- AI Fire
- Posts
- 🍏 Apple’s $900M AI Toll Booth
In partnership with
Apple just proved you don't have to build the smartest AI to make nearly a billion dollars from it, you just have to own the "toll booth" everyone uses to buy it!
What's on FIRE 🔥
IN PARTNERSHIP WITH HUBSPOT
Want to get the most out of ChatGPT?
ChatGPT is a superpower if you know how to use it correctly.
Discover how HubSpot's guide to AI can elevate both your productivity and creativity to get more things done.
Learn to automate tasks, enhance decision-making, and foster innovation with the power of AI.
AI INSIGHTS
Apple may look behind in the AI race, but it’s still winning in one important way: distribution. According to new data from AppMagic reported by the WSJ, gen AI apps paid Apple nearly $900 million in App Store fees in 2025.
And around 75% of that revenue came from ChatGPT. Even if Apple isn’t leading model innovation, it still controls one of the most powerful toll booths in tech. Here’s how the fee structure works:
Standard commission: 30%
Small Business Program (< $1M revenue): 15%
Most revenue driven by subscription-based AI apps
As AI apps become daily tools, subscription revenue compounds quickly, and Apple collects a percentage every month. ChatGPT accounts for roughly 75% of Apple’s AI-related App Store income. xAI’s Grok contributes about 5%.
Monthly revenue from generative AI apps grew rapidly through 2025, rising from $35M in January to a peak of $101M in August, before slowing slightly as downloads stabilized.
Interestingly, Apple is investing far less in AI infrastructure than competitors. Distribution can sometimes be more profitable than invention.
PRESENTED BY WISPR FLOW
Stop typing prompts. Start talking.
You think 4x faster than you type. So why are you typing prompts?
Wispr Flow turns your voice into ready-to-paste text inside any AI tool. Speak naturally - include "um"s, tangents, half-finished thoughts - and Flow cleans everything up. You get polished, detailed prompts without touching a keyboard.
Developers use Flow to give coding agents the context they actually need. Researchers use it to describe experiments in full detail. Everyone uses it to stop bottlenecking their AI workflows.
89% of messages sent with zero edits. Millions of users worldwide. Available on Mac, Windows, iPhone, and now Android (free and unlimited on Android during launch).
AI SOURCES FROM AI FIRE
1. 5 Must-Have Rules for Building Working AI Agents in Your Daily Life (I Said 'Must-Have'). Almost nobody talks about what actually keeps agents alive. Making it reliable is different. Here are the 5 rules most people miss.
2. ALL Working Prompts in Only 5 Minutes with NotebookLM (Included 1,500+ Free Prompts) Your prompts shouldn’t feel lost. Use NotebookLM to turn a massive open-source prompt database into a private system that finds the right prompt when you need it.
3. Create 1–30 Minute AI Videos on Your Phone. Way Easier Than Expected (No Camera, $0). A complete 2026 workflow to write, animate, and edit high-retention faceless YouTube videos using only free mobile tools.
4. How to Build a One-Person AI Business Faster Without the AI Agent Hype. Some solo founders are scaling faster than small teams. The biggest shift in business right now isn’t AI agents. How to actually build a one-person business faster.
AI COMMUNITY WORTH JOINING
Not sure why but every day, I receive many DMs asking very specific questions: “How to use AI for marketing? Which tool works best for automation? How to build an AI workflow for a specific job?”
Most of the answers are already inside our community. I’ve organized tutorials, case studies, workflows, and tool breakdowns in one place so you can search by keyword and quickly find what you need.
You don’t need to scroll endlessly or start from zero. Just type the tool name, job role, or problem, and relevant guides appear. Like this:
And if you still can’t find what you need, you can message me directly inside the community. I read messages and often create new tutorials based on member questions. There is no fee to join.
If you want one place where most AI questions already have answers, this is it.
TODAY IN AI
AI HIGHLIGHTS
🧩 Cursor admits its new “frontier” coding model is partly built on Moonshot’s Kimi 2.5, then added heavy training. In the AI arms race, who is really building vs remixing?
📱 Anthropic launched Claude Code Channels (preview). You can message your Claude Code session via Telegram or Discord, and it replies instantly. See it in action here.
🌍 OpenArt just launched Worlds that lets you turn one prompt or image into a fully explorable 3D environment, even walk inside scenes. Try using it for free here.
👀 OpenAI is expanding ChatGPT ads to Free and Go users in the US. Sponsored results now appear under answers, clearly labeled. If you see this, don’t be surprised.
🏡 A tech CEO used ChatGPT to sell his home for $100K more than realtor estimates. It suggested pricing, staging, even paint ideas. No expensive middlemen. Full story.
🧠 Mark Zuckerberg is building an AI agent to help him be CEO. Employees are already using personal agents that talk to each other. Meta becomes fully AI-native.
🤖 Elon Musk wants Tesla and SpaceX to build their own chips. The proposed “Terafab” could power huge AI workloads on Earth & space. No timeline yet.
💰 Big AI Fundraising: Halter, an AI startup in livestock management, reached a $2B valuation after a new funding round led by Founders Fund, signaling strong market confidence in AI agriculture.
NEW EMPOWERED AI TOOLS
🎯 MasterClass On Call analyzes your meetings live, giving instant guidance and post-call insights to improve leadership, communication, and decision-making automatically. See it in action for free.
🧠 Lokuma is a design intelligence layer for agents like OpenClaw, Claude Code, or Codex. Turning raw outputs into something clear, structured.
⚡ Fractal is the fastest way to ship exceptional ChatGPT apps. It handles all architecture planning, coding, testing, and deployment.
🧩 Vite+ is a new unified toolchain to web application development. Manage your runtime, package manager, and frontend stack with one tool.
AI CHARTS
Click to play this for real (using your keyboard)
A world model is an AI system trained to simulate reality. It builds an internal representation of how environments behave, allowing the AI to test actions before taking them. Examples of what a world model tries to capture:
How gravity affects objects
How humans move and interact
How environments respond to actions
How cause leads to effect over time
This capability is critical if AI is going to move beyond chat interfaces. Large language models are powerful, but they lack grounded understanding of the physical world.
Embodied intelligence requires prediction across time, space, and motion, not just words. We’re starting to see early signals of world-model-like capabilities across AI research and products.
Packy McCormick argues that world models could dramatically accelerate progress toward AI that can operate in the real world. Robotics, logistics, manufacturing, healthcare, and autonomous systems all benefit from this shift.
That creates a powerful feedback loop: Simulate → test → learn → improve → repeat.
We read your emails, comments, and poll replies daily
How would you rate today’s newsletter?Your feedback helps us create the best newsletter possible |
Hit reply and say Hello – we'd love to hear from you!
Like what you're reading? Forward it to friends, and they can sign up here.
Cheers,
The AI Fire Team
Reply