Google I/O: Day 1 Reactions, Goldman to Lead SpaceX IPO | Diet TBPN

We're both in suits today. I like it. It looks good. Some, we were in UnHerd.

People have been saying.

So they can't tell us apart? Is that what you're saying? That? Oh, okay. Anyway.

But people have been saying I need something like a swear jar.

Okay.

When I don't wear a suit.

Oh, okay. I'd like that.

And so, so yeah, something to consider. It should probably be a pretty big jar.

$20 to the OpenAI nonprofit. That's what you gotta do. No, we were in UnHerd. We, I mean, we can pull up the full post later in the show, but Funny. I mean, it's a good, good analysis of like, you know, what we've taken from ESPN, what works about live streaming. This is from Alice Key. My latest for UnHerd is on why tech shows like TVPN are rerunning the sports media playbook invented by ESPN and why it's working. They're hard to tell apart in their matching suits and floppy haircuts. I don't know if that's good or bad, but we don't always have matching suits. Usually Jordi's the casual one, but yesterday we were both casual. Today we're both in suits.

We do still get the brother thing a lot.

But I think it's a term of endearment. I think it's positive. Anyway, we have to react to Google I/O. Of course, there's a whole bunch of announcements, some really exciting stuff, some stuff that people are having mixed reactions to. We'll take you through it all. But first, we need to watch this video about— Humanoid robot. Ramp? Why is ramp?

No, I'm kidding.

No, I'm crying. Watch this video.

Let's get some sad—

You got to go to the beginning. You're spoiling it. You got to go to the beginning. Doing pretty well. Moving pretty quickly. Little bit of a— oh, catches itself. Catches itself. Not bad. Not bad. Okay. Seems like a full recovery. Seems like you're ready to go. Another one. I'm liking it. Yeah! And then— Yeah! Yeah! Yeah! Yeah! Not good. And then I wonder what it's thinking because you would think that there would be some programming.

Oh, they got to cut the music.

They gotta cut the music.

Don't let the music keep playing while your boy's down.

He just gets carried off like this. It's so crazy. He just carried off like this. Anyway, Singularity Delayed.

Yeah, it's possible that the robot died from embarrassment.

Yeah, or maybe it was damaged. It's possible that as it hit the ground, it was just actually taken out. Anyway, Google I/O. Bunch of different announcements. Brandon Garel on our team posted on the TBP newsletter some reactions, sort of bucketed it into 4 key areas. Intelligent eyewear. This is an interesting one. I want to go into this Gemini Omni. We talked about the videos, we played a little bit of that yesterday. Upgrades to Gemini LLMs. Those have been mixed reactions from developers. We'll go through that. And then antigravity, which is an interesting place with an interesting history. So let's start with intelligent eyewear. If you had to pick Warby Parker, Gentle Monster. Have you heard of Gentle Monster before? I've heard of Warby Parker. I know the story. I'm a fan of it, of the business story.

I'm super familiar.

I haven't worn glasses in a very long time, so I'm not really in the market. But the Warby Parkers, I've always enjoyed the way they thought about the brand. And I've also been impressed by the way they built that business. They were early to the D2C boom. And then didn't some of the founders move over to Harry's? Is that the same team or is that a different—

maybe I'm thinking of— they were certainly— when I think of D2C, I think of Warby. I think of Allbirds. I think of Everlane.

Yes. But when you think of the last 2 out of those 3, the market caps are sub-$100 million. Allbirds was trading at, what, $20 million or something, and then spiked because of the AI thing. But Allbirds, Everlane, not really sustainable businesses. Warby Parker, on the other hand, current market caps—

Allbirds has pretty much been—

Cookin's sitting at $3.5 billion.

Allbirds has pretty much been— down only.

Down only. This is the pump, the AI pump.

On their Neo Cloud, no surprises there.

Well, have they given us an update on how they are rolling out Kubernetes, how it's going, building their Neo Cloud? Did they get allocation? Are they racking Cerebras? Are they racking GB200s? What are they racking and how fast are they getting power?

It would be funny if Jensen ends up having to talk about Allbirds on the earnings call today.

Didn't they?

There's a new—

Yeah, didn't they also like fully rebrand the name? It was gonna be like Bird AI or All AI. —like, they were moving—

Allbirds still exists, but they basically kept the public entity, and that's what they're building the Neo Cloud through. So fun. Well, lots of fun.

Uh, Warby Parker, uh, resilient. I mean, uh, in 2021 it was a $6 billion company. Now it's a $3 billion company. Uh, not the best scenario, but surprisingly resilient, I think, in a time when a lot of people wrote off a lot of the standalone direct-to-consumer. It was like either get rolled into a bigger company or go or face the fate of the public markets. But Warby Parker has a deal with Google and Samsung. Google says, "We're partnering with Samsung, Gentle Monster, and Warby Parker on new intelligent eyewear. Here's a sneak peek at two designs from this fall's upcoming collections." And people are— "I like featureonomics from Sam." "Kind of crazy that you can wear your favorite Mag-7 on your face now." You can. The Gentle Monster one does a really good job of hiding the camera. I imagine that it will have a light to tell you if it's recording, but if someone wore these from a distance— also Meta Ray-Bans have done sort of the hard work of becoming the first face computer. So when you see Ray-Bans and they're a little thick, you start immediately thinking, oh, should I be looking for a camera lens? Am I being recorded? But the Gentle Monster design, the silhouette, doesn't scream technology, it doesn't scream wearable face camera. And so these are going to be a little bit more stealthy. Warby Parkers look nice, but the camera bump on this— you zoom in on the Warby Parkers—

you know, it makes a lot of sense that the Googles and the Metas have to go and partner on different silhouettes. Yeah. My expectation, my uninformed expectation, is that Apple will just make Apple Glasses, right? They will probably— I, it's hard to see them taking the route, at least early on, of partnering and allowing another company to influence the design language. But it makes a lot of sense that Meta would partner with Luxottica. Okay. First, look at the camera bump on this.

If you zoom in as far as you can, I don't know if we can zoom in any further, but the camera is actually not flush with the frames. It's actually protruding a little bit. Yeah, you can see it right there. Interesting design choice. I wonder how that will catch the light, how that will reflect in the real world. But this was all from a joke from Abdu, who says, OK, so Apple has Carl Zeiss, Meta has Ray-Bans and Oakley. Google has Gentle Monster and Warby Parker. Boring. Which company is going to be bold enough to slap wearable technology into some 3M safety glasses? Would you rock these, Jordi? I like those. 3M safety glasses? I like those. You know what I'm talking about, right? You're working with a buzz saw.

I do know. Dust in your eyes. But these look cool. These are sporty. I'm much more likely to just commit to the bit. Just go full planker. And be wearing a face computer.

Full cyberpunk. Full clank. Yeah, full clank out. Full cyberpunk. I think that might be the move. I don't know. For some company, a challenger company could potentially do that. Maybe Friend or something. Anyway, what else? So Warby Parker traded down on the news, which Shiel Monat was surprised by. Why is Warby Parker down 14%? They announced a partnership at Google I/O that's been in the works for a while. Is it because they aren't available yet? And our friend Rat King, Mike Isaac, says, OK, Google AI glasses. With Warby Parker officially coming for Meta Ray-Bans. Google also said it would bring Gemini to glasses this fall with Samsung Electronics and the eyewear companies Warby Parker and Gentle Monster. The glasses, which work similarly to Meta Ray-Ban smart glasses, come with a camera, microphone.

Yeah, at what point, at what point does Google just buy Warby Parker, right? It's a $3 billion company. It's actually done quite well. Yeah. Over the last, uh, over the last 6 months, it's up 43% in the last 6 months, although it's been almost flat this year. I would say I expect that smart glasses are gonna have product market fit among people that need to wear glasses first, right? If you already have to wear glasses all day long for your vision, why not throw some smart features in there? It's gonna be harder to get someone that doesn't need glasses to add a new device to their rotation. Right? And so, you know, Warby Parker's done quite well and has been surprisingly resilient, but they have incredible distribution, and I wouldn't be surprised if they get sniped at some point.

I mean, deeper integration into a traditional, I don't know, like workflow. A lot of the Google I/O, we'll get into this, but was talking about Spark, the personal AI assistant, and when I think about— SimSpark. It's an API.

I know, I know. Oh, just look in Omni.

Yeah, there's a lot of names. It's Google. There's a lot of products. You're referring to, of course, Nathan Clark's post. It's in Gemini. Just created an AI Studio. Oh, it's for your personal Google account. For workspace, you need Gemini Business. No, not Gemini Advanced. That's AI Pro now, unless you need AI Ultra. Oh, agents? You do that in Spark, actually. No, not Gemini API Managed. And it's the typical. Meme. The interesting thing is that I do think Meta Ray-Bans, like it was always like, okay, you have a deep integration with WhatsApp, you have a deep integration with Instagram DMs, maybe Facebook Messenger. Some people are still using that. But in terms of like wiring into your life, there are way more people that see Google Docs, Gmail as like the central node in their personal life. Like people, like people think of like my— the— all the stuff I have saved on the desktop of my MacBook is like my core repository. A lot of people think, okay, for the important stuff, I'll put it in Google Docs or Google Drive. And then most things flow through Gmail. Most things flow through iMessage. There are some people that just are like, yeah, WhatsApp is the number one screen time for me. That's where I really organize things. But Meta doesn't really have this. Like knock-on effect of like, oh yes, you've, you're, it's not necessarily an enterprise-level productivity suite, but there are people who are like, yeah, I'm using Apple Mail, iMessage, I save my files in Apple Files, Apple, you know, the cloud storage I use, my camera roll, super important. So an AI agent running through the Apple ecosystem can be valuable and an AI agent running through the Google ecosystem can be valuable. The Meta smart glasses, it's a little bit trickier to go and do anything because you're just like sort of bumping up against the walled gardens, right? Yeah. But investor Nick doesn't like them for aesthetic reasons. He says these Google X Warby Parker glasses are horrific looking compared to these Meta Ray-Bans. Someone is probably going to lose their job over this. I don't know that they look that much worse. I don't know. Ray-Bans are a very iconic silhouette and they do look good. So we'll see, we'll see how the response goes. I think, uh, from a product perspective, there's obviously fertile ground. On the flip side, the Wayfarer is just such an iconic— it's more iconic than anything Warby Parker has produced. And that's just sort of the reality of brand building over a decade versus a century or something like that, however long Ray-Ban's been around. Long time. Uh, anyway, uh, Genie 3. You can now simulate real places by grounding Genie 3 experiences with Street View imagery. Google is sitting on a motherlode of real-world data. I was always thinking YouTube was gonna be so valuable for Omni and VO3, VO4 maybe in the future. I hadn't considered Street View as a trove of data. Demis seems very data-pilled. He seems— a lot of the Mag Seven CEOs seem very data-pilled. There's that story about Mark Zuckerberg screen recording or logging all the computer use from all the Meta employees, these important troves of data are increasing in value and Street View certainly seems like it's one of them. This is cool. I wonder how interactive this will be, how this actually instantiates into a game. It's a great demo.

What does it take to build a game on this? Yeah. Do they allow people to build games on top of this?

Yeah. I just think about— I don't know. I mean, Dennis has a background in games and he was sort of alluding to the fact that he might go back into games at some point or or at least be able to like scratch that itch again. Famously, he wrote a programmatic code to generate vomit in a roller coaster simulator. Very fun story. But again, when I think about Roller Coaster Tycoon, which was not— I don't think he was actually working on that game. It was a similar theme park simulator. But we are moving back into the simulator world. But the mechanic is what is so enticing to gamers often. When I think about the games that I've spent a long time with, some of them have incredible graphics, AAA graphics. Some of them have 2D graphics, but the mechanic is great. And so that is what gets me to keep coming back.

The legend Bobby Chipman in the X chat says, "Can't wait for smart glasses to fully replace my monitors." Yeah, maybe you'll need augmented reality or something.

Meta Ray-Ban display certainly going that direction. The Orion. I've been surprised. Wasn't the first episode we ever did, we were talking about Orion and they still haven't shipped it, right? I mean, they shipped a smaller version, the Meta Ray-Ban displays, which have sort of the Call of Duty HUD. It's not full augmented reality. I was expecting— I was expecting— we demoed the Orion headset and it has a bit of a narrow field of view, but it really can put a screen right in front of you. And I assumed that, you know, everyone was saying it's really expensive, it's clunky, it's not ready for primetime. But, you know, look at how fast things are going. In a year, maybe two, We'll get it. And maybe that's coming at the next Meta Connect. Maybe this summer we'll see it, but haven't been that many rumbles on it. And then obviously the massive pitch shift to AI CapEx might have taken a backseat. I don't know. I'm certainly hopeful. I like AR and VR. I think, I think we're overdue for a new fun product. I'm still waiting for the next Apple Vision Pro, Apple Vision Air, something just lighter. That's all I want. Cheaper maybe, but lighter and Same screen, screen was great. Anyway. We know, John. Gemini Flash 3.5 looks pretty neat according to Tenebris and extremely fast, but still largely the sort of incremental progress we've come to expect from Google. Generally a pretty disappointing I/O. Now what's interesting is that Gemini 3 felt like a new base pre-train, felt like it had some of that big model smell, felt like it was, you know, really delightful to talk to. And I think a lot of people were expecting Gemini 4 here, We're still waiting for the next iteration here. And also— Yeah, we're still waiting for Pro. Yeah, Pro isn't out. But people are speculating that 3.5 Pro won't necessarily be a new pre-train. And so it seems like there's a little bit of research being at odds with the product cadence. Google I/O is scheduled probably like 2 years in advance. And whether or not the training run finishes on time is a little bit harder to— package up and nail on a specific time. We see this with the independent labs or the OpenAI, Anthropic, the other labs, XAI, like they're launching models very much like when they're done and then they will like instantiate like something that looks like a conference around it or maybe a video or a blog post, a model card. But if you're grinding towards a specific date and the specific model isn't quite ready, you come out with something that's looks a little bit more incremental. People were really, really honing in on the fact that the cutoff date was January of 2025, right? Was that the date or was it December of 2025? Either way. January. Yeah, I don't know. I don't know how much cutoff dates matter because all these models can query the web and get up-to-date information.

Yeah, overall reactions from developers across the board were not good. Not good. Yeah, Prakash here, underwhelming. Cursor ranked it on CursorBench. Uh, it is below Composer 2. Is that, is that a fair thing?

I mean, I'd like to see you rank another, another live stream on TVPN Bench. It doesn't match up, you know.

It's like, no, I mean, they have, they have, I guess it is all the other, they have all the other frontier.

And some of them are ahead of CursorZone models on CursorZone.

It's just one, it's just one data point.

But yeah, yeah, yeah, yeah, but here's the other thing.

It's 4 times— it costs 4 times— it underperforms Composer 2 even though it's roughly 4 times more expensive.

Interesting, interesting. Yeah, I feel like for a long time Google's positioning was, you know, frontier or near frontier but best possible pricing, and this marks sort of a shift in the strategy perhaps.

Yeah, overall starting to make more and more and more sense why Google has put so much capital and resources behind Anthropic, right?

Yeah, Prakash says, seems to indicate that DeepMind is constrained by data rather than compute for what they intend to do, hence the TPU sales. Rest of Google now shipping their org chart. Ben Thompson talked about that a little bit and there was some context on like, you know, we were asking the question like, will there be AI fatigue from stuffing AI in every product surface area? Allie K. Miller shares one of the loudest applause clauses in the entire Google keynote. Nishtha put on the Gentle Monster plus Gemini glasses, tapped the side to summon Gemini, and all in one prompt said, take a photo and put a cartoon blimp in the sky that says Google I/O 2026. And within seconds, the preview of the edited photo from Nana Banana appeared on her watch. I want to spend less time on screens. AI is really coming everywhere, and so much is driven by voice AI as the interaction mode. Very cool demo, impressive technology, but Greg's Gadgets says, "These companies truly have no idea what regular people want." Because, yeah, that is a little bit of a niche use case. You need to be more creative with it for when you would actually use that because this is a very— it's a perfect demo of the product and the functionality, but it lacks that, like, creative spark of, like, "Yes, I did want a picture of that on my wrist at that key moment in time." If you're not doing a demo, his point is that regular people would not be excited about that particular feature. SpaceX IPO, we're getting more details by the day.

The other thing that was going pretty viral, the last thing on I/O, was that Google Anti-Gravity team flashed a codecs folder in their actual demo video. Gergely says, "I had to do a double take in the second minute of the launch video for Anti-Gravity. You can see people use codecs." on the Antigravity team. Did no one double-check the launch video at the very least? Not a huge surprise. Obviously Antigravity looks, a lot of people were saying, looks quite like Codex. So clearly— More than Windsurf.

I feel like it would be like a, they would've just rebuilt Windsurf. I don't know. We'll have to see. No.

So, but anyways, this isn't a huge surprise, right? Google's been using a bunch of Anthropic models. Clearly they're using, a ton of different models and products internally.

So what was that drama with Steve Yeager going back and forth with Demis about like what teams are using what models and stuff? There was a big back and forth, big dustup on the timeline like a month ago about like whether or not Google's employees were deploying AI efficiently or broadly. Some of them aren't and some of them are. And Demis chimed in and said like, this is just complete wrong. And everyone's using AI. I don't know, goes back and forth. There's also people are benchmarking OmniFlash, which looked amazing when we saw the videos. There was a, there were a few like little quirks. Some people in the chat were saying that the firing order of the V8 was not correct. Maybe it was only a V6. Yeah, it was missing 2 cylinders. It was missing 2 cylinders, but it looked good to me. I don't know. But now people are actually comping it to SeaDance 2.0, which obviously has much looser content restrictions, uh, because I, I guess just like Hollywood can't file a lawsuit in China. I, I'm not exactly sure how that works because CDance seems to be available in America.

It, it seems like maybe— Oh, I think, I think Chinese businesses have been relatively immune to U.S. copyright law for a very, very, very long time.

And it also might just take like years to file a lawsuit, do discovery, actually go through and litigate.

Oh yeah, you can just go, like there's malls in China where you can go to a Nike store and Nike has nothing to do with it. And yet all the products— They've been selling Swatch APs over there for decades. Yeah, yeah. Just ask Rolex and Patek how they're—

Sure, sure. Yeah, I've heard fake cars too. Like you can get a full replica of like a G-Wagon that's just made in a factory and then you could buy it, bring it over here, and you take it to a Mercedes dealership and they're just like, this is not a Mercedes, but it looks like one, like, like, you know, to the millimeter from the outside. But internally it's just, it's just frauding. Anyway, SeaDance 2.0 looks great. OmniFlash looks great as well. These are both like super useful. We'll see how they actually play out and how they get implemented, how they get used. The interesting thing will be like, like at what point, like it still takes a long time to generate videos. Very hard to get them right. The last 90, like we're at 99% fidelity. But when you click in, you start noticing little details. When will we be in a paradigm where you ask a question and you actually get an explainer video, 6 minutes, 10 minutes like you would on YouTube? Very computationally expensive, very difficult to maintain the logic. Like what is the deep research report of OmniFlash? These 8-second, 10-second, 20-second videos are impressive. But not perfectly substitutable for a 20-minute YouTube video because of the time and the level of detail that you can go into. Some people that are looking for information about a V8 engine, they want a breakdown that lasts 20 minutes. And so that's the next benchmark. We got to move the goalposts. SpaceX IPO. The prospectus is incoming, is incoming according to ZeroHedge, as soon as May 20th. That's today. We will see Goldman lead left. This was a surprise. Michael Grimes has worked with Elon for a long time at Morgan Stanley. There was some back and forth. He went back to Morgan Stanley. There's a question about whether or not there would even be a lead left because it's such a big IPO. Maybe they all share equally. Obviously, they're all going to make a ton of money off of this. So good news from start to finish. But it is interesting that Goldman was selected. Do you have a soundboard cue you want to play? I'm always ready, Jon. Katie Roof has a scoop. Uh, the scoop athlete of the century, Katie Roof, has a scoop on the biggest venture returns ever. Founders Fund and Valor are set to make more than $60 billion in gains on the SpaceX IPO. Sequoia, more than $20 billion.

Yeah, was this, was this somewhat of a reaction to, uh, D1 getting a lot of credit earlier in the week, right? They're set to, uh, to to generate roughly $20 billion in returns, and maybe some of these other funds thought to put their hand up and say—

I think that at this scale, like, there are so many LPs in these funds that are getting updates, and they've known the numbers for a long time. They've known the ownership, the holdings, and you do some back of the envelope, and you get some pretty huge numbers. Will be very interesting. Huge for Sean Maguire, huge for Luke Nosek, and a lot of other folks over at Founders Fund and Antonio Gracias at Valor and all the other Founders Fund folks.

Yeah, look, Sequoia, Founders Fund. They needed a win. They needed a win.

I mean, you go back, like, you know, these investments were made like 2004, 2010. Like, it was not obvious. Certainly there was no Starlink narrative when these were made. There was no space data center narrative. This was a rocket company that was blowing up rockets left and right and not quite getting to, massive business. You really had to be a believer, and they were.

Packy was having some fun on the timeline.

He said, "Megafunds are too big to generate returns.

They're basically just fee collectors." And of course, they're printing.

Yes, they are printing. Jensen Huang talked about the quarter that Nvidia just had. He said the buildout of AI factories— boo! The largest infrastructure expansion in human history. Yay! Is accelerating at extraordinary speed. Yay! Not a fan of the AI factory terminology, but good that there is progress being made. Agentic AI has arrived, he says, doing productive work. True. Generating real value. True. And scaling rapidly across companies and industries. Also true. Lots of good stuff. NVIDIA net income It rose to $42.96 billion. They almost hit $43 billion. Not too bad. A year earlier, they were doing just $18.8 billion in net income. Huge, huge increase. Really wild.

Really wild. Printing.

Definition of printing. All good news. The stock is sort of up and down, basically flat. But NVIDIA revenue jumped 85% to $81. $1.62 billion from $44 billion last year, the company said. Shocking! Yeah, great stuff. Steve Wozniak, the co-founder of Apple at Grand Valley State University— he talked about AI, and unlike Eric Schmidt, he did not get booed offstage; he actually got cheered for his comments. Wait, this was who?

Was it Tyler? Steve Wash-niac? No, no, I'm a big fan of Steve Wozniak.

I love the Woz. Shots fired. Let's play the clip from Grand Valley State University on Instagram here.

You all have AI.

You all have AI.

Actual intelligence.

Oh, mic drop. Knee slapper. Hey, playing to the crowd. He knows the audience. He knows the audience.

We're trying to figure out how to make a brain. Software, hardware, synapse chips, And I was at a company where the engineers figured out how to make a brain. Takes 9 months. Yeah.

Knee slapper. But he knows the audience. He's delivering the right thing. Is he AGI-pilled? Is he superintelligence-pilled? Probably not. But it is— regardless, I think it's potentially the right framing for the crowd. It's knowing the audience and that is a way to bridge to a broader conversation about AI, a broader conversation about how humans fit into a post-AGI world. I don't know, we'll have to go watch the full clip eventually.

Let's close it out with this video from Tyler that we can pull up.

Okay, what is this?

This is the video I was referencing with Marcus. I'm very concerned about these gentlemen and what they're doing. Pull it up.

What's happening?

Let's get— Like, this is insane. Insane contact. I mean, the helmet is getting dented. I think this is breaking.

Not an issue, skill issue. They really hitting that, aren't they?

But we'll try it out after we—

we have the gong mallet. Yeah, we'll see you tomorrow. Goodbye. Cheers.