ongoing joke and truth about me and my TV consumption, right? Maybe this is part of how I was raised. My parents were super strict, and I couldn’t watch TV except for sports. That’s why I love all these sports on TV, and I really don’t watch anything else.

I read all the time. I got grounded as a kid by not being allowed to read, and it was effective. It’s not like I associated reading with punishment. It’s because I read 10 books a week, and I’m not even exaggerating. So to me, Twitter just slots right into that.

Yeah.

And the idea of sitting down—I can barely make it through. I never finished Succession. It’s like, “This feels like work. Not doing it.” I got a season and a half in, much less all the slop that’s on all these other streaming services.

But why do I still write about it? Why do I think Netflix is compelling? This is a bit where I’m not writing from personal experience. I’m writing from looking at the data. We know about the amount of TV that’s watched. We know about the habits, and thinking about things like advertising and all those sorts of bits and pieces. It goes into why I feel okay talking about the new iPhone even though I don’t have one, because I’m thinking about it from a structural perspective.

Mm-hmm.

Certainly, it’s useful and beneficial, and sometimes I do have blind spots about stuff that I don’t watch or because I missed something from personal experience. I acknowledge that’s the case. But the fact of the matter is, according to the numbers, I’m the outlier, not everyone else.

Yeah.

And you saw this in the late 2000s with Facebook versus Twitter. There were a lot of people in tech, on Twitter of course, who couldn’t figure out why Facebook was so popular relative to Twitter.

I remember, I signed up for Facebook right when it came out, as well as Twitter, and never really touched it for ages. I remember being in Taiwan. Even then, even though I wasn’t thinking in Stratechery terms specifically and had no idea I’d write an article about this, it just struck me how other people—I was an English teacher back then—and all the other English teachers were obsessed with Facebook. They were on it all the time.

Mm-hmm.

And I’m like, “Twitter is so much better. It’s so much more interesting.” Again, I see that and recognize I’m the weirdo here, not all these other people.

Right.

What was Facebook back then? Facebook pretty quickly became photos. People preferred photos over text. The text was interesting. When Facebook was just text, they were onto it, but then photos, and then video comes along.

You see the rise of TikTok really blindsiding Facebook, not just by virtue of being video, but also video from anywhere. And actually, this was probably one of my best calls, even though I only got it 70% right, back in 2015: Facebook has to go beyond the social network. Social—actually, social networking—is not the end game. It is a way to get scale so that you can just be pure entertainment. And entertainment—for me, Twitter is entertainment.

I was going to say.

I’m a weirdo. For everyone else—

We find Twitter entertaining and enriching, but most people don’t have the energy.

Right. They say you gotta pray that you can do what you love. Probably the closest bit of doing what I love in my life is the fact that I get to justify reading Twitter 24/7 because I can tell myself it’s for my job.

Yeah. And it gets dangerous. It can get abused. I’m speaking from personal experience there. But yeah, that’s the reason it resonates with me.

We mentioned this over the summer. Even though Steve Jobs liked to think of Apple products as productivity enhancers and bicycles for the mind, the iPhone is ultimately a consumption device more than a productivity device, and that’s what made it the most successful consumer electronics product in world history.

The reason I say we don’t talk about it enough is because of the shift to video. It’s undeniable that the shift is already happening. Like, everyone—

Oh, it already happened. I mean, TikTok is dominant.

In the world—

Reels is all video.

—is reading less than ever. Yeah. And consuming more video than at any point in human history. There are overwhelming social implications for that shift, and then business implications. But it’s not clear that it’s fully reflected in how people are forecasting what the world might look like in an AI universe, which is what your piece was focused on this week with YouTube.

Well, you just zoom out, and you’re trying to make forecasts not just for today, tomorrow, or next year, even though you want to dismiss products forever based on how they are right now.

Oh, boy.

Yeah.

We’re going to address that later in the show, don’t worry.

All right. Fair enough. But if this overall concept is right—and to be clear, for this article, and I probably should’ve been clearer about this, I’m talking about consumer AI: what people use, what resonates, what makes money.

If you look back to the history of mass media, newspapers to magazines to TV, if you look back to the history of user-generated media, which is probably a better name than social networking because that actually captures TikTok, which really isn’t a social network at all, in all cases, text is there, it’s important, but it’s superseded by images. And they’re both massively superseded by video.

That is actually the dominant one that matters most, that takes the most share. So if you apply that to AI—which you might not, but let's apply it to AI just for our amusement here—we have text. ChatGPT dominates text. I think there's a good chance it will always be the go-to text source.

Yeah.

Text usually always comes first because text is easiest. That's just been the case for all the mediums as well: text was easier in the analog world. The printing press came years before the camera, which came before video. It was easier on the internet to send text, so we got text social networking before we got images, before we got video, and we needed bandwidth—we needed 4G at a minimum—all these sorts of pieces. And with AI, we got text first, and then we have images. What was the last real viral moment for ChatGPT?

Studio Ghibli.

Well, we can talk about pronunciation later in the show as well.

Great.

People like images.

4. Gemini Makes Images Viral

Well, and speaking of the images issue, this again speaks to my life as a wordcel. I was completely unaware of this until I was prepping for this specific podcast episode, but Google's Gemini app has been number 1 in the App Store for the past couple of weeks. Per CNBC, they launched an updated version of its AI model called Gemini 2.5 Flash Image, which has a new image-editing feature called Nano Banana. The app allows users to blend multiple photos together and use prompts to create such things as digital figurines from photos that a user uploads. The product helped bring in 13 million new users to the Gemini app in 4 days, and according to Josh Woodward, a VP at Google, users have uploaded 5 billion images in less than a month to use the Nano Banana editing feature.

No, the Nano Banana thing is incredible. You should try it. It works incredibly well, it's really easy to use, and I think it's a pretty big deal. You see apps pop up to number 1 in the App Store for a day or 2 because the App Store rankings have a very short half-life. You can have an app go viral, it's up there for a day or 2, then it tumbles down the charts.

Mm-hmm.

The fact that Gemini's actually been number 1 for weeks, basically since this came out, and there keeps being a viral meme—to this point, it's not just about the digital figurine. It's a new idea, everyone does it, and it goes around. There appears to be some sort of sticking power, and part of this is that Google is really good at this.

Yeah.

The Nano Banana capabilities are really good. We saw this a little with Midjourney. Midjourney always ended up stuck, buried in Discord. They probably needed an app sooner than they did to sort of move things along. But there were a lot of people spending a lot of time on there just going through images and creating images. People like images. And this shouldn't be a surprise if we look at the history of media. But that means we're only halfway, because the end state is video.

Mm-hmm.

5. Google Takes The Video Lead

Video is what people like the most. And you might sit here and say, "Oh, but people aren't going to like AI video," to which I would go back 10 years: did you forecast that people would actually be spending way more time on user-generated videos on TikTok that are 30 seconds long than they would be—

Yeah.

—on produced content? Be careful in your presumption around human preferences. Broadly speaking, we know that humans like video. So given that, and again, maybe they won't like AI video, but I think they might. Who's going to win that? Google not only has the channel with YouTube; its video-creation capabilities are amazing. Nano Banana is amazing. There are also some other pretty good image-generation tools. Midjourney is still really, really good, to take one example. OpenAI's image generation already had its viral moment; it's still pretty good.

Yeah.

I think Google's is better, but it's competitive. Video's not competitive. Google's so far ahead—astronomically. OpenAI tried to get ahead of it with its Sora launch last year, and Google—

Mm-hmm.

—just embarrassed them with Veo 2.

I remember, yeah.

And since Veo 3 has come out, it's amazing, and now they've actually productized it. And here's the big deal. This is why I wrote this article even though it's late. Google stock actually has been on a tear for a few months. This is not a Facebook, "I called it at the bottom, I'm going to pat myself on the back for the next 5 years." I am late to this.

The reason I'm late, and the reason I wrote it now, is because the fact that Google has the capabilities has never been in question. DeepMind—that's why Elon Musk and Sam Altman started OpenAI ages ago, because it was clear even a decade ago that Google was the best-positioned to win this—

Mm-hmm.

—and they've gotten refocused, and they're shipping good models, even though they started out so embarrassingly with the whole Bard thing and the launch. That was still one of the craziest keynotes that I've ever seen. A guy went up to do a presentation, and he said, "Let me demo it," and went back to the counter. There was no phone there to do his demo on. It was amazing.

Was this the thing they did in Paris—

Yes, yes.

—trying to respond to OpenAI? Yeah.

It was crazy. People were presenting; they didn't know what was on the slides. It was one of the wildest keynotes I've ever seen. I don't even know if it's on the internet anymore. I should have captured—

Scrubbed from the internet.

Yeah, it is—

Yeah.

—it’s crazy. So they're still behind, but—

But that's where they were.

That's where they were.

They were flat-footed—

So their models have been getting better.

—and people were like, "Okay, how did you let this happen?" But now here we are.

They dropped these image models. They dropped these video models. But the question was, can they actually productize them? Can they actually ship? And what was such a big deal about this YouTube announcement last week was not just that YouTube is the obvious channel, but that they actually shipped it.

Mm-hmm.

And if they've figured out how to productize and ship—not just fun little Veo 3 applications in YouTube—

But—

—which is great.

—but what did they actually ship, though, as far as YouTube is concerned, just to remind people?

Some of this stuff is live, and some of it's not, so fair enough. We can wait, check in in a month or 2, and see how much actually got out the door. But the stuff was there. I think it was far enough along that I'll call them on it if none of this actually sees the light of day. But—

But they are AI tools for creators, essentially, to incorporate AI, is that right?

Well, you have Veo 3 that anyone's going to be able to use. You can use a prompt to make videos, or you can just upload a bunch of clips and it will put the clips all together into a narrated video, and then you don't have to organize it or whatever it might be.

Actually, the one that I think is pretty funny, and I'm actually kind of excited about, is text to song. This might be a good Sharp Tech marketing tool. You could clip a rant or something like that—

Yeah.

—and it will make it into a song.

Terrific.

Which, again, you could see a lot of misses with this, but you could imagine a couple of hits will be pretty spectacular.

No, I think the next anti-monopolist rant is going to be a big hit on YouTube. I can't wait.

Yeah, no, just brace yourself. There's going to be some viral videos that go around of someone taking Trump or something or other and just making a song out of it, and it's going to be—

Mm-hmm.

—very funny. So that's coming. And those are more consumer-y, easy-access bits and pieces.

6. YouTube Monetizes Creator Commerce

The big one that blew my mind was this auto-tagging capability for creators, who feature products. Actually, the highest-end creators make more money off YouTube than on YouTube. You watch MKBHD, and he's doing the Samsung thing. That's where he's making his real money, even more so than the YouTube ads.

That was actually an interesting bit about the presentation. YouTube wants into that business, and they're framing it as, "We're helping you, freeing up time."

"We're going to make this easier."

Yeah, there's a bit from YouTube's perspective which is, "You're making a lot of money on our platform that we're not getting a piece of." But—

You could certainly see, for lower-level creators who don't have the capability to do those brand partnerships, it will be better. You can also see some of those brands going to the high-end creators and saying, “Can you just go through YouTube? We don't want to deal with you individually.” And then suddenly their cut is taken. So there's a little bit of tension there.

Mm-hmm.

But as part of this feature, you can insert a link to your video. That can be one of the ways your monetization can be gated. How many click-throughs did you actually generate? How many people went through? Did you actually drive sales? You get more sophisticated advertising than just, “Did you say the spiel in the middle of the video?”

Yeah.

To do this, you have to upload it to YouTube, and then in the YouTube editor you have to go find the spot, insert the link, and then cut the link off at the right time. It's not that it's hard; it's just annoying, right? It takes a while. You have to go in and do it, and then if you change the video, you have to do it again, and all this sort of thing. And YouTube's like, “Look, no, we can take care of it. Give us the link. We will figure out the right spot, we'll put the link in, and then we're good.”

Mm-hmm.

And let's say at 20 minutes, it's like, “Oh, the AI can see he's talking about a Samsung phone. I have a Samsung link here. I'm going to put the link in the video.” Super basic. Super basic and incredibly profound because—

Because—

We talked about this in the context of Meta a year ago, which is my contention that every pixel on Instagram going forward is going to be monetizable. They will be able to recognize every item in every video and attach a link to it or create some sort of bidding process. The actual business model, we'll see what that sort of looks like.

The reason I frame that in the context of Meta is I've gotten a lot of traction by being pro-Meta for a long time, for sure, and also the sense is Meta's founder-led; Meta can ship. Here we are a year later, Meta's AI is, like, a total mess—

Languishing, yeah.

It's kind of in disarray. Again, I go back to that. My interview with Mark Zuckerberg was so weird this year. It was pre-reset, and I felt there were so many things he didn't talk about, like this, like using AI and VR, which are back to his talking points. A lot of the Meta keynote last week was Zuckerberg talking points, which is great—I love to see it—but the reality is they're kind of a mess—

Yeah.

And Google shipped it. Google—they shipped a very small version that is very pro-creator, but it's a straight line to what's happening.

The big blob—

The AI is identifying a product and inserting a link.

…is moving. Yeah. Well, so to recap, first of all, just loving your evolution here, growing from a little calf out there in the pasture to a big, strong Google bull on the other line.

Google's always had the pieces to do it. The question was, could they ship? And YouTube, at least, is shipping. And then you back up, and it's not just that part of Google is shipping. Arguably the most important part for winning the future of the consumer is shipping. That's what YouTube is.

Well, that's the key. That's what you synthesize with your article. If you look at the history of media, AI video is probably going to trump AI text on the consumer side as far as a market opportunity. Then you look at the advantages Google has, and what we've seen even over the last 2.5 years, Google is going to be much better in AI video, at least for the foreseeable future, than any of its competitors.

Right. And, by the way—

And—

They have all of YouTube video for training data.

Exactly.

Everyone kind of scrapes it, and as everyone is also using YouTube video data, I imagine Google is much more efficient and much more capable of using it by virtue of owning it.

But also, Google has a clearer path to monetizing all of these advantages than any of its AI peers, at least as we sit here in 2025, and that's the other piece that came through.

Well, Meta's the other one. This is the reason I was bullish in the Meta article. I kept saying, “Oh, this also applies to Google. This also applies to Google.” I just sort of gave Meta the benefit of the doubt that they would be better at shipping.

Mm-hmm.

And that was wrong. So go read the Meta AI Abundance article and just search and replace Meta for Google, and all this stuff applies. They have the—

Yeah.

…surfaces to manifest it. They have the business model to manifest it. AI is going to be one of the biggest gifts to advertising ever. It already is in terms of generating the ads themselves. It's huge.

Yes. And I think in the Meta article that you're referencing, you said that Meta has a chance to be the most valuable company in the entire world.

Ah.

So find and replace, and Google has a chance to be the most valuable company in the entire world, for everybody keeping track at home. Speaking of monetizing opportunities, though, I want to read this note from Joshua. He says, “How big of a threat to Amazon is Google increasing the affiliate ad surface in YouTube? AI product recommendations seem like an emerging and relatively underappreciated use case.”

And that is the end of the free preview. If you'd like to hear more from Ben and I, there are links to subscribe in the show notes, or you can also go to sharptech.fm. Either option will get you access to a personalized feed that has all the shows we do every week, plus lots more great content from Stratechery and the Stratechery Plus bundle. Check it out, and if you've got feedback, please email us at email@sharptech.fm.