- AI Fire
- Posts
- 🌶️ OpenAI’s Secret $18B 'Jalapeno' Gamble
In partnership with
Project Nexus Snag: Why Broadcom is demanding an $18B commitment from Microsoft and what "Jalapeno" chips mean for OpenAI's profitability!?
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
OpenAI’s big dream of building its own custom chips (codenamed Project Nexus) has run into a bit of a speed bump. To get the first phase moving, they need $18 billion.
Broadcom is willing to finance the first 1.3 GW phase (costing $18B), but only if Microsoft agrees to buy 40% of the initial batch. Right now, Microsoft is holding the pen, and they aren't signing.
OpenAI’s custom inference chip, codenamed "Jalapeno," is designed to power ChatGPT with way higher efficiency than standard Nvidia GPUs.
Microsoft wants "versatile" data centers that can run anything, while OpenAI wants "specialized" centers built only for Jalapeno.
OpenAI is projected to burn over $200 billion through 2029. If they can't build their own chips, they have to keep buying them from other people at a premium. Building Jalapeno is their best shot at making the company profitable in the long run. Without this deal, their path to "superintelligence" gets a lot more expensive.
It’s easy to see these headlines and think there’s trouble in paradise, but this is actually a pretty normal "growing pain" for a partnership this big. Microsoft and OpenAI are still very close, but they’re starting to realize that building the physical world (data centers and power plants) is much harder and riskier than writing code.
🎁 Today's Trivia - Vote, Learn & Win!
Get a 3-month membership at AI Fire Academy (500+ AI Workflows, AI Tutorials, AI Case Studies) just by answering the poll.
What's the main goal behind OpenAI building AI chips? |
PRESENTED BY BELAY
AI can help you move faster, but real leadership still requires human judgment.
The free resource 5 Traits AI Can’t Replace explains the traits leaders must protect in an AI-driven world and why BELAY Executive Assistants are built to support them.
WORTH SAVING: PRO AI GUIDES & TUTORIALS
1. Top 9 Claude Code Skills & Plugins That Make AI Web Design Actually Look Designed! From simple prompts to full UI control. These 9 frontend tools, Skills, Plugins, and CLIs will unlock a level most people never reach.
2. Complete Beginner’s Guide to Local AI Agents in 2026: Everything You Need to Know. Learn how to build your own 24/7 AI agent locally in 2026. This guide uses local AI agents using OpenClaw, Claude Cowork, local models,...
It feels like Claude is dropping new features almost every single day. At this point, following the Claude ecosystem is starting to feel like following Google’s ecosystem years ago. And honestly… it’s getting hard for beginners to keep up.
That’s exactly why we made this video. Not just to explain one feature. But to give you the bigger picture behind Claude: how the ecosystem fits together, which tools actually matter, what beginners should focus on first,…
If you want one solid foundation before Claude becomes even bigger, save this guide. You’ll understand Claude far differently after reading it.
TODAY IN AI
AI HIGHLIGHTS
👨💻 For the first time, Anthropic’s Boris Cherny did a fully unscripted livestream using Claude Code exactly the way he works every day. Really worth watching, video here.
📱 If you still believe every screenshot online, this might change your mind. ChatGPT’s newest image model is making fake screenshots almost impossible to spot, like this.
🧠 Claude Managed Agents just got 4 big upgrades that let you build agents that actually get smarter over time. Try it via the OSS repo at github/anthropics/skills.
🚀 Anthropic just doubled Claude Code’s 5-hour limits for all paid users. They also removed peak-hour restrictions, and significantly increased Opus API rate limits.
🏃 Google’s $9.99-per-month AI health coach is launching on May 19. It knows your sleep, workouts, stress levels, then gives personalized health advice in real time.
🏛️ Ex-OpenAI CTO finally spoke out in Musk’s lawsuit against OpenAI. She accused Sam Altman of lying about safety reviews and creating chaos across leadership.
💰 Big AI Fundraising: China’s Moonshot AI raised $2B at a $20B valuation after its Kimi models exploded. It reportedly hit $200M ARR, showing massive demand for cheaper open-source AI models.
NEW EMPOWERED AI TOOLS
💼 Claude Agents for Financial Services gets 10 pre-built Claude agent templates for investment research, KYC screening, month-end close.
🎨 Google Pomelli Catalog gets brand-consistent campaign assets, ad creatives, and photoshoots generated.
🤝 FlowMarket is a network of AI agents that automatically discover, match, and generate B2B deals, without the need for large sales teams.
🌍 Lingo.dev v1 helps teams build consistent AI translations with brand voice rules, localization APIs, and quality scoring.
AI BREAKTHROUGH
OpenAI is currently facing a massive wave of lawsuits from families alleging that ChatGPT, in some cases, actively helped users plan self-harm ideation. So, it responded with Trusted Contact to put a human "safety net" inside. How it works:
Adult users can opt-in to name a specific friend or family member. If a conversation turns toward self-harm, OpenAI encourages the user to reach out directly.
When the system flags a serious safety risk, a human safety team reviews the incident → OpenAI claims they strive to do this in under 1 hour.
If the human reviewer confirms the risk, the "Trusted Contact" gets a text or email. Crucially, it does not include chat logs.
The feature is 100% optional. You can toggle it off, or simply open a secondary account to bypass the safety person entirely.
OpenAI is effectively offloading the final check-in responsibility to your friends and family. It’s a clever way to ensure that if a tragedy occurs, the company can point to the feature and say, "We tried to tell someone."
For families already suing the company, this might feel like "too little, too late," but for the 920M weekly users, it’s a reminder that while ChatGPT can be a coder, a writer, or a tutor, it is not a therapist.
Talk to your AI tools the way you'd talk to a colleague.
You don't send a colleague a three-word brief. You explain the context, the constraints, what you've already tried. But typing all that into ChatGPT takes forever — so you don't.
Wispr Flow lets you speak your prompts instead. Talk through your thinking naturally and get clean, paste-ready text. No filler words. No cleanup. Just detailed prompts that actually get you useful answers on the first try.
Millions of users worldwide. Works system-wide on Mac, Windows, and iPhone.
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