Transcription
I'm cancelling Cursor, Cloud Code, and Tragic, and going all in on Google's AI ecosystem. Google, the company that invented the Transformer architecture and then let OpenAI eat their lunch. The company that had infinite money and still fumbled the bag. But something has changed. Sergey Brin is back in the office writing code. Google is desperate to win. And if I'm reading the science right, this is the biggest asymmetric bet you can make in tech right now. I understand their demos are fake, their launches are confusing, and they kill products for fun. Their ecosystem is a mess. I get it. But before you think Google sucks, hear me out. These are my six reasons why the giant is back. Plus the massive risks that still worry me.
Reason number one, the vibe shift. You can feel it. OpenAI is bleeding talent, declaring code a threat. Meta is running out of steam, and Anthropic is burning through its cash runway. Meanwhile, Google has entered wartime mode. Look at this guy, Sergey Brin, co-founder of Google. Worth 250 billion.
>> Man, I hear the reports. You and I have talked about it. You're going to work every day.
>> Yeah, [clears throat] it's been, you know, some of the most fun I've had in my life. I retired like a month before COVID hit in theory. I want to do something else. I want to hang out in cafes.
Right now, he could be on a private island sipping margaritas. Instead, he came out of retirement not to sit in meetings, but to write code.
>> This guy named Dan and I, I ran into him at a party and he said, you know, look, what are you doing? This is like the greatest transformative moment in computer science ever.
This is founder mode activated. This is the signal to pay attention. Google has the data, and they have the compute. Unlike startups, they own the hardware stack. This allows them to do crazy things like release massive models for free while everyone else is counting tokens. And most importantly, Google finally wants to win. They are tired of losing. All over Twitter, Google engineers are replying, trying to figure out how to improve the product, asking developers what they want. They finally ship like a startup again.
Reason number two, the tech. For a long time, Google's offerings were lacking. No AI code editor, no coding agents, no CLI. But recently, they caught up big time. Let's start with Gemini CLI. While the developer community often considers Cloud Code as the gold standard, Gemini CLI has quietly become a top-tier contender. It's powered by one of the best models out there. It's open source, it's highly customizable with extensions. And unlike the competition, it's free and has generous limits. You can get a ton of work done without hitting a paywall every 10 minutes. It's slick, it's easy to use, and it feels like a powerful tool designed specifically for developers.
Then there is Anti-Gravity. This is Google's new AI-first IDE. We all know that Cursor is the king of code editors right now. But Google isn't trying to copy or match Cursor in this game. Google is coming for the crown. Anti-Gravity is a bold bet. It's designed for an agentic-first workflow. You aren't just writing code. You're managing a team of autonomous agents. I want to explore deeper and understand if this style of development is just a hype, or if it could actually be the next evolution of how we build software. And Anti-Gravity supports Cloud as well. So we get the Google ecosystem features, the deep integration, and access to the best models from the competition. Plus, at some point, you need real infrastructure because you're already in Google's ecosystem. Graduating your project from a prototype to a production app is more seamless. You aren't stitching together five different vendors. It's one pipe. And finally, there's Project Jewels. It's their AI agent. You assign it an issue, and it goes off, creates a branch, fixes the bug.
Reason number three, the bleeding edge, the toys. Google is dropping experiments that are genuinely fun. Again, first Whisk. You upload a subject, scene, and style, and it remixes them based on a prompt. Additionally, you can animate or refine the generated image. Essentially, it's an experimental generative AI tool from Google Apps that focuses on prompting with images rather than just text. Then we have Flow. This is an AI filmmaking tool. It's quite powerful and looks a bit intimidating at first, but basically, you can bring in assets and use VO to generate clips. Nano Banana Pro for images and actually direct a video. It feels like the future of filmmaking, and I want to be on the bleeding edge of video-making tools. Next up, there is Nano Banana and VO, which are the best image and video generators out there. So many steps ahead of the competition. They can be used for rapid prototyping, ideation, increasing product development for creatives, you name it. It's not just "make a picture of a cat." It's "change the lighting on this specific product shot" or "move this object to the left." Those tools understand the spatial 3D world. And of course, Anti-Gravity, which is a bet towards a new way of coding and developing software. The downside, of course, is that a lot of those experiments will be killed. I just hope they don't kill the wrong products. And all these developers and experimental tools are available at an extremely affordable price.
This brings us to packaging and value. So, reason number four, packaging and value. At this point, I have a bit of subscription fatigue. Pay for ChatGPT, pay for Cursor, pay for Cloud Code, pay for Perplexity. Why pay for all of them if I can pay for Google only? The value is insane. Instead of ChatGPT, we have the Gemini app. Instead of Cloud Code, we have Gemini CLI. And instead of Cursor, we have Anti-Gravity. Then, on top, we get credits for Nano Banana, V, NotebookLM, 2TB of Google storage, discounted YouTube Premium, and Gemini baked into Docs, email, and Chrome for roughly $20. It's crazy. Gemini seems to be the best value for your money right now.
I'm saving the strongest reason and my concerns for last. First, let's talk about reason number five. Infinite money. Although too much money can be a curse, I think this tweet summarizes it pretty well: "Google can win because they can afford to burn." Google can afford to burn money in order to train the best image generation model, the best video generation model, and offer lots of their developer-oriented tools free or at higher limits. This has a double effect: getting way ahead of competition in multimodality and growing their ecosystem. Why can Google afford to offer that bundle for $20? They're subsidizing our productivity to win back market share. This matters for us because it means higher limits. While Cloud is telling you, "You've hit your message limit. Come back at 4 p.m.," Gemini is like, "Keep going."
Reason number six, ecosystem and integration. I want to pick one ecosystem and get actually good at it, rather than being mediocre at five different tools. It's less mental overhead to keep up with one ecosystem instead of three. Basically, I get more value from time invested in theory, anyways. And because I use so many Google tools day-to-day, such as Gmail, Docs, now I have Gemini integrated in all of them. I don't copy-paste to ChatGPT anymore. I just highlight a paragraph in Docs and ask Gemini to rewrite it. It has context. It sees what I see. And looking forward, Anti-Gravity, because Google owns the browser Chrome and the IDE Anti-Gravity. The integration potential here is pretty big. We're talking about agents that can control the browser natively for testing in ways that a third-party app like Cursor just can't match because they don't own the platform. It's the Apple effect with different players.
This brings us to deep research. Google has the Knowledge Graph. They have spent 25 years mapping the entire internet. So, Perplexity, for example, is renting and building the map. Google is the map. The results that Deep Research and also AI Search generate are extremely powerful, and Google has all the right ingredients to be the best in class in these areas. Next up, NotebookLM. Because Google owns YouTube, the integration is native. I can paste a video link into my NotebookLM and it actually understands it, summarizes it, and answers questions. This is a workflow that OpenAI literally can't copy because they don't have the platform access. NotebookLM can also generate slides, infographics, videos, podcasts based on your Google Docs. The examples you see on the screen are based on the script for this video. And we have to talk about vertical integration. OpenAI and others have to pay the Nvidia tax. They buy or rent GPUs. Google makes the chips. They own the TPUs. This allows them to offer these crazy high limits and free tools. They can afford to offer a 2 million token context window. They can afford to offer what others cannot.
But it's not all great. Here are the ugly parts. Google tools feel very corporate. They are trying to be fun, but you can feel that decisions go by 15 managers. The disclaimers everywhere, the censorship, although that's getting better. At times, the simplicity is missing. The ecosystem is confusing. There are 100 ways to do the same thing. What's the difference between AI Studio, Vertex AI, Workspace Studio, Gemma, Gemini, and how to access all those? There are also some embarrassing issues. The dictation on the Gemini app on the web is surprisingly bad for a feature that important. And most importantly, the graveyard. This is the big one. I'm afraid that I'll fall in love with Anti-Gravity, and Google will kill it in 18 months because it didn't reach 15 billion users. This is the risk that I'm taking.
But all in all, I think Google deserves a chance. So, this is Project All In on Google. The rules: 90 days, no Cursor, no ChatGPT, no Cloud Code. The goal: to see if the all-in ecosystem actually makes sense, to see if Google is so back or it's so over. Thanks for watching, and see you in the next one.