Is Google Now the AI Leader?

0:00Today on the AI Daily Brief, is Google now the AI leader? Before that in the headlines, And Anthropic closes a monster new round, plus the latest in the talent wars. The AI Daily Brief is a daily podcast and video about the most important news and discussions in AI.

0:23All right, friends, quick announcements before we dive in. First of all, thank you to today's sponsors, Super Intelligent, Fanta, Robots and Pencils, and Blitzy. 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, shoot us a note at sponsors at AIDailyBrief.ai. We are now officially selling into 2026. With that, let's get into today's episode. Welcome back to the AIDailyBrief Headlines Edition, all the daily AI news you need in around five minutes. Today, our main story is all about whether Google is now in the AI lead, and so appropriately, a lot of our headlines have to do with gyrations and moves among the big labs.

1:00We kick off today with a story that I mentioned in passing yesterday as part of the AI predictions, but will now give full coverage, which is that Anthropic has officially closed their latest round of fundraising at a whopping$183 billion valuation. The company brought in$13 billion in fresh capital, co-led by iconic Fidelity and Lightspeed Venture Partners. There was a laundry list of participating funds, but very notably, this round includes a lot of private equity, sovereign wealth, and retirement funds alongside the usual VCs. This feels in many ways like the round where the AI industry as a whole, not just open AI, have definitively outgrown the capacity of Silicon Valley venture funds alone.

1:37Anthropic's previous round was$3.5 billion at a$61.5 billion valuation back in February, meaning they've more than tripled in size. Now, part of why they were able to do this is that the company has gone from$1 billion to$5 billion in ARR year to date. And much of that revenue is coming now from high-ticket customers. Anthropic now claims 300 ,000 business customers and said that their list of large accounts above 100 ,000 in annual spend has increased sevenfold this year. In a blog post, Anthropics CFO Krishna Rao said, we are seeing exponential growth in demand across our entire customer base.

2:11This financing demonstrates investors' extraordinary confidence in our financial performance and the strength of their collaboration with us to continue fueling our unprecedented growth. Claude Code has become a major business line, bringing in$500 million all by itself with usage up 10x since its full launch in June. The TLDR is that a company that already had the wind in its sails just got a massive new war chest to continue its surge. Anthropic, however, was far from the only major lab with big news this week. OpenAI is kickstarting their next era with new CEO of applications Fiji Simo officially at the helm.

2:43On Tuesday, OpenAI announced a big acquisition in leadership restructuring to build out their applications team. The company bought analytics platform Statsig for$1.1 billion in an all-stock deal, marking one of OpenAI's largest acquisitions to date. Statsig specializes in A-B testing and other product optimization services. In addition, Statsig CEO Vijay Raju will be brought in as OpenAI's CTO of applications. He'll have product engineering responsibilities for ChatGPT, Codex, and other forthcoming product lines. Statsig will continue to operate as an independent entity and service their existing customer base out of their Seattle office.

3:16Other personnel changes include engineering lead Srinivas Narayanan moving on to the C-suite as OpenAI's CTO of B2B applications, reporting directly into COO Brad Lightcap. Kevin Wheel, OpenAI's chief product officer, will be moving to the research side of the business, also becoming the company's VP of AI for science. His former product team, including the head of Chachi Bittinic, Turley, will now report directly into CMO. Now, we saw that this was coming, but it is very clear that OpenAI is significantly delineating between the applications and research side of the company. Speaking of talent, the exodus of AI talent continues at Apple with four new departures.

3:53Bloomberg reports that Apple's lead AI for robotics researcher, Zhang Zhang, has left to join Meta's robotics studio. Separately, a trio of AI researchers at Apple's in-house LLM team have left for greener pastures and Anthropic and OpenAI. The reporting states that the Foundation models team has lost roughly 10 members, including their chief, over recent weeks. The departures come as Apple reportedly mulls outsourcing their LLM needs to Google or Anthropic. Sources also said that Apple's staff departures are expected to continue, with basically every other researcher actively interviewing at other AI companies.

4:24Said the sources, the poor response to Apple's intelligence and the company's potential shift towards using third-party models have contributed to worsening morale. Yet when it comes to worsening morale and big questions around talent, the biggest lens right now is squarely focused on meta. Reporting is rife with claims that there are rifts forming inside the new superintelligence lab. The Verge, for example, filled in the gaps from last week's reporting that a trio of researchers had left the team. They reported that two had never formally begun their roles before deciding to return to OpenAI, while the third was a confirmed departure, with a staff member working a month before changing their mind.

4:58TechCrunch added that ScaleAI's former senior VP of GenAI Product and Operations, Ruben Mayer, had left the company after two months. Mayer was brought across as part of the$14.3 billion scale aqua hire. Still, the reporting is a little strange. TechCrunch's Sources said that Mayer was never a part of TBD Labs and had instead been assigned to oversee AI data operations teams, but Mayer himself disputed the source, stating that his initial position was to, quote, help set up the lab with whatever was needed rather than just data. He also said that he was, quote, part of TBD Labs from day one rather than being excluded from the core AI unit.

5:30He said that he did not report directly to Alexander Wang and was very happy with his Meta experience, instead leaving for a personal matter. In addition to the departure, TechCrunch reports that Meta is now partnering with alternate data labeling vendors, including Surge and Mercor, instead of working exclusively with Scale. A third piece from the Financial Times said that ChatGBT co-creator Shang-Zhiya Zhao threatened to quit within days of signing with Meta. Now, Zhao was one of the highest-profile recruits to Meta and really cemented that they were having a lot of success in poaching top talent from OpenAI.

5:59FT sources said that Zhao went so far as to sign employment paperwork to return to OpenAI within days of his recruitment. Shortly afterwards, Meta appointed him chief AI scientist and published an introduction video featuring him sitting next to Alexander Wang and Mark Zuckerberg. One investor close to Meta leadership commented, There's a lot of big men on campus. For their part, Meta does seem to be getting a little frustrated with the frequent coverage, with a spokesperson stating, We appreciate that there is outsized interest in seemingly every minute detail of our AI efforts, no matter how inconsequential or mundane, but we're just focused on doing the work to deliver personal superintelligence.

6:33Look, at the end of the day when it comes to these questions, there are two things. The first is that when you spend the type of money and are as aggressive as this company was in recruiting, that makes this all interesting cannon fodder for media stories. It's unavoidable, even if the Occam's razor explanation for this is that you're not going to be able to be this aggressive without having some people change their minds. In general, I personally am a fan for Occam's razor explanations, and my best guess is that we're in some sort of settling phase, rather than there being abnormally utter chaos given the extremes of the situation.

7:05The second thing to note, though, is that the way to get people to shut up is to put out stuff that works. Now, we're barely months into this. It would be unrealistic to expect even an incredibly performant team to be dropping models and applications this fast. But the big do or die moment for Meta Superintelligence Labs that will absolutely crater this and all other discussion, like the discussion that's also happening around Meta apparently considering partnerships with OpenAI or Google to power their products, the only thing that's going to squash that ultimately is what they put out. That is certainly what we will be watching for the rest of this year.

7:35But with that, we wrap today's headlines. Next up, the main episode. If you are a regular listener, you will have heard about Superintelligent's 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, Superintelligent 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.

8:09These 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. 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 Super Intelligence Agent Planning Suite, we've built a custom GPT to answer your questions.

8:37Just go to bit.ly slash super super agent. That's bit.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. 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, Vanta gets you audit-ready fast and keeps you secure with continuous monitoring as your models, infra, and customers evolve.

9:19Fast-growing customers like Langchain, 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-T-A dot com slash NLW to save$1 ,000 for a limited time.

9:53AI isn't a one-off project. It's a partnership that has to evolve as the technology does. Robots and Pencils work side-by-side with clients to bring practical AI into every phase. Automation, personalization, decision support, and optimization. They prove what works through applied experimentation and build systems that amplify human potential. As an AWS-certified partner with Global Delivery Centers, Robots and Pencils combines reach with high-touch service. Where others hand off, they stay engaged. Because partnership isn't a project plan. It's a commitment. As AI advances, so will their solutions.

10:27That's long-term value. Progress starts with the right partner. Start with Robots and Pencils at robotsandpencils.com. 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, bringing in their development requirements. The Blitzy platform provides a plan, then generates and pre-compiles code for each task.

10:58Blitzy 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 SDLC 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. Visit blitzy.com and press Book Demo to learn how Blitzy transforms your SDLC from AI-assisted to AI-native.

11:34That's B-L-I-T-Z-Y dot com. Welcome back to the AI Daily Brief. Today, we are nominally talking about Google's big win in their antitrust case. TLDR on that is that they are not going to be forced to sell Chrome. So we will talk about the implications of that, what it means for their positioning vis-a-vis AI, why Chrome might matter in the AI race. But I thought it would be interesting to use this as a chance to do a little bit of a retrospective around where Google has been. I think when it comes to the major labs, no one has gone through such wild swings of where they sit in the AI race as has Google.

12:09There is a growing chorus of people who believe that Google is now firmly in the lead when it comes to AI. That's showing up not just in random tweets, but on prediction markets. Since all the way back in May, Google has been slightly in the lead when it comes to the polymarket prediction market for which company will have the best AI model by the end of 2025. But since the release of GPT-5, the number has absolutely skyrocketed up to Google having a 75 % chance. Now, on the one hand, this shouldn't be all that surprising. Google has had a longer relationship with generative AI building than most of their competitors.

12:44They had the talent, the money, the motivation, and the drive, which is why it was so shocking how far behind they found themselves when ChatGPT was launched. The piece that we're looking at is from Forbes in February of 2023. It was called AI First to Last, How Google Fell Behind in the AI Boom, and was one of a series of pieces with similar themes around that time. The piece discussed Google's early lead in AI. They talked about how just a few months after becoming CEO back in 2016, Sundar Pichai declared that Google would become an AI-first company. Now, interestingly, even back then, they already felt like they had their back to the wall at least a little bit.

13:22Amazon's release of Alexa had taken Google by surprise, feeling like they should have been first to market with that product. Forbes wrote, seven years later, Google finds itself in a similar position, again beaten to market in a field it should have dominated. But this time, it's worse. The usurper is OpenAI, a comparatively small San Francisco startup and not a deep-pocketed giant like Amazon. The product is ChatGPT, a bot that can generate sitcom plots, resignation letters, lines of code, and other texts on almost any subject conceivable as if written by a human, and it was built using a technology breakthrough Google itself had pioneered years ago.

13:55Now, of course, part of what made this all the more frustrating was that Microsoft, by dint of their relationship with OpenAI, had moved into more of a software leadership role than they'd had in some time. And for much of 2023, Google's story in AI was all about catching up. There were a bunch of product announcements with names that you might or might not remember, for example, BARD, but nothing that could be considered even close to GPT-4 level, in other words, close to the state of the art, until the very end of the year when Google announced Gemini. Now, I was already six months into the show at that time, and I remember that the general sentiment was that it was very much a pressured and rushed announcement forced by how far ahead OpenAI was.

14:36Although it was announced in December, it was announced with a bunch of pre-recorded demos, and the most performant model wasn't even available yet. That wouldn't come until the beginning of the next year. And when it did, it did little to improve Google's rough position. First of all, Google got itself completely caught up in the culture war by releasing an image generation model that tried extremely hard to add diversity into historical images in places that were completely inappropriate. For example, putting people of color into Nazi uniforms as emblematic examples of early 1940s German soldiers.

15:10Hold aside the political dimensions of this, it was bad enough that Google pulled the model entirely and later apologized for releasing something that was so far out of sync with reality. A couple months later, we had the first incarnation of Google's AI overviews. While they've now become totally de rigueur and very much a part of the search experience, at the beginning, all that anyone could talk about were things like where Google suggested that people use glue as an ingredient and in pizza. Now, whereas the problems with the image generator were extremely widespread, Google did try to say that the examples going public with the AI overviews were more isolated, but still it all contributed to this narrative of Google being incredibly far behind.

15:47Behind the scenes, though, there was a lot going on at the company. Some of that was personnel. One of the issues with Google had been that AI was so big at the company that it was spread across a lot of different divisions rather than consolidated and focused. Over the course of 2024, there was a set of at sometimes painful consolidation to bring everything basically under DeepMind and DeepMind leader Demis Hassabis. For example, it took until October of 2024 for Google to fold the team that was responsible for the Gemini app into DeepMind officially. But where things really started to turn around for Google was, of course, the release of Notebook LM.

16:27Notebook LM was first announced earlier in the year as part of a string of announcements and was frankly kind of buried when it came. However, in the fall, they released a new feature called Audio Overviews, which was basically a way to auto-generate a podcast based on any subject matter. You could drop a set of documents or source links into Notebook LM, and in a couple of minutes, you would get an audio conversation between two hosts explaining the material in ways that were engaging and really fell into all the tropes of podcasting. Now, this is the type of thing that could easily just become a novelty, something that people tried once or twice and then went away from.

17:02That didn't happen. And I think part of why it didn't is that Google really had unlocked a different medium for consuming information that some number of people found really useful. For example, for me on the podcast, it was sometimes a more engaging way to deal with complex technical topics that I wanted to then share with this audience. I knew law firms who were using it to help summarize case files. You had students who were using it as a study tool. And the point was that for the first time, Google AI had a genuine bona fide hit on their hands that people were really thrilled about. I think it was also a testament to the goodwill and good brand that Google has built over the last decades that there was almost a sense of relief of Google getting back in the AI game and doing something well, as though the aberration had always been when they were behind.

17:46As the year closed out, it was clear that heading into 2025, AI competition was going to be the big mission. In a December town hall, CEO Sindhar Pichai told the company, I think 2025 will be critical. I think it's really important we internalize the urgency of this moment and need to move faster as a company. The stakes are high. These are disruptive moments. In 2025, we need to be relentlessly focused on unlocking the benefits of this technology and solve real user problems. And indeed, 2025 has been an absolute flurry of products. One interesting inflection point came at the I.O. conference in May.

18:20Google made an odd decision heading into the conference, releasing AI mode for search the prior week. When he took the stage for his keynote, Pichai said, Normally you wouldn't have heard much from us in the weeks leading up to I.O. because we'd be saving our best models for the stage. But in our Gemini era, we're just as likely to ship our most intelligent model on a Tuesday in March or announce a really cool breakthrough a week before. We want to get our best models into your hands and our products ASAP, so we're shipping faster than ever. And ship they did. Google coverage listed 100 different things announced at I.O., and a very meaningful portion of them were AI-related.

18:53And there was definitely a sense that Google was starting to get its groove back. Now, there are two big areas where Google has been very on the ball this year. One of them is multimodal, which we'll talk about in just a moment. And where I think more than anything else, the sense of Google maybe now being in the leader or pole position with AI comes from. But it's also important to note that they recognized quite early in the year that agentic coding was going to be a massive, massive theme. Back in February, they released a version of Gemini 2.0 that was fine-tuned for coding. Again, CEO Sundar Pichai tweeted, New Gemini 2.0 updates for developers.

19:28Starting today, you can use Gemini Code Assist for free in your favorite IDE. It's Gemini 2.0 fine-tuned for coding, including 180k code completions per month, 90x what others currently offer. Plus, Gemini 2.0 Flashlight is now GA and cost-effective for projects that use long context windows. And so what I think is notable here is that one, they understood that models that were specific to coding were going to be important. And two, they knew that they wanted to compete on cost because the increase in performance of these models was going to ultimately lead to a big increase in the tokens that would ultimately be consumed.

20:01A couple months later in May, they released an updated Gemini 2.5 Pro, once again focused on coding. Google DeepMind CEO Demis Hassabis tweeted, very excited to share the best coding model we've ever built. Today, we're launching Gemini 2.5 Pro Preview IO Edition with massively improved coding capabilities. It's especially good at building interactive web apps. A couple months later in June, N. Taylor Mullen tweeted, this is truly a proud moment for me. For the past few months, we've pushed nonstop to build Gemini CLI. It's been a journey and I couldn't be happier to finally share it with all of you.

20:33Okay, so the point here is clearly that Google was focused on this coding use case. And if you need evidence of whether it worked, look no further than an announcement from July where Google shared that they had jumped from processing 480 trillion tokens in May to 980 trillion by July, 104 % growth in just a couple of months. I think a huge part of that was the emergence of this coding use case. Still, it's impossible to talk about Google AI right now without recognizing that where they have really started to break out into a distinct lead is around multimodal, specifically when it comes to video, image generation, and world models.

21:09For a while, all the buzz in video was around OpenAI's Sora. Over time, however, Sora stayed very much in a behind-the-scenes-for-paid-partners-only kind of mode, and when Google released VO3, it had one very specific transformative difference that made, as it turns out, all the difference in the world. That was the ability to generate sound and video at the same time. For the first time with AI video generation, people didn't have to create the video clips and the audio clips separately and lay them together with some program. Instead, they could generate in a single shot the full clip, including video and audio.

21:45AI-generated video absolutely exploded from there. I did a show back in July about 10 AI video trends taking over the internet, and if you've spent any time on TikTok or Instagram Reels or any social media in the last three months, you have absolutely been inundated with VO3-generated videos. Of course, it's not just video. More recently, we've got Genie 3, which is Google's experimental world model where you can generate entire interactive environments with just a prompt. This is still a little bit more firmly in the realm of the future and the experimental, rather than something you're seeing people deploy for a lot of different use cases right now.

22:20But it's still contributed to this sense that when it comes to multimedia and multimodality, Google is just on a different plane. And then, of course, more recently, you'll know if you've been listening to the show, that everyone's been talking about Nano Banana, or as it's technically called, Gemini 2.5 Flash Image, or something along those lines. Now, what made Nano Banana different is its ability to do editing of a particular image in a way that is extremely high fidelity to the prompt. It captured more attention than any image generation model we've seen since the Studio Ghibli trend of earlier in the year from OpenAI.