- AI Wire
- Posts
- 🔥 OpenAI’s money leak
🔥 OpenAI’s money leak
The numbers everyone is whispering about
Welcome to AI Wire — your smart shortcut to all things AI, without the jargon.
What we’ll cover today:
🔥 OpenAI’s money leak
✂️ ChatGPT’s writing fix
📚 Harvey’s legal AI boom
🎥 Jeff Su’s must-watch guide
The leaks show OpenAI is paying Microsoft huge revenue share fees while still burning more on inference than it earns.
They paid Microsoft $493M in 2024 and $865M in the first 9 months of 2025 just in revenue-share.
The revenue share math between OpenAI and Microsoft is complicated, so the public numbers do not show the full picture.
If the top AI company is struggling with margins, the smaller players built on its models are in even tougher shape.
Will AI stay expensive to run? |
FROM OUR FOUNDER…
THE SMARTEST WAY TO GROW YOUR NEWSLETTER 🚀
Tired of low-quality subs and wasted ad spend? I’ve scaled AI, fintech, SaaS, and multiple niche specific newsletters with real, engaged and high intent subscribers. Would love to do the same for you!
✅ 40%+ open rates: no bots, no junk leads.
✅ No ad management fees: just pay per engaged subscriber.
✅ First-party data & proprietary methods ensure quality.
✅ Risk-free scaling: we’ve built and grown newsletters ourselves.
📩 Want high-intent subscribers that actually engage? Let’s talk.
ChatGPT can now avoid em dashes if users disable them in custom instructions.
The change fixes one of the biggest giveaways of AI written text.
It also shows how hard it is to control writing style in large models.
Personalization is becoming a core product feature, not a nice extra.
Do you trust AI to match your writing style? |
Harvey has grown to $100 million ARR and 700 clients by solving real legal workflows.
The company’s advantage comes from workflow data and a true multiplayer legal platform.
Strict global data rules force Harvey to run compute in many countries, which is costly but defensible.
Legal AI adoption is still tiny, so Harvey is capturing a massive market long before competitors catch up.
When will AI handle complex work? |
🛠️ TOOLS
⚡ Group Chats in ChatGPT – AI powered shared conversations for teams →
🛠️ Swytchcode – Converts API docs into an AI playground →
✍️ SourcePilot – Personal AI editor for better writing →
📚 RESOURCES
🚀 10 Ways to Use ChatGPT – Tutorial on advanced ChatGPT usage →
🎬 Cinematic AI Videos Guide – Learn to create cinematic AI videos →
⚙️ Build Apps With AI – Complete guide to AI app building →
📘 AI Habit Log Tutorial – How to build an AI powered habit log →
Jeff Su explains ChatGPT in the simplest way possible. It shows when to use each model, when to use search, how Deep Research works, and how Canvas helps you think better. You should watch this because it teaches you how to get real results from ChatGPT without any confusion.
What do you think about today’s edition? |
Was this forwarded to you? Sign up here.
AI Wire News.
Signing off