The AI Daily Brief: Artificial Intelligence News and Analysis - Is Pixel 10 the AI Phone iPhone Never Was?

Starting point is 00:00:00 Today on the AI Daily Brief, why the Pixel 10 is the AI phone that the iPhone couldn't be. Before then, in the headlines, Meta's Super Intelligence team takes shape. The AI Daily Brief is a daily podcast and video about the most important news and discussions in AI. All right, friends, quick announcements before we dive in. First of all, thank you to today's sponsors, KPMG, Blitzy, Vanta, and Superintelligent. To get an ad-free version of the show, go to patreon.com slash AI Daily Brief. And if you are interested in sponsoring the show, send us a note at sponsors at AI Daily Brief.A.I. We are basically full for the rest of 2025 and now selling starting into

Starting point is 00:00:43 26. And with that, let's get into today's show. Welcome back to the AI Daily Brief Headlines Edition, all the daily AI news you need in around five minutes. We kick off today with some reporting out of meta. Now, meta has obviously been in a big transition with its AI strategy over the course of this year. As we've covered extensively, their rating expeditions across Silicon Valley have totally repriced AI talent in a huge, huge way. As their superintelligence group has come together, we've started to see leadership and who fits where, but what we're getting now is exactly how that group is going to be structured.

Starting point is 00:01:17 And to some extent, if this was just normal internal politics, I wouldn't really think it's all that worthy of spending a bunch of time on, but because of how it's been reported, with basically Wall Street taking it as a sign of chaos and even as a withdrawal from AI, I think we need to do a little bit of setting the record straight. So basically, at the end of last week, the information published it as a scoop in a piece they titled Meta Plan's fourth restructuring of AI efforts in six months.

Starting point is 00:01:43 The headline implied that this has been some chaotic process, and even more it implied that this was another big shift after and separate from the superintelligence effort in general. That doesn't seem to me to be what's going on here. What it looks like to me from the outside, and of course I don't know, I could be totally wrong about this, is that now that they've hired all this talent, they've made a decision about how they're going to organize the whole effort, and this is it. In other words, this is the natural, inevitable follow-up to the big shift, which was the creation

Starting point is 00:02:13 of this superintelligence team, not some new effort that's like less than a month after that team came together. It's now getting shifted again. In any case, after the rumors came out, on Tuesday, meta-formally announced internally this new structure and explained the purpose for each group. Meta-superintelligence labs will now comprise of the TBD lab, led by Alexander Wang, which will oversee their large language models, including Lama, Fair, which is their fundamental AI research lab that's responsible for longer-term research projects, products and applied research, which will be led by Nat Friedman, former GitHub CEO and prolific AI investor, which will basically be charged with taking meta's models and bringing them into consumer products,

Starting point is 00:02:50 and MSL Infra, which will focus on supporting meta's gigantic data center and AI infrastructure buildout. Alexander Wang, who also serves as Meta's chief AI officer, wrote in a memo to staff that the restructuring is supposed to accelerate the company's pursuit of superintelligence. He said, Superintelligence is coming and in order to take it seriously, we need to organize around the key areas that will be critical to reach it, research, product, and infrastructure. According to Bloomberg, no layoffs were part of Tuesday's reorganization.

Starting point is 00:03:17 However, per New York Times sources, quote, some AI executives are expected to leave, and meta is also looking at downsizing the AI division overall, which could include eliminating roles or moving employees to other parts of the company because it has grown to thousands of people in recent years. This led to follow-ups from the Wall Street Journal like this one. Meta freezes AI hiring after Blockbuster Spending Spree. Quote, the hiring freeze which went into effect last week and coincides with broader restructuring of the group

Starting point is 00:03:42 also prohibits current employees from moving across teams inside the division. This led to this type of reporting from Axios, Wall Street faces an AI reality check from Meta. Meta stock fell this week on reports of the company looking at downsizing its AI division. Axios writes, this is what Wall Street fears most, a pullback in AI spending from the companies propping up the entire stock market. Still again, going back to the idea that this is just the shakeout from

Starting point is 00:04:06 and culmination of those previous efforts rather than some new chaos, the New York Times also wrote that their sources said that the restructuring is likely to be the final one for some time. Bloomberg framed it as, Meta is hoping to stabilize its AI efforts after months spent poaching dozens of top AI researchers from competitors with lofty pay packages. Look, if you're trying to understand the reporting and the reception, of all of this, the big signal takeaway is that Wall Street is extraordinarily jittery and very much concerned by how much AI spending is at the center of the stock markets quote-unquote success right now. And so anything that has even a width of companies heading in a different direction is causing shockwaves.

Starting point is 00:04:44 Now, if it makes you feel any better, this has been happening all year. Remember when there were reports that Microsoft was pulling back on their data center spending, just because of a couple contracts that they let lapse because their priorities changed. There was a couple week period there where everyone was absolutely freaking out, and that ended up being a big nothing burger as well. In any case, the point is that we've yet to really see what this new meta group is going to do, but I'm sure we will get more information about what their plans are soon. Now, speaking of the infrastructure side of Meta's buildout, Louisiana regulators have approved three new gas plants to power their largest data center. Entergy Corp is planning on building the facilities alongside Meta's 4 million square foot complex in rural Louisiana.

Starting point is 00:05:22 Mark Zuckerberg announced the data center known as Hyperion in July. lie, sharing that it would be comparable to the size of Manhattan. At full capacity, it's expected to consume as much as 5 gigawatts of power. Now, speaking of data centers, interesting one from OpenAI. In an interview on Wednesday, OpenAI's chief financial officer, Sarah Fryer, said that Open AI is currently not actively looking at renting out spare data center capacity to other AI companies, but said that, quote, I do think about it as a business down the line for sure. Now, at this point, obviously Open AI does not have a lot of spare capacity that they could be renting out even if they wanted to. Each new model release sees them hit resource limits, and it sounds like

Starting point is 00:05:58 the design of GPD5 was even constrained based on how much compute the company anticipated having available. At the same time, as Sam Altman made clear in that dinner last week, OpenAI is planning on spending trillions on data center construction in the not too distant future. Friar seemed to endorse this scope commenting, I think we are so in the early innings. Lots of prognostications want to say we're on a sugar rush. We're not. It's more like the railroads or the build out of electricity than anything I've seen. The internet, in hindsight, was actually a relatively Kappex-like build-out, so I think we're just getting started. In terms of AI, demand is voracious right now for GPUs and for compute. The biggest thing we face is being constantly under compute. That's why we launch Stargate.

Starting point is 00:06:35 That's why we're doing the bigger builds. And we're just getting started. A follow-up from the Aqua hire stories from last year and earlier this year. Character AI is apparently in talks to tell themselves or raise more money as they figure out what's next. Sources speaking with the information said that the company is holding discussion with potential acquisition partners and bankers to sound out a deal. Executives have also spoken with investors about raising a few hundred million dollars at a valuation above a billion. Now, Character AI is in a strange position in the industry.

Starting point is 00:07:02 Last September, founders Noam Shazir and Daniel DeFretas agreed to join Google in one of the first high-profile aqua-hire deals in AI. After they left, Character AI's roughly 70 employees were left as the owners of the company. Now, these aqua-hire stories usually end with the startup winding down, paying out employees with the money they gain from the deal or selling the company for parts. Winsurf was a recent example of the latter exit strategy. Despite losing their founders, however, Character AI has continued to have a viable business and even expects to hit a $50 million ARR by the end of the year. The issue is that their services are starting to get more costly to run,

Starting point is 00:07:35 and it's unclear whether they have the pricing power to charge more than their current $10 a month. Now, there's another question about why this information is leaking now, and whether that means they're in more trouble than it even seems. But still, to the extent that we are interested in how these weird sort of aqua-hire things play out, it's a good one to watch. Lastly, in an indicator of the growing viability of general purpose agents, the company behind the Manus agent have hit 90 million in ARR. Now, one of the reasons that I'm watching them is that while we have seen the absolute total breakout use case of agentic coding, it's still very unclear how general purpose agents are

Starting point is 00:08:08 going to play out and if there are even going to be a thing outside of the context of the foundation labs are only as offerings from the foundation labs. 90 million annualized at this speed is no small feet, however, so maybe there are actually some legs there. In any case for now, that's going to do it for today's AI Daily Brief Headlines edition. Next up, the main episode. What if AI wasn't just a buzzword, but a business imperative? On You Can with AI, we take you inside the boardrooms and strategy sessions of the world's most forward-thinking enterprises. Hosted by me, Nathaniel Wittamore, and powered by KPMG, the seven-part series delivers real-world insights from leaders who are scaling AI with

Starting point is 00:08:44 purpose. From aligning culture and leadership to building trust, data readiness, and and deploying AI agents. Whether you're a C-suite executive, strategist, or innovator, this podcast is your front row seat to the future of Enterprise AI. So go check it out at www.kpmg.org.us slash AI podcasts or search you can with AI on Spotify, Apple Podcasts, or wherever you get your podcasts. This episode is brought to you by Blitzy, the Enterprise Autonomous Software Development Platform with Infinite Code Context. Blitzy uses thousands of specialized AI agents that think for hours to understand enterprise-scale code bases with millions of lines of code. Enterprise engineering leaders start every development sprint with the Blitzy platform,

Starting point is 00:09:27 bringing in their development requirements. The Blitzy platform provides a plan, then generates and pre-compiles code for each task. Blitzy delivers 80% plus of the development work autonomously while providing a guide for the final 20% of human development work required to complete the sprint. Public companies are achieving a 5x engineering velocity increase when incorporating Blitzy as their pre-IDE development tool, pairing it with their coding co-pilot of choice to bring an AI-Native STLC into their org. Blitzy is providing a limited time, 30-day free proof-of-concept for qualifying enterprises. The team will provide a 5x velocity increase on a real development project in your org.

Starting point is 00:10:01 Visit blitzy.com and press book demo to learn how Blitzie transforms your STLC from AI-assisted to AI Native. That's BLITZY.com. As a founder, you're moving fast towards product market fit, your next round, or your first big enterprise deal. But with AI accelerating how quickly startups build and ship, security expectations are higher earlier than ever. Getting security and compliance right can unlock growth or stall it if you wait too long. With deep integrations and automated workflows built for fast-moving teams,

Starting point is 00:10:33 Vanta gets you audit-ready fast and keeps you secure with continuous monitoring as your models, infra and customers evolve. Fast-growing customers like Langchane, writer and cursor, trusted Vanta to build a scalable foundation from the start. And look, as someone who lives in the world of enterprise procurement, I love how Vanta makes it easy to get compliance right. The last thing you need when you're trying to win that big deal is to have it scuttled by something that Vanta has solved for over 10,000 companies. Go to vanta.com slash NLW to save $1,000 today through the Vanta for Startups Program and join over 10,000 ambitious companies already scaling with Vanta. That's v-a-n-ta.com slash NLW to save $1,000 for a limited time. If you are a regular listener, you will have heard

Starting point is 00:11:16 about Super Intelligence Agent Readiness Audits at this point. But I wanted to tell you today about the full suite of Agent Readiness products that go beyond just the initial readiness report. Over the last six months, Super Intelligence has built out an entire agent planning suite. We help you move from discovery to planning to implementation. After you've completed your agent readiness audits, we help you double-click on your most important use cases with what we call our use case planning reports. These reports are going to help you understand what sort of technical preparation you need to do to be ready for a use case, what challenges you might face in implementation, and whether you should be thinking about building, buying, partnering, or some combination.

Starting point is 00:11:54 After that, you can even get a spec document in what we call our technical blueprint that gives either your developers or the developers of the partner you work with what they need to build exactly the agent that you're looking for. If you want to learn more about superintelligence agent planning suite, we built a custom GPT to answer your questions. Just go to bit.ly slash super super agent. That's bit.l.ly slash super super agent, all one word. And if you have any questions, the agent can even help you book an appointment with our team. Welcome back to the AI Daily Brief. We have had a lot of big macro sort of narrative discussions this week, trends in how people are thinking about the AI industry overall, questions about an uncertain future and what it means.

Starting point is 00:12:40 not to mention more bubble talk than you can shake a stick at. So I have to say I was quite excited to have an actual product announcement to cover today, with the product that I'm referring to, of course, being Google's Pixel 10. Now, as much excitement as there is about the future of AI devices and whatever Sam Altman and Johnny I have cooking over in San Francisco, the reality is that for most people, they're going to experience AI and AI features first, probably through their phones. For Normies at this point, mobile is by far their most important computing experience, and that is of course why it seemed like Apple's whole AI strategy, or Apple intelligence strategy, as it were,

Starting point is 00:13:18 centered around bringing actually useful, simple, AI-enabled experiences to regular consumers via the iPhone. That was the promise at least, but as it's happened, it has just been delay after failure, after delay after failure, ultimately underwhelming and leaving Apple farther behind than ever before when it comes to this AI race. Now, Google has so many things going on, not only in the AI space but beyond, but even just in the context of their AI efforts, something like their Pixel 10 announcement almost got buried

Starting point is 00:13:48 compared to everything else. In fact, as you'll hear, a lot of the discourse at least from the AI insiders, was all about sadness about something that didn't seem to be released. In an event, what we're going to do is we're going to talk about the Pixel 10, what features it has, what people are excited about how they're discussing it, and what it means for the state of AI hardware. Now, the reality is, Google's AI rollout to the Android ecosystem has been an ongoing project. Gemini has been enshrined as the default assistant for several months, and there have been numerous little integrations here and there. This release, however, is not part of that incremental

Starting point is 00:14:20 rollout. It is instead a complete overhaul of Google's flagship handset in both hardware and software to be AI first. And if you had any doubt about how the market was receiving this thing, the very first line of the Wall Street Journal article entitled, by the way, Google is beating Apple on smartphone AI was this. The race to develop the killer AI powered phone is on, but Apple is getting lapped by its Android competitors. We'll come back to that discussion, but let's first talk about the features. One of the big ones is called Magic Q, and that's the name they've given to Google's agentic assistant. It can rifle through your apps to find certain information and offer context-sensitive assistance. For example, if a friend texts you to ask

Starting point is 00:15:00 where you made dinner reservations, the assistant will search your calendar and deliver the information in a pop-up. Initially, the feature, it sounds like, will be limited to Google's family of apps, but given that they have calendar, Gmail, and so much more, there's still quite a few integrations there. In the context of that Apple competition, this was obviously core functionality that was promised with Apple intelligence but was never delivered. What's more, when it comes to the user experience of AI, a small but important point is that the feature is going to run as a passive AI service rather than needing the user to prompt it. Next up, we have a feature called visual overlays that can use the camera as a live input for AI queries.

Starting point is 00:15:37 Google showed an example of pointing the camera at a pile of metric wrenches to see if anyone will fit on a half-inch bolt. Now, on the one hand, this is sort of table stakes. This is the type of thing that people have been excited about with ChatGBTBT for a while, but obviously the integration directly into the operating system brings the utility of this type of feature to another level. Another feature is that the Gemini Live AI will be able to detect your tone. For example, being able to figure out if you're excited or concerned

Starting point is 00:15:59 and adjusting your outputs to match. And this, I think, gets to one of the most important underlying subtext of this whole launch, which is that this is basically all about exploring how much new and important context the mobile interface can provide when it comes to AI assistance. When you're prompting a regular generic chatbot, you have to provide all the context yourself. In other words, you have to explicitly type or say via voice, what's your feeling, the type of output you want. you have to explicitly share with it context from other documentation. Now, yes, that is changing a little bit as memory becomes a more important part of these systems,

Starting point is 00:16:38 but by and large, each prompt is tabula rasa when it comes to the context that you might want to bring to it. Your mobile device, though, is one big bucket of personal context. Mobile devices are our command and control centers. They have information about our friends. They have a record of what we've been doing, what we've been looking for. All of these systems potentially give any sort of AI that's interact. acting with you via a mobile experience, more context that can be used to make the output better. So this tone detection, for example, is another expression of using the particular form factor

Starting point is 00:17:08 of mobile to get more context, but really I think the important thing is that underlying idea of being able to get more context. There's also a live translation feature for phone calls, an AI enhanced note-taking feature, and several other minor quality of life additions. Basically, it sort of sounds like Google took the Apple Intelligence announcement video from last year and actually built all of the tech. Still, maybe the thing that's generated the most conversation so far is the camera and photo editing suite. The core offering that Google is promoting is a feature called camera coach. It does things like tell you how to properly frame a photo, give an interesting photo suggestion, and automatically choose the best picture

Starting point is 00:17:42 from a series of shots. One of the big features getting a lot of attention is 100x zoom. Google shared a demo of a full frame photograph of a car, then showed that the photographer had taken it from nearly a mile away. The interesting thing here is that this is actually a sneaky AI feature. Instead of achieving this with optical lenses, Google is using AI to generatively fill in the details as you zoom in. It's basically a sci-fi enhanced tool. Now, because this is AI not optical, it does feel like the sort of thing that we'll need to get our hands on to know how useful it really is. But if you're looking at the trend lines, it's still interesting to note that Google is starting to use ImageGen for live camera use. The other big photography feature is AI picture

Starting point is 00:18:21 which Google is calling Edit by Asking. Now, for this, you prompt the AI to edit images as we've come to expect from these modern models. Google gave the example of removing the glare, brightening the photo, and adding clouds to the background to show how specific you could get with this, but it also showed how you could go more generic, taking an older photo and simply saying, restore this old photo. You can also apparently do the old prompting trope of just asking the model to quote-unquote make it better, and you'll get a decent result. Now, to the extent that this announcement had some disappointments, the big one was that we didn't

Starting point is 00:18:53 get any explicit word about nanobanana. Nanobanana is an image generation model that was released in stealth on L.M. Arena earlier this and, as we discussed yesterday, had been absolutely blowing people away. People were especially impressed by its ability to do very precise edits with incredible model consistency and prompt adherence. Now, even before this, people pretty quickly started to suspect it was a Google model, which seemed to get confirmation when a number of different Googlers, including Logan Kilpatrick, started tweeting out just the banana emoji.

Starting point is 00:19:21 And yet, when the pixel launch came and went without any sort of explicit banana reveal, there was a fair bit of frustration on the timeline. DeVeed Mori wrote, Okay, let me be honest. Am I the only one underwhelmed by Google's event today? I was expecting more from Gemini, like Gemini 3.0, notebook L.M improvements, the nano-banana release, etc. In reality, they've just announced features seen a year ago and small improvements for daily usage.

Starting point is 00:19:44 Where's the big thing we were expecting? Is this the open AI contagion? That said, to some people, the edit by asking feature looks pretty close to the capabilities of nanobanana, meaning that it's possible that we actually did get a new state-of-the-art image generation model, just not a formal announcement. And I think that to the extent that that is the case, which I'm not sure about, it does make a particular sort of sense in how Google was approaching this particular announcement. Those of you who are listening to this show are used to following AI model announcements

Starting point is 00:20:12 like it's a sports league. And some of the events that we have, such as Dev Days or Google I.O, are exactly catering to this audience of us. The Pixel 10 announcement wasn't that, though. This was the rollout of a new flagship phone. In fact, if you need a sense of just how Normie it was supposed to be, it had Jimmy Fallon hosting the event. There's probably not a realistic world in which Google is going to get Jimmy Fallon to explain why the editing feature is called Nanobanana, or mention that codename anywhere near a product aimed at the average consumer. By the way, for what it's worth, and for the completeness of coverage, people did not love the Jimmy Fallon aspect of this.

Starting point is 00:20:47 Summing up a lot of people's sentiment, tech cruncher Sarah Perez wrote, Google, sorry, but that pixel event was a cringe fest. Sarah writes, The awkward event made Google feel out of touch. It also suggests that the company felt it needed buzz to cover up for the lack of tech advances, which is not the case. Basically, they were arguing that there was enough tech here to not need this sort of celebrity production.

Starting point is 00:21:08 They write, instead it went for buzz with paid celebrity appearances, including event host Jimmy Fallon and others like Stefan Curry, podcaster Alex Cooper, the Jonas Brothers, and more. The result was a watered down cringing at times almost QVC-like sales event, which Reddit users immediately dubbed unwatchable. Jono Tan wrote, I used to wish Apple would bring back live presentations, but after watching the Pixel 10 event,

Starting point is 00:21:29 turns out they made the right call keeping them recorded. Again, I think what's relevant for our audience is not a critique of the particular marketing efforts here, but an acknowledgement of the intention of the audience. This is very clearly aimed at a general consumer. audience. Moving back briefly though to one more thing that matters to this particular audience, a feature that's going a little under the radar is that the phone is powered by Google's new tensor G5 chip. The chip features an AI core that's 60% more powerful than its predecessor. Google claimed this allowed them to run all the AI features on device powered by a version of

Starting point is 00:21:59 Gemini Nano. Google actually took a hit on overall performance by using this chip, meaning that they're making a big bet on on-device AI. Now, Google's custom AI Silicon is starting to get more attention. The latest generation of their Trilium TPU seems to be delivering some impressive performance for AI inference, and Google has started to ramp up their orders with TSMC. On a recent podcast appearance with A16Z hosted by Eric Torrenberg, Dylan Patel of Semi-analysis talked about how the biggest threat to Nvidia is this sort of custom silicon that companies like Google are doing. It's the biggest thing, right, is when we look at orders from Google and from Amazon,

Starting point is 00:22:37 right, especially. And meta, their custom silicon is, not Microsoft, their custom silicon kind of sucks. But the other three, they're really upping their orders massively over the last year. You know, Amazon is making millions of Traneum. Google's making millions of TPUs. TPUs clearly are like 100% utilized, right? Yeah.

Starting point is 00:22:58 Trinium's not there, but I think Amazon will figure out how to do that. And Anthropic will. So I think that's the biggest threat to Invidia. that people figure out how to use custom silicon more broadly. Now, what's interesting about this again, connecting it back to the broader Apple conversation, which started it, is that while it's always been Apple's strategy to own the entire stack, because they haven't been able to actually compete on the AI models themselves, Google now has a much more integrated AI stack of a type that honestly would make Apple of yester,

Starting point is 00:23:31 you're jealous. They're making the chips and the devices to deliver their own models direct to consumers. And so maybe that brings us back to wrap up to the question of how much it actually matters for Apple. Google was clearly gutting for Apple intelligence with this release, mirroring all the features they couldn't deliver on. Google's head of devices, Rick Osterow, even took a not-so-veiled swipe at Apple during the launch event saying there has been a lot of hype about this and frankly a lot of broken promises too, but Gemini is the real deal.

Starting point is 00:23:56 The really interesting question is that right now we simply don't know how much AI features actually can drive iPhone sales. Remember, despite breaking their promise on AI. Siri and delivering a very underwhelming iOS, Apple's iPhone orders have actually gone up. There's also the fact that if we want a really clear expression of this sort of gap between Silicon Valley's take on AI and everyone else's take on AI, look no farther than general consumers talking about these devices. Reddit, for example, absolutely hates AI. And because of that, they've been loudly complaining about the Pixel 10. In fact, you're seeing a bunch of posts where

Starting point is 00:24:31 people who are genuinely excited about the features and who don't have a stake in the AI culture wars, being kind of confused about why people are so angry. One post was called, there's too much hate on the pixel 10. It reads, so I know a lot of you guys don't like AI or anything that has AI, but aren't these new AI improvements on the pixel 10 genuinely just a nice new feature or improvement? It seems like people just default to thinking the product is bad as soon as they see AI in the marketing. Look, when it comes down to it, the mobile handset wars are about much more than AI features, as much as all the companies are focused on those features as their new storytelling. For a shocking number of people, it's just about what color your messages bubble is.

Starting point is 00:25:08 Branding, my friends, is a powerful, powerful force. But still, it's pretty undeniable that Google is now delivering on a lot of the features that Apple seemed to indicate that they thought would be important to the average consumer who only cares about performance and outcomes and not whether something is called AI or not. It'll be interesting to see if and how this changes the balance of handset buying in the market, but for now, that's going to do it for today's AI. Brief. Appreciate you guys listening or watching as always and until next time, peace.

The AI Daily Brief: Artificial Intelligence News and Analysis - Is Pixel 10 the AI Phone iPhone Never Was?

There aren't comments yet for this episode. Click on any sentence in the transcript to leave a comment.