（预告）OpenAI 的后续冒险、Apple 的商业秘密诉讼、大型机问答、Meta、夏令时

Hello, and welcome to a free preview of Sharp Tech. Hello, and welcome back to another episode of Sharp Tech. I'm Andrew Sharp, and on the other line, Ben, how you doing?

Feeling incredibly smug.

Okay.

Smug and proud of myself right now. You've made my day. No, you haven't made my day; you've made my year.

How so?

You are on your traditional summer trip somewhere in New England. I don't want to know. Beautiful setting, great cabin, great video background, and your internet is absolute garbage. We are on the fifth take here trying to record this podcast. I think we're on your cellular hotspot right now. And do you know why your internet is garbage right now?

Because I do not have Starlink, that's why.

Because you're on cable internet.

Mm-hmm.

Almost certainly, what is happening is your neighborhood is saturated and you're just randomly getting dropped. All the things that happen to me. And do you know what's not happening to me on any conversation for the last 9 months?

Oh, boy. Living the fiber dream out there.

Dude, I'm just saying, it was ridiculous that I spent the money to get this done. But every time someone else suffers, I just feel great about myself. So, thank you. I appreciate that retroactive justification for me spending ridiculous money to get fiber into my house.

Absolutely. Look, I'm glad that our collective pain as a podcast team—as I sit here squirming with the worst effing internet on the face of the earth—at least validates your individual choices out there in Madison. I'm ready to hop on a flight to get back to my fiber internet in Washington, D.C. But here we are, and we have a lot to get to.

But yes, we should look into backup Starlink for you as you travel going forward. I'm telling you, greatest technology ever. Anyhow, I do have to know—

It's never sounded better to me, just for the record.

Before we begin, I'd like to formally recant a take.

That's right.

What is this? Is this—

Taking it on the chin left and right here. No, look, a couple of weeks ago on the podcast, you asked me whether OpenAI had been overtaken by Anthropic in the tech takeability rankings, and in a regrettable moment, I said that, yes, they probably have. Anthropic is the new number one for tech takeability, at war with the U.S. government every other week. What an idiotic thing for me to say, or even think—to consider the possibility that anyone in the history of the world would ever unseat OpenAI in the tech takeability rankings.

I regret all of it. I apologize to Sam, to the audience, to everyone. I regret the error.

I regret your apology. No, this is what you should have said.

You should have said, “You know, Ben, just like you are proud of your fiber internet, I am proud of the role I played in OpenAI's recent news. There was a moment where they were just not making enough news. To motivate them, to give them a kick in the rear end, I said that Anthropic was more takeable than OpenAI.”

Did I believe that? Of course I didn't believe that. No one will ever surpass OpenAI. But they needed some motivation.

I was coaching.

I delivered it, and they came through. So, good job, Sam. I'm glad you take coaching. You are definitely number one. I never questioned you.

Filled with the fire, ready to go get sued by Apple and be great again. Here we are. I love it, yes.

Anyhow, where do we start? Do we start with Apple?

1. OpenAI Enters Hardware

Well, no, we're not going to start with Apple. We're going to start with the competition that Apple is worried about. Obviously, this is not the only OpenAI hardware news of the past 7 days, but I want to start with a headline from Mark Gurman. He wrote, “OpenAI's first device will be a mobile, screenless speaker built as an AI companion.”

He writes at Bloomberg, “OpenAI's much-anticipated push into consumer devices is slated to begin with a mobile, screen-free smart speaker designed to be a new type of home computer for the AI era. The product, still under development, is meant to serve as a human-like AI companion that lives in the home. It will help control smart home appliances, play media, answer questions, respond to messages, and tap into the range of capabilities offered by OpenAI's ChatGPT.”

Ben, what do you think of that specific idea as an initial foray into the hardware market for OpenAI?

Well, I don't know. You wanted to put this number one. You said you were excited about it, so I want you to go first before I give my reaction.

I'm excited because I'm excited about all of OpenAI's hardware. I feel like we've been in a sunken place in the hardware space for the last 5 to 10 years. There hasn't been much new. And if I'm being 100% honest, my experience with the Vision Pro, which was a product I was deeply skeptical of when it hit the market and have come to love, has just made me excited to try something new.

And so I like this idea because it's going to ship, and they're actually moving forward with the hardware dreams, which may ultimately be crazy but also may be more useful than the description makes it sound as I read that story in Bloomberg. So I'm just excited that this is apparently going to be real sometime in the near future.

2. AI Makes Stuff Fun Again

Well, I'll tell you what I'm excited about. There's a bit about AI in general that makes me feel like a kid in the '80s or '90s. Just the bit about messing around and playing with stuff.

I can imagine my 17-year-old self. I was talking to someone—I know lots of people, probably lots of people listening to this, who first got online in the 1990s and were like, “Oh, I know what I should do. I should start a small business helping small companies get websites online. I could build websites. I bet I could make a bunch of money doing this.”

In the Strategy Interviews, it's come up with at least 3 or 4 people.

Oh, sure. Yeah.

I feel like right now I want to go build AI apps for people. I could go out and solve your problems. Software is amazing. It could do all these sorts of things. It's just this idea, this excitement, and what's the best way to configure your setup and do all these sorts of things? All of this—it's so refreshing. Stuff that's fun again.

And I think that sort of speaks to your point a bit: stuff that's fun again. Well, there's one other thing from the '80s and '90s that this news reminded me of.

Oh, boy. What's that?

The boombox. Back in the day, you used to carry around the boombox. You had your big D batteries or whatever filling it up, which would last for 4 hours. You would carry it around not just to listen to music, but also when your new favorite song came on the radio, you had to run over and hit record for your mixtape.

Oh, my God. This is just 5 years before my time. I kind of missed the boombox era.

Oh, you missed out. This was before the initial wave of MP3s and all this. I've talked about this before. People would have directories of FTP servers where they would list their address, and they'd have a list of all the songs they had. You would just copy this IP address. You would go to someone's computer and download songs off of their computer. Absolutely insane.

Yeah.

This is the world we lived in before, when the internet was awesome and we didn't have to worry about security. That's why we had to worry about security a lot very quickly. This is how people lived. This was even before Napster—then Napster comes along and makes it easier.

Mm-hmm.

All these things. But before then, how did you get songs that you liked? Yes, you literally would carry a boombox—

Tape the radio.

—around with a tape in there. When a song came on that you liked, you would run, hit record, and then you would have a 60-minute tape. Every time they had a new tape format that was extended, you could have a 90-minute tape instead of a 60-minute one.

Mm-hmm.

And you would have the voice-over of the DJ over the beginning of the song, which meant you were missing 15 seconds because you were running the boombox. I can't believe you never experienced this. Of course, you had the tape player in your car. That's what it was for.

That's right.

So this is the first thing I thought of: this idea that you're carrying around a device that's battery-powered and is sort of with you. I also love it. Any longtime readers of Strategy are probably going to know where I'm going with this, which is the original Amazon Echo and Alexa. That came out 3 or 4 months after the Fire Phone came out.

Mm-hmm.

The Fire Phone is arguably the single worst product I've ever used in my life.

It was so bad. It wasn't just misguided from a strategic perspective; it was horrible to use. Everything about it was bad. It was one of the worst products ever.

So this was before my time. What were they even trying to do with the Fire Phone? How were they planning to differentiate?

I was just trying to remember. They had some weird new input methods. I can't even remember it; it's mostly wiped from my memory.

But it was just awful.

It's cauterized.

Yeah.

It was so bad, of course, they didn't sell any. The reviews were terrible. They didn't sell any, and it didn't make any strategic sense for Amazon to be making a phone. Then, 3 months later, they came out with the Echo and Alexa, which everyone, sort of primed—no pun intended—by the Fire Phone, was ready to reject and say, “This is a terrible idea.” I'm proud to say I was like, “Actually, this is pretty great.”

You were in.

This is everything. It makes so much more sense in a way the Fire Phone didn't. One of the ways in which it made sense is that there's one environment in everyone's life where, number 1, you don't necessarily have your phone with you because it's charging. Number 2, you're not going to feel weird talking to it because no one's around. That is the home.

Yeah.

I think this idea of something to talk to and interact with in the home makes sense. It has continually made sense with products that frankly sucked, right? You use an old-school Alexa now in a world of LLMs, and it's actually incredible they sold as many as they did, given how crappy it is. By the way, I had a killer Alexa use case this week.

What's that?

I was getting my hair cut. I just go to the local Great Clips or whatever, which is, by the way, technology I love. You can sign in on the app, put yourself on the waiting list, and get a notification when you need to run over there. It's literally across the street. You get a notification, walk over, sit down, get your hair cut, and you're out in 15 minutes. Amazing. Love it.

There you go. No more need for magazines in the waiting room. Just put yourself on the waiting list.

That's right. You don't have to call and make an appointment and all those sorts of things. I need a haircut, so I just go in the app, hit a button. Love it.

Great.

Anyhow, I'm in the chair, and she's cutting my hair. She's like, “Alexa, skip this song,” and she's complaining to the other hairdresser: “We've heard that song 57 times.” I'm sitting there thinking, “That is one of the best Alexa use cases I've ever seen.” Her hands are busy cutting my hair. The poor people in a hair salon have to listen to the same songs over and over and over again.

Over and over. Yeah.

Yeah. She yells at Alexa, skips the song, and goes to the new one. Great. Loved it. Great use case. Anyhow, this idea of talking in your home makes a lot of sense. What you get with a speaker is really good microphones and a good speaker. But what's interesting is that, for all these devices—the Alexa, Google Home, and these others—the number 1 use case has ended up being music. The hairdresser, right?

That's what I was going to say. The hairdresser that you're describing is certainly how my family has used Alexa. My kids love—or my son, at least, my 3-year-old son loves using Alexa because he can talk to it.

I'm going to start giving him tips on songs that he should request.

It's actually fairly annoying. So, look, if that's the baseline, then it is useful, but I'm curious to see what they build on top of that as well.

3. The Microphones Matter

I think Steven Sinofsky had a tweet that was really insightful. The most important component for most people in the way these products were actually experienced was the speaker. The most important component for this device is going to be the microphones.

Okay.

Because it's a device that is meant to be interacted with. It's something that listens to you, and that sounds weird and creepy, right? Your speaker listening to you. But in this case, everything's weird and creepy until it's so awesome and great that you realize, “How could I live without this?” Right?

Totally. This is where I go to the Vision Pro experience. It's like, “Why would I ever do that?” Now I do that 5 nights a week, so let's see what it turns into here.

4. Portable AI Gets Interesting

Right. I like the idea of starting here. A phone is too much; it's too aggressive. This is something that's there when you want it, and it's not there when you don't want it. I also like the idea of porting it around. The ideal future is that you can just talk to your house.

By the way, that's coming for the home inventory app. I have some of the components in place.

Hands on top of it.

Yeah.

Don't worry, everybody.

This idea that you can just have this sort of ambient bit—you can fake it till you make it as far as an ambient AI if you just carry the AI around with you—I think it makes a ton of sense. I was just talking to Gruber about this, and we were debating whether this is similar to a HomePod or not. To me, the fact that it doesn't plug in, that it's battery-powered, is actually the most essential feature.

Why?

It's there when you want it because you can bring it with you. It's there when you want it, and it's not there when you don't. I just think it makes all kinds of sense to start. It's possible it doesn't work, and that's fine; you haven't gone all the way down the road of trying to build a phone and get an ecosystem and all those sorts of things. I think it's a great V1 project to start with, and I'm excited about it.

Yeah. I missed the note about it being battery-powered and not plugged into the wall. That makes me more excited as well, because I was looking at all this reporting—

That's the whole thing that made me think of a boombox: the fact that you can carry it around. That's as critical.

Well, then I was thinking about it, and I was like, “If this is really useful and something I enjoy using, am I going to have to buy 5 versions of this for different rooms throughout my house?” But no, you could pick it up and carry it with you as you go.

And this gives a signal if people want a phone. If suddenly you realize people are taking this with them everywhere, bringing it in their car, then you get a real, useful signal from usage about whether there's something there—whether there's something that people are interested in. I think the battery-power component is critical. It's a critical distinction from the Alexa devices of the world and the HomePods of the world.

Mm-hmm.

Again, it only would work in your house, but yeah, I think it makes a lot of sense. I'm excited about it.

5. OpenAI's Hardware Business

Let's see what it turns into. As far as a hardware business generally, we got this question from Pfizer Frank: “So what does a successful hardware product look like for OpenAI? For example, how important is it to sell X billions in product over the next 5 years versus establishing OpenAI as a product OS versus using this as another data-gathering venture?” Any thoughts on that? How are you thinking about what this should be and what the goal should be for OpenAI over the next couple of years?

It's a really good question. Building a hardware business is really hard. Where are they going to sell these? Actually, that's another thing I like about this product a lot, by the way: it's the sort of product that's going to be a great present.

Oh, yeah.

It's easy. There's something about it that's easy in a way that a phone—and getting cellular service and all those sorts of things—isn't. You're going to bring it home, connect it to your Wi-Fi, and it's just going to work. You're going to sign in with OpenAI, and all this will be done with your phone.

And it's the sort of thing that you won't necessarily want to buy for yourself, but if someone buys it for you, you'll be psyched to just play with it. That's why it's a fantastic present. That's a great call.

Right. The problem with the phone is that you're making a bid to replace the phone, right?

Mm-hmm.

No one carries 2 phones. This is an add-on device. Now, that said, this is a great device in the context of doing hardware. Should they even be doing hardware at all? If they do hardware, what should the business model be?

And this—you go back to my critique of Meta, right? You're a service that should be serving everyone. Hardware is fundamentally constraining your market because you're not going to sell hardware to everyone. So what are we doing here? You should hope someone else makes the hardware, and you could be a service sitting on top of it. Doesn't that apply to OpenAI? And yeah, you can make that argument.

Has this whole effort been worth it just from the context of pissing off Apple?

This actually ties into one of the concerns I've had about this hardware effort even before this lawsuit was filed: wouldn't it be better to be AI on all Apple devices—

Mm-hmm.

—instead of trying to replace the iPhone?

You're presenting a competitive threat that they're going to keep at arm's length at best. Obviously, they're currently in litigation, but—

Right. Even before we got there—

—you should be partnering with them on the software side as much as—

Yeah.

—you possibly can.

And I think the pushback there is a couple of things. Number 1, if you truly believe that AI presents the opportunity for a new paradigm—

Mm-hmm.

—Apple's not going to deliver that paradigm. In fact, they're going to be an obstacle in delivering that paradigm. So there's a degree of conviction and belief that Apple's not going to do this.

Yeah.

The flip side, or the implication of that, is there's an opportunity to do this—

Mm-hmm.

—because everyone in the market is not going to get that. And maybe the partnership with Apple was a spur in this regard in realizing, look, Apple—

Yeah.

—it's great we have this partnership with Apple. We're only ever stuck behind Siri, and people purposely invoke us.

Yeah.

If Apple gave any indication or signal that they would let us be the AI on their devices, of course we wouldn't build our own devices because that is obviously the best opportunity. If there was an opportunity in the iPhone to say, "Replace Siri with OpenAI," that's the best.

Maybe we can live there, sure.

That's not going to happen. The fact that that's not going to happen means Apple is going to double down and bet on themselves. That also means there's an opportunity. Now, there's also the Anthropic angle, which is we're not going to win the whole market anyway.

Mm-hmm.

But then there's a number 3 business model. How are these companies ultimately going to monetize? I've been making the case for a while that the economic imperative for Anthropic and OpenAI is to own the consumer touchpoints.

Yeah.

This is why, at the end of the day, they are going to go after software. That's why they are going to go after Microsoft. That's why they're going after this. And in that context, that's why they're going to go after Apple.

To the extent that models are commoditized, your model needs to be integrated with the end-user experience such that they can't swap it out even if they wanted to.

Mm.

This is a tension underlying this. This is the story in tech.

Yeah.

This is why Satya Nadella is writing essays every week about the importance of not giving your data away to these other companies. There is a brewing fight for consumer touchpoints, and these companies are coming for it. In that sense, this makes sense.

Now, do you price it at cost so you get—

Right.

—people on your service? Do you price it at a premium so you can actually make some money and try to be the next Apple? I go back and forth on that. I think you could probably start by pricing at a premium to see if you have a market and play it out over time.

But that is a critical decision. Are we trying to be Apple? Are we trying to be Google? That's going to be an important question.

Right.

And how do you take any sort of market share initially if you're charging a premium for a service that people—

Well, because the initial cohort of people buying this is not going to care about price.

Just nerds like you and me.

It's really about you and me and everyone in tech, however many they can manage to sell in the first year. This is all within reason, right? I'm assuming this is going to cost a few hundred dollars as opposed to—

Okay.

—a few thousand dollars like the Vision Pro.

Yeah.

But the technology, I don't think, sounds too complicated. It's really mics and speakers, right?

All right, 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.